Smart search for your catalog
Hybrid search (BM25 + embeddings) with an optional AI layer that understands intent, recommends products and rescues zero-result searches. It plugs into any store with an embeddable widget.
76,000+
Indexed products
~60 ms
Per search
BM25 + AI
Hybrid engine
The search engine that turns your catalog into sales
Native search engines return zero results on a typo or a natural-language query, and search SaaS tools are black boxes that charge you per query without ever giving you your data. AgentSearch combines the precision of BM25 with the semantic understanding of embeddings, adds a per-channel AI layer and hands you back control: your engine, your synonyms, your curations and your analytics.
Relevant results from the very first keystroke
Instant suggestions, typo tolerance and results ranked by real relevance, not by exact text match.
- Hybrid BM25 + vector search: finds what the customer means, not just what they type
- Real-time autocomplete and suggestions as you type
- Typo tolerance and synonym understanding
- Responses in ~60 ms even with catalogs of 76,000+ products
A search chat that recommends products
The AI layer (Azure OpenAI gpt-4.1) is optional and enabled per channel. It understands query intent and chats to recommend exactly what the customer needs.
- Interprets natural-language queries and infers intent
- Recommends products in a conversational chat
- Rescues zero-result searches by suggesting alternatives
- Enable or disable it per channel, without touching the search engine
Integrates into any store without touching the theme
An embeddable JS widget (Shadow DOM, ~30 KB, no React) that injects into Shopify, PrestaShop or any platform. With dynamic facets and filters that adapt to your catalog.
- ~30 KB widget with Shadow DOM: zero conflicts with your CSS
- Dynamic facets by brand, category, type and attributes
- Editable synonyms and curations: pin or hide products and add banners
- Install with a single line of code, no template changes
Everything a modern search engine needs
A single engine covering relevance, AI, integration, editorial control and measurement.
Hybrid search
Typesense engine that combines BM25 (text relevance) with vector embeddings for semantic understanding. The best of both worlds in a single query.
AI layer and chat
Optional, per-channel AI that understands intent, suggests results when there are zero matches and offers a conversational chat that recommends products.
Embeddable widget
A ~30 KB JS widget with Shadow DOM, no React, that injects into any store (Shopify, PrestaShop and more) without touching the theme or causing style conflicts.
Facets and synonyms
Dynamic facets by brand, category, type and attributes, plus editable synonyms and curations: pin or hide products and add banners whenever you need to.
Search analytics
Full, first-party analytics: queries, zero-result searches and the search → click → cart → purchase journey. Your data, with no SaaS black boxes.
Multichannel
A single engine serving several channels at once. Index once, search from anywhere, with independent configuration per channel.
Ready to search in three steps
From your catalog to an AI-powered search engine without rewriting your store.
We index your catalog
We connect your product source and index your entire catalog into Typesense, with hybrid BM25 + embeddings search ready from the start.
We tune relevance and AI
We set up facets, synonyms and curations, and enable (or not) the AI layer and the conversational chat channel by channel to fit your business.
We integrate the widget
We inject the embeddable widget into your store with a single line of code. You start searching in ~60 ms and measuring every query from day one.
Frequently asked questions
Which platforms can AgentSearch integrate with?
Any of them. The widget is pure JavaScript with Shadow DOM (~30 KB, no React), so it injects into Shopify, PrestaShop, WooCommerce or custom builds without touching the theme or causing style conflicts.
Is the AI layer mandatory?
No. The AI (Azure OpenAI gpt-4.1) is optional and enabled per channel. You can start with just the hybrid BM25 + embeddings search and add intent understanding, zero-result suggestions and the conversational chat whenever you want.
Does it work with large catalogs?
Yes. In real production it serves a store with more than 76,000 indexed products, responding in around 60 ms per search and serving several channels with a single engine.
Do I have access to my search data?
Absolutely. AgentSearch includes first-party analytics: queries, zero-result searches and the full search → click → cart → purchase journey. It replaces the black boxes of SaaS search engines and gives you control of your data.
Beyond the search box
Services to deploy, migrate and integrate AgentSearch into your operation without friction.
Search for your ecommerce
We deploy the search engine in your online store (Shopify, PrestaShop, WooCommerce or custom) and tune it to turn every search into a sale.
Learn moreMigrate from your current search
We migrate from Algolia, Doofinder or any SaaS with no downtime and no data loss. We keep your history, synonyms and curations, at a better price.
Find out moreWidget & integration
An embeddable ~30 KB JS widget with Shadow DOM that drops into any store without touching your theme. One engine for all your channels.
See moreTurn your catalog into a search engine that sells
Tell us about your catalog and we'll show you AgentSearch in action with hybrid search and AI in a tailored demo.