Init Live Search – AI-Powered, Related Posts, Slash Commands
Deliver an ultra-responsive search experience to your visitors — no page reloads, no jQuery, no lag. Init Live Search is a modern, lightweight, and fully accessible live search solution for WordPress — now with native Block Editor (Gutenberg) blocks and Abilities API support (WordPress 6.9+), an optional MySQL FULLTEXT search index, optional Meilisearch integration, tag-aware matching, SEO metadata support, ACF integration, WooCommerce product filters, and customizable UI presets. It replaces the default with a clean, intuitive search modal powered entirely by the WordPress REST API. Everything loads in real-time — with zero disruption to browsing flow. Perfect for content-heavy blogs, WooCommerce stores, or even headless sites. Every interaction is fast, fluid, and designed to work across devices. Want typo-tolerant, sub-50ms relevance ranking on top of that? Connect your own self-hosted or cloud Meilisearch instance in a few clicks — Init Live Search will automatically prefer it for search, and just as automatically fall back to the built-in database search if it’s ever unreachable. Your visitors never see a broken search box. Prefer working entirely in the Block Editor? Three dynamic blocks — Live Search Box, Related Posts, and AI Related Posts — live under their own Init Live Search category in the block inserter, each with a live server-side preview as you configure it, and each sharing the exact same rendering code as its shortcode counterpart. For AI agents and automated tools, two read-only abilities (search and related posts) are also exposed through the standard WordPress Abilities API on WordPress 6.9+. It also brings AI-powered related posts and an advanced keyword generator — giving your visitors smarter ways to discover content. This plugin is part of the Init Plugin Suite — a collection of minimalist, fast, and developer-focused tools for WordPress. GitHub repository: https://github.com/brokensmile2103/init-live-search What’s New in Version 2.0.0 Abilities API support (WordPress 6.9+): search and related-posts exposed as two read-only, discoverable abilities (init-live-search/search-posts, init-live-search/get-related-posts) — optional and fully backward-compatible on older WordPress versions Block Editor (Gutenberg) support: three new blocks — Live Search Box, Related Posts, AI Related Posts — under their own block category, each with a live editor preview and sharing the same rendering code as their shortcode counterparts Requires at least raised from 5.9 to 6.9 to support the Abilities API. Requires PHP remains 7.4 What’s New in Version 1.8.x & 1.9.x Optional FULLTEXT Search Index (1.9.5): opt-in MySQL FULLTEXT-indexed table for Title/Excerpt/Content, replacing slow LIKE '%term%' scans on large sites. Off by default, auto-builds in the background via WP-Cron once enabled (no SSH/WP-CLI needed), with wp init-live-search fulltext-reindex available for a manual/faster build. See “FULLTEXT Search Index” below for details. Optional Meilisearch Integration: connect your own Meilisearch server (self-hosted or cloud) as the primary search engine Typo-tolerant, relevance-ranked, sub-50ms search results straight from Meilisearch Automatic, transparent fallback to the local database search whenever Meilisearch is disabled, unreachable, or misconfigured — search availability is never compromised New Meilisearch settings tab: Host URL, Index Name, Search Key, Admin/Indexing Key, request timeout, one-click “Test Connection”, and (1.9.5) a one-click “Reindex Now” button for sites without WP-CLI/SSH access Posts are synced automatically on publish/update/trash/delete (non-blocking — won’t slow down the editor) WP-CLI command wp init-live-search meili-reindex for bulk-indexing or rebuilding the entire index Sensitive indexing key can be defined via the INIT_LIVE_SEARCH_MEILI_ADMIN_KEY constant in wp-config.php instead of the database, for extra security AI-Powered Related Posts: brand new [init_live_search_related_ai] shortcode Uses multi-signal scoring (tags, series, title bigrams, same_keyword via ACF, category, views, comments, freshness) Shares templates with [init_live_search_related_posts] (no extra styling needed) Fully filterable via new developer hooks: ai_candidates, ai_signals, ai_weights, ai_score Advanced Keyword Generator: upgraded algorithm for admin keyword suggestions Replaced TF-IDF with BM25 term weighting Added NPMI (fixed probability base) and Log-Likelihood Ratio (Dunning) for collocation strength Generates bigrams and trigrams for richer, more specific keyword suggestions Cross-document frequency penalty down-ranks phrases that are too generic across the site MMR (Maximal Marginal Relevance) selection ensures diverse, non-redundant final keywords Title-only source: clean signal, no excerpt noise, works reliably across all site types Unicode-safe, locale-aware stop words (Vietnamese & English), and soft fallback mode 404 Smart Redirect Added “Auto Redirect 404 to Best Match” mode driven by Init Live Search scoring Fully respects “Post Types to Include” settings Uses unified resolver + filters for extensible post-type handling Safety checks to avoid loops, invalid targets, and cross-type mismatches Developer Filters Expansion New filters added for AI related posts and keyword signals Complete list now includes over 20 filters (*_fallback, *_post_ids, *_results, *_weights, *_commands, etc.) Developers can hook into candidate pools, signal scores, and schema output with fine-grained control Performance Optimizations Smarter candidate pooling for related posts (recent + context-based) Pre-cached scoring loop for AI signals to minimize queries Safer regex handling in keyword processing to avoid PCRE errors Reduced memory footprint in bigram statistics without sacrificing accuracy Backward Compatible Enhancements [init_live_search_related_posts] and [init_live_search_related_ai] now share the same rendering pipeline Existing templates, schema, and CSS continue to work without modification Auto insert related posts still works and can be switched to AI mode via shortcode override Features Packed with everything a modern live search needs — and more: Live search via REST API (no admin-ajax, no jQuery) NEW: Block Editor (Gutenberg) blocks — Live Search Box, Related Posts, and AI Related Posts — under their own block category, each with a live server-side preview in the editor NEW: Abilities API support (WordPress 6.9+) — search and related posts exposed as discoverable, executable abilities via wp_get_abilities() and the wp-abilities/v1 REST namespace NEW: Optional MySQL FULLTEXT search index — faster Title/Excerpt/Content matching than LIKE queries on the built-in database search, auto-builds in the background NEW: Optional Meilisearch integration — typo-tolerant, relevance-ranked external search with automatic fallback to local DB search Smart tag-aware search mode (title + tag match) SEO metadata support: Yoast, Rank Math, AIOSEO, SEOPress, TSF ACF field matching and customizable filters Synonym expansion with fallback scoring logic Slash commands: /recent, /popular, /tag, /id, /sku, /price, /coupon, etc. WooCommerce support: search by product, SKU, brand, attribute, variation, coupon Clean modal UI with keyboard navigation (↑ ↓ Enter Esc) Optional voice input (SpeechRecognition) Tooltip Quick Search (select text to trigger) Favorites system via heart icon or /fav commands Infinite scroll and smart category filter (no extra API) Deep linking: ?modal=search&term=... New [init_live_search] shortcode: insert input or icon anywhere UI presets: default, fullscreen, topbar — with full CSS override Local caching + analytics (CSV export, no personal data) Developer-ready: custom slash commands, REST filters, JS hooks Dark Mode Support Enable dark mode for the modal by either: Adding the dark class: document.querySelector(‘#ils-modal’)?.classList.add(‘dark’); Or using a global config: window.InitPluginSuiteLiveSearchConfig = { theme: ‘dark’ }; Options: dark, light, auto Admin Settings Choose post types to include in search Enable an optional MySQL FULLTEXT search index for faster Title/Excerpt/Content matching, with automatic background indexing Connect an optional Meilisearch server (Host, Index, Search/Admin keys, timeout, one-click connection test, one-click Reindex Now) Configure modal triggers (input focus, triple click, Ctrl+/) Enable slash commands (e.g. /recent, /tag, /id) Enable support for + and - keyword operators (must-have, must-not-have) Set default slash command to run on modal open (only if slash is enabled) Set debounce time and max results Choose search mode (title-only, tag-aware, full content) Define custom ACF fields to include in search (optional) Enable Search in SEO Metadata (Yoast, Rank Math, etc.) Toggle excerpt display below each result (1-line contextual snippet) Toggle fallback logic (bigram/trim) Enable synonym expansion and manage synonym mappings (JSON editor) Enable Search Analytics to log queries (no personal data stored) Set max words for tooltip search Enable voice input (SpeechRecognition API) Enable result caching (localStorage) Choose frontend UI style (default, fullscreen, or topbar) Allow theme override via init-live-search/style.css Automatically insert related posts after content or comments (optional) Option to disable all built-in CSS completely Add default UTM parameter to result links Define or auto-generate keyword suggestions FULLTEXT Search Index (Optional) The built-in database search normally uses LIKE '%term%' queries, which can’t use a regular index and get slow on sites with a large number of posts. Tick FULLTEXT Search Index (Settings → Init Live Search → General) to sync Title/Excerpt/Content into a dedicated indexed table (your wp_posts table is untouched) and match with MySQL’s MATCH() AGAINST() instead. Once enabled, the index builds itself automatically in the background via WP-Cron — no SSH/WP-CLI needed — or you can speed things up with wp init-live-search fulltext-reindex. If your server doesn’t support FULLTEXT, or the index isn’t built yet, the plugin simply keeps using the standard LIKE-based search — nothing ever breaks. Tag search, SEO metadata, ACF fields, synonyms, and the +/- operators all keep working the same either way. Meilisearch Integration (Optional) Init Live Search works great out of the box with zero setup — the built-in database search handles everything by default. But if you want faster, typo-tolerant, relevance-ranked search at scale, you can connect your own Meilisearch instance in a few minutes. How it works: Install and run Meilisearch yourself — self-hosted (a small VPS is plenty for most blogs) or via Meilisearch Cloud. This plugin does not host or manage a server for you. Go to Settings → Init Live Search → Meilisearch, enter your Host URL, Index Name, and a search-only API key, then enable the integration. Index your existing published content either by running wp init-live-search meili-reindex via WP-CLI, or by clicking the “Reindex Now” button right on the Meilisearch settings tab (handy if you don’t have WP-CLI/SSH access) — it runs in the background (~200 posts every 5 seconds) with a live progress status. That’s it — search requests are now answered by Meilisearch. New, updated, and deleted posts stay in sync automatically. Built for safety, not lock-in: If Meilisearch is disabled, unconfigured, or fails to respond (timeout, wrong key, server down), the plugin automatically and silently falls back to the local database search — visitors never see a broken search box. Reindexing (via the button or WP-CLI) is always manual-only, and auto-stops with a clear error after 3 consecutive failed batches instead of retrying forever. Your sensitive indexing key can live in wp-config.php (INIT_LIVE_SEARCH_MEILI_ADMIN_KEY) instead of the database. Turn it off any time — nothing about your site’s core search behavior depends on Meilisearch being present. Keyboard Shortcuts Arrow Up / Down — navigate between results Arrow Right — add selected result to favorites (if not already added) Arrow Left — remove selected result from favorites Enter — open selected result or submit Escape — close modal and reset state Slash (/) — start a command instantly (e.g., /recent, /id 123) Developer Reference: Shortcodes, Filters, and Hooks Shortcodes [init_live_search] Display a search icon or input anywhere that opens the Init Live Search modal. Attributes: – type: icon (default) or input – choose between a clickable icon or a search box – placeholder: (optional) text inside the input if type="input" – label: (optional) adds a label next to the icon if type="icon" – class: (optional) add custom classes like dark, my-style, etc. – stroke_width: (optional) set the stroke width for the search icon (default: 1) – radius: (optional) override the border radius of the input form (default: 9999px from CSS; only applied if value differs) [init_live_search_related_posts] Display a list of related posts (static HTML) based on post title or keyword, optimized for SEO and fully themable. Attributes: – id: (optional) the post ID to find related posts for (defaults to current post) – count: (optional) number of posts to display (default: 5) – keyword: (optional) override the keyword used for finding related posts – template: (optional) layout template to use — default, grid, classic, compact, thumbright – css: 1 (default) or 0 – disable default CSS if you want to fully style it yourself – schema: 1 (default) or 0 – disable JSON-LD ItemList output for SEO schema [init_live_search_related_ai] Display a list of AI-powered related posts using multi-signal scoring (tags, series, title bigrams, same_keyword via ACF, category, views, comments, freshness). Uses the same templates as [init_live_search_related_posts], so no extra styling is required. Attributes: – id: (optional) the post ID to find related posts for (defaults to current post) – count: (optional) number of posts to display (default: 5) – post_type: (optional) restrict results to one or more post types (default: post) – template: (optional) layout template to use — default, grid, classic, compact, thumbright – css: 1 (default) or 0 – disable default CSS if you want to fully style it yourself – schema: 1 (default) or 0 – disable JSON-LD ItemList output for SEO schema Filters for Developers Init Live Search includes many filters to help developers customize behavior and output at various stages of the search flow. Full documentation (with code samples & advanced usage): Using Filters in Init Live Search Popular filters init_plugin_suite_live_search_enable_fallback Enable or disable fallback logic when few results are found. init_plugin_suite_live_search_post_ids Customize the array of post IDs returned from the query. init_plugin_suite_live_search_result_item Modify each result item before it’s sent in the response. init_plugin_suite_live_search_results Filter the final array of results before being returned. init_plugin_suite_live_search_query_args Modify WP_Query arguments for different commands. init_plugin_suite_live_search_ai_weights Adjust AI scoring weights for related posts. (…and more in the full docs) REST API Endpoints Fully documented, lightweight, and API-first endpoints. Ideal for headless or decoupled builds. All endpoints are under namespace: initlise/v1 Full documentation (with examples & parameters): REST API Endpoints in Init Live Search Popular endpoints /search?term=example Standard search query (uses plugin settings like post types, search mode, fallback…). /id/{id} Fetch a post by ID (returns permalink). /recent Fetch the most recent posts. /tax?taxonomy=category&term=slug-or-id Fetch posts by taxonomy (category, tag, or custom). /related?title=page-title&exclude=ID Fetch posts related to the current page title. /product?... Fetch WooCommerce products with flexible query parameters. (…and more in the full docs) License This plugin is licensed under the GPLv2 or later. You are free to use, modify, and distribute it under the same license.
Top keywords
- search84×3.51%
- live40×1.67%
- live search34×1.42%
- posts34×1.42%
- init29×1.21%
- related29×1.21%
- optional27×1.13%
- related posts24×1.00%
- init live22×0.92%
- init live search22×0.92%
- default21×0.88%
- meilisearch18×0.75%
Savanto – AI Search & Chat
A visitor searches your site for “comfortable shoes for hiking” and gets nothing back. WordPress search matches words, not meaning, so anything phrased differently to your product titles returns an empty page. Savanto replaces that search. It understands what people are asking for, adds a chat assistant that answers from your own content, and suggests related products. You do not need an OpenAI account, an API key, or any technical setup. Install the plugin, activate it, and click Connect to Savanto. You sign in or create a free account, approve the connection, and your site is live. There is no key to copy. Savanto runs the AI on its own servers, so you never manage tokens, usage limits, or billing with OpenAI, Anthropic, or anyone else. How Savanto Differs From Other AI Plugins Most WordPress AI plugins are toolkits. They connect your site to an external AI provider, but you bring your own API key, configure embeddings, choose models, and build the experience yourself. That suits developers. It does not suit a store owner who wants better search. Savanto is a finished product rather than a toolkit. There is no API key from OpenAI, Anthropic or Google to supply, no third-party AI account to manage, no vector database to stand up, and no prompts or models to choose. Everything runs on Savanto’s own infrastructure, so there are no usage bills arriving from providers you did not sign up with. Savanto uses different models for different jobs, each chosen and tuned for the task: ranking search results, holding a conversation, matching products. You get that work without configuring any of it. Other AI plugins need your own keys and some developer knowledge before they do anything. Some support platforms sell AI as an add-on from around $29 a month. Savanto gives you search and chat on a free plan that does not expire, with nothing else to set up. Paid plans start at $9 a month when you need higher limits. Semantic Search Replaces WordPress search with search that reads intent. Someone types “something warm but not too bulky” or “gifts under $50 for someone who likes cooking” and gets sensible results. Typos, vague descriptions and full questions all work without tuning. Chat Assistant An assistant on your site that answers questions from the content you have synced, suggests products, and points people to the right page. It works from your own material, so it is not guessing at what you sell. Recommendations Suggestions drawn from what a visitor is actually looking for, rather than from cookies or browsing history. Related products, posts and pages that a category-based list would have missed. Live Agent Handoff The assistant recognises a question it should not answer alone and hands the visitor to your support team, so nobody reaches a dead end. Automatic Sync Publish a page, product or post and Savanto picks it up. Nothing to re-index by hand, and search results stay current as your content changes. Built for WooCommerce Savanto reads your whole catalogue: titles, descriptions, prices, categories and custom fields. Visitors search in their own words and still find the right product, which means fewer searches that end in nothing. Privacy Savanto indexes only publicly available content. It does not touch private posts, drafts, user accounts or personal data. Performance Search runs on Savanto’s infrastructure with global edge delivery, so it adds no load to your WordPress hosting. Responses typically return in under 100ms. Enterprise plans include a contractual uptime SLA with service credits. Compatible With WordPress WooCommerce Elementor Gutenberg Advanced Custom Fields (ACF) Developer API available at savanto.ai/docs/api for custom integrations. Getting Started Install the plugin, activate it, and click Connect to Savanto. You are live once you have signed in and approved the connection, and Savanto begins indexing your content. Most sites are fully searchable within minutes. Guides for installation, search, chat, recommendations and troubleshooting are at savanto.ai/docs/wordpress. The free plan needs no credit card and does not expire. Paid plans start at $9 a month. Use of Third Party Services This plugin connects to Savanto’s servers to provide its search and chat features. When activated, your publicly available site content (pages, posts, and products) is sent to Savanto’s servers for indexing and processing. Savanto – Terms of Service | Privacy Policy No data is shared with OpenAI, Anthropic, Google, or any other third party AI provider. All AI processing is handled through Savanto’s own infrastructure. Privacy This plugin connects to the Savanto cloud service (savanto.ai) to provide its search and chat features. Data Sent to Savanto When you connect and sync content, the following data is transmitted to Savanto servers: Post and page content (title, body, excerpt, permalink) Product information for WooCommerce (name, description, price, SKU, categories, images) Taxonomy terms (categories, tags, product attributes) Post meta that is marked as public Data NOT Collected User personal information (emails, names, addresses) Order details or customer data Passwords or authentication credentials Private or draft content (only published content is synced) Comments or user-generated content Chat Widget The chat widget loads from Savanto’s CDN (cdn.savanto.ai). When visitors use the chat: Their questions are processed by Savanto’s AI No personally identifiable information is collected Chat conversations are not stored unless you enable analytics in your dashboard MCP Server (WooCommerce) If you enable the MCP Orders domain for WooCommerce: Order lookups are performed locally on your WordPress server Order data stays on your server and is not sent to Savanto The Savanto cloud only routes requests to your server How Order Lookups Work: When a customer asks about their orders through the chat: The question is analyzed by Savanto’s AI in the cloud If it’s an order question, Savanto sends a request back to YOUR WordPress site Your WordPress plugin looks up the order using WooCommerce Only the relevant order summary is returned to complete the AI response The full order database is never accessed by or copied to Savanto servers. Data Retention Synced content remains in Savanto’s cloud until you disconnect or delete it Disconnecting removes all your indexed content from Savanto servers You can request complete data deletion by contacting
[email protected] External Service This plugin connects to the Savanto cloud service to provide its search and chat features. The plugin and cloud service are developed by the same team. By using this plugin, you agree to Savanto’s: Terms of Service Privacy Policy GDPR Compliance Savanto is designed with privacy in mind. Since no personal user data is collected or transmitted, standard GDPR consent mechanisms for visitor data are not required. However, you should update your site’s privacy policy to disclose that content is processed by a third-party AI service.