Search engine for your data
Search over your data that gets better every time an agent uses it.
Relevan indexes your documents, products, profiles, and proprietary data behind an agent-native search API. When agents retry, refine, and find the right result, Relevan learns what worked, so the next search gets there faster.
Beta access is limited. New sign-ups join the waitlist when the round is full.
Think Algolia, Elasticsearch, or OpenSearch, built for AI agents, with relevance learning included.
Live comparison
Regular search vs. Relevan
This is your catalog, indexed in Relevan. Both agents start from the same query. One keeps broadening it and loses precision: three searches, ending at 22 results with a portable speaker tied for first place. The other reads which words matched, drops the one that matched nothing, and lands it in two searches and 6 results. That recovery is the first half of the loop. The section below shows what Relevan keeps from it.
noise-cancelling headphones for travel
Run the search to compare regular search with the Relevan Protocol.
This walkthrough is scripted against a fixed product catalog, so it plays the same way every time. Every query, score, and protocol response is replayed from the real search engine — a test fails if any of them drifts. Create an index in the console to run it against your own data.
The learning loop
Every search teaches the next one.
Relevan sees the whole session on its own: every query an agent sends, and every refinement it makes when the first attempt misses. You report what the result led to with one call to the feedback endpoint, as a click, an agent judgment, or a conversion. Relevan keeps what those two signals agree on.
The demo above ended on a gap: travel matched nothing in this catalog, so the agent had to work out what the word meant. This is what Relevan learns from that session.
The next agent that asks for something similar starts where the last one finished.
What Relevan is
Not web search. The search engine inside your product.
Relevan does not crawl the web or read pages for you. Relevan searches the data you load and control: your documents, in your index, behind your API key. It holds the place Algolia, Elasticsearch, and OpenSearch hold today, with two differences. The response tells an agent why a result matched and what to try next. And the ranking learns from the searches your agents already run.
What you can search
Load the data your agents keep guessing about.
Product catalogs
Every item, attribute, and variant your agents must find and compare.
Documentation
Guides, references, and release notes your support agents answer from.
Recruiting profiles
People, skills, and experience, matched on what a role actually needs.
Support content
Tickets, macros, and help articles behind an agent that resolves them.
Internal knowledge
The documents your company holds, searchable without a new pipeline.
Fewer searches. Better answers. Relevance that improves.
Fewer searches per answer
Agents recover on the second try instead of the fifth.
Lower token spend
Stop burning tokens on arbitrary query rewrites.
Better answers
Give agents the right evidence before they respond.
Relevance that improves
Every search your agents run makes the next one land closer.
Index your data. Let your agents teach it.
Relevan is a search API for teams building agentic search, internal knowledge agents, and AI product experiences. Load your data and see where retrieval breaks today.
- Create an index in the console
- Load your documents through the ingest API
- Search the index from your agent
- Report what the result led to with one call