An AI search plugin that works on any website — not just the big platforms

Every website has a search function. Almost none of them work well. The search bar on a typical site — whether it's built on WordPress, Shopify, Wix, Squarespace, or a custom stack — does one thing: keyword matching. A visitor types words, the search engine checks which pages contain those words, and it returns a ranked list. This has been the model since the late 1990s, and it has barely evolved. The ranking algorithms have gotten more sophisticated, the indexing has gotten faster, but the fundamental interaction — type keywords, get links — hasn't changed. Meanwhile, how people search has changed dramatically. Visitors in 2026 don't think in keywords. They think in needs, questions, and descriptions. "Something for my sister's birthday under $50." "Which plan includes API access?" "Show me projects similar to the loft renovation in Brooklyn." These are real searches that real visitors type into real search bars, and they almost always return garbage — or nothing.

The gap between how people search and how search widgets work is where revenue disappears. Studies consistently show that visitors who use site search convert at 2-3x the rate of visitors who browse — because search signals intent. A visitor who searches is telling you what they want to buy. But that conversion premium evaporates when the search returns irrelevant results. A visitor who searches "lightweight stroller that folds small for travel," gets a page of tangentially related results, and leaves is the most expensive kind of lost sale: a motivated buyer who told you exactly what they wanted, and you failed to match it. Multiply that across hundreds of daily searches and the leak is substantial — invisible because your search analytics only track what was searched, not what wasn't found.

Platform-specific search plugins have tried to improve on this. Shopify has apps like Searchanise, Algolia, and Doofinder. WordPress has Relevanssi, SearchWP, and ElasticPress. Wix has its native site search. Each improves the ranking algorithm for its specific platform — better weighting of product attributes, faceted filtering, typo tolerance. These are genuine improvements. But they're still keyword-matching engines with better math. They still return a list of links on a search results page. They still require the visitor to translate their intent into keywords the search engine can process. And they only work on the platform they were built for. If you run sites across WordPress and Shopify, you need different search solutions for each, configured separately, with different feature sets and different price points. There's no unified search experience across platforms.

An AI search plugin changes the interaction model entirely. Instead of a search box that returns links, visitors interact with a conversational AI that understands natural language, searches your content and product catalog, and presents results visually — as Dynamic Product Cards with images, prices, and action buttons — right inside the chat. The search happens through conversation, which means it's iterative. "Do you have running shoes for flat feet?" shows product cards. "Something lighter, for indoor use." The AI narrows the results. "How does that compare to the CloudStep?" The AI provides a side-by-side. Each exchange builds on the previous one. The visitor refines their search by talking, not by retyping queries and clicking back.

What makes this work as a universal plugin — not tied to any single platform — is that the AI's knowledge comes from two sources that are platform-independent. The first is auto-crawl. You paste Asyntai's JavaScript snippet into your site, and the AI crawls your pages: product listings, service descriptions, blog posts, FAQ pages, documentation, landing pages. It reads what's publicly visible and builds a searchable knowledge base from it. This works the same way whether your site runs on WordPress, Shopify, Wix, Squarespace, Webflow, Magento, or static HTML. The crawler doesn't care what CMS generated the page. It reads the content, understands the structure, and indexes it for natural language search.

The second source is the Real-Time Data Feed — a URL you provide that returns your product catalog as structured data (JSON, CSV, or an API endpoint). This is where the AI gets current prices, stock levels, image URLs, and product attributes. The feed is platform-agnostic by design: any system that can export products as JSON or CSV works. WooCommerce has a REST API. Shopify has a product CSV export. BigCommerce, Magento, and PrestaShop all have export functions. Even a Google Sheet published as CSV works. The AI indexes the feed and uses it alongside crawled content to answer product searches with accurate, real-time data. When a visitor asks "do you have that in blue," the AI checks the feed — not a cached index from last week.

Dynamic Product Cards are the visual layer that turns search results into a browsing experience. When the AI finds products matching a visitor's query, it doesn't just list names and links. It renders each product as a visual card inside the chat: product image, name, price, a short description, and an action button. Multiple matches display as a horizontal carousel the visitor can swipe through. The cards make the chat feel like a curated product display rather than a text exchange. For ecommerce sites, the cards show prices and "View Product" buttons. For service businesses, they can show service descriptions and "Learn More" buttons. For portfolio sites, project images and "View Project" links. The card format adapts to whatever content the AI surfaces.

The conversational format solves a usability problem that even the best search results page cannot: contextual refinement. On a traditional search results page, every new search starts from scratch. The visitor types a query, reviews results, goes back, types a different query, reviews again. There's no memory between searches, no building on previous context. In a conversation with the AI search plugin, refinement is natural and cumulative. "I'm looking for a winter jacket." The AI shows options. "Something warmer, for below-zero temps." The AI narrows down. "Do you have the Expedition Parka in large?" The AI checks stock. "What's the return policy on outerwear?" The AI answers from the site's FAQ page. One conversation, four different types of queries, each building on the last. No search bar could handle that sequence because search bars don't have memory.

For site owners who operate across multiple platforms, the single-snippet model eliminates fragmentation. An agency that manages client sites on WordPress, Shopify, and Squarespace doesn't need to learn three different search plugins, maintain three different configurations, or pay for three different subscriptions. One Asyntai account, one snippet per site, and the same AI search experience everywhere. The configuration — knowledge base sources, data feeds, Custom Tools, widget appearance — is managed from the Asyntai dashboard, not from within each platform's admin panel. This is especially valuable for businesses that run a main site on one platform and a blog or documentation site on another. The AI search works across both, using content from all crawled sources.

Custom Tools extend the AI search plugin beyond discovery into action. Once a visitor finds what they're looking for, the AI can do more than link to a product page. With Custom Tools, it can call your own API endpoints: check real-time inventory for a specific size, calculate shipping costs to a given postal code, look up an order status, verify whether a promotional code applies. Each tool is an HTTP endpoint you configure in the Asyntai dashboard — the AI calls it during the conversation when the visitor's question requires it. This turns the search widget into a full shopping assistant or support agent, capable of answering questions that require live data from your backend. And because Custom Tools use standard HTTP, they work regardless of your platform — any system that can expose an API endpoint can power a Custom Tool.

Language support is built in, not bolted on. The AI search plugin supports 36 languages natively. It detects the visitor's language — from browser settings, from the page language, or from the language of their query — and responds accordingly. A German-speaking visitor asks "Gibt es eine leichte Regenjacke unter 100 Euro?" and the AI searches your English-language product catalog, finds matching lightweight rain jackets under the equivalent price, and responds in German. The product cards show data from your feed (product names, prices in your currency, images), while the AI's conversational text is in German. You don't need to maintain a translated catalog. You don't need a multilingual search plugin. The AI handles the language mapping automatically, which means your single-language product catalog is instantly searchable by visitors in 36 languages.

The question of whether an AI search plugin is worth the investment comes down to a simple ratio: how many visitors search your site, and how many of them find what they're looking for? Most site analytics tools track the first number but not the second. You can see that 15% of your visitors use search, but you can't easily see how many of them found a relevant result. The "zero results" metric is a start, but it misses the more common failure: searches that return results, just not relevant ones. A visitor who searches "gift for someone who likes coffee" and gets a page of loosely related results might click on one, realize it's not right, and leave. That search "worked" by the platform's metrics — it returned results and generated a click — but it failed the visitor. AI search doesn't just reduce zero-result searches. It increases the quality of every search result by understanding what the visitor actually means.

Installation complexity is the most common objection to adding a new search solution, and it's the objection that matters least here. The AI search plugin is a single script tag — fewer than 100 characters of code. You paste it into your site's HTML. On WordPress, that means dropping it into a header/footer plugin or using the Asyntai WordPress plugin. On Shopify, it goes into theme.liquid or through the Asyntai Shopify app. On Wix, it's the Custom Code section. On Squarespace, Code Injection. On a custom site, paste it before the closing body tag. There's no package to install, no build step to run, no API key to configure on the server side, no database table to create. The snippet loads asynchronously, so it doesn't affect page load speed. It doesn't conflict with existing search functionality — the AI chat widget sits alongside your current search bar, giving visitors a second, more capable search path.

The difference between AI search and traditional search becomes most visible with ambiguous or complex queries — which are exactly the queries that indicate the highest purchase intent. "Something for my living room that ties together gray furniture and warm wood tones." "A laptop for my daughter starting college, needs to run design software." "Waterproof speaker that sounds decent, not too big, for camping." These are real purchasing decisions expressed as natural language. No keyword search can handle them because they're not keyword queries — they're descriptions of needs that require reasoning to match against a product catalog. The AI reads the description, infers the relevant attributes (color compatibility, software requirements, size and durability), and searches the catalog for products that match those inferred criteria. Then it presents the matches as visual cards, so the visitor can evaluate them immediately.

Content-rich sites benefit from AI search in ways that ecommerce stores might not immediately recognize. A law firm with 200 articles about different legal topics can give visitors an AI search that answers "do I need a permit to build a deck in Ontario?" by finding and summarizing the relevant article. A university with thousands of program pages can help prospective students who search "master's program in data science with co-op" find the right program without navigating a complex directory. A recipe site can handle "quick weeknight dinner with chicken and whatever's in a basic pantry" better than any category filter. In each case, the AI search plugin doesn't replace the existing site navigation — it augments it with a natural language path that works especially well for visitors who know what they want but don't know where to find it on your site.

The economics of site search have always been lopsided. Visitors who search are your highest-intent audience, yet most sites spend their design and development budget on navigation, landing pages, and checkout flows. The search function gets a default text input and whatever results the platform generates. An AI search plugin rebalances that equation by giving your highest-intent visitors a search experience that matches their intent level. They're ready to buy or engage — and now the search actually helps them get there. For the cost of a Standard plan, every search on your site becomes a conversation where the AI understands the query, searches your live catalog, shows visual results, and follows up. It's not a better search bar. It's a different kind of search — one that works the way your visitors already think.

The sites that see the most impact from AI search share a common trait: their visitors describe needs rather than products. Ecommerce stores where shoppers say "something comfortable for standing all day at work" instead of searching for a specific brand and model. Service businesses where clients describe situations — "I think my furnace is making a weird noise, it's about 15 years old." Educational platforms where learners search by goal — "learn Python for data analysis, I know some SQL already." Real estate sites where buyers search by lifestyle — "family-friendly neighborhood near good schools, under $500K." In every case, the visitor's query is rich with intent that keyword matching throws away. The AI search plugin captures that intent, reasons about it, and delivers results that match what the visitor actually means. That translation — from visitor intent to relevant content — is where the value lives, and it works the same way regardless of what platform your site is built on.