Use case · Product teams building a chat-over-the-web experience
A Perplexity-style answer UI
A Rust backend that fetches a handful of pages per query and returns them shaped for prompt-stuffing, so your UI owns the chat and the citations.
The problem
You want a "search the web and answer with citations" experience inside your own product. The hard, repetitive part is the middle: take the query, get URLs, fetch each page, strip the boilerplate, chunk it, embed it, and keep every chunk tied to its source URL. Writing that in the hot path, per query, is exactly the plumbing nobody wants to own.
How Polymathy helps
- One call to GET /v1/search returns the top ten pages already fetched, chunked, and embedded.
- Each chunk arrives as (source_url, text), so your citation footnotes are a projection of the response, not a separate bookkeeping problem.
- Your UI and prompt template stay in your codebase; Polymathy is just the retrieval box behind them.
What the flow looks like
Query → Polymathy → chunk map → your prompt template → your LLM → answer + citations rendered in your UI.
A good fit when
- The corpus is the public web, reached through SearxNG
- You want to own the chat UI, prompt, and citation rendering
- Your stack is Rust or you want a single self-hosted binary
Reach for something else when
- You want a turnkey chat product you can deploy today (look at Onyx)
- You need a hosted SaaS with a dashboard and SLAs
Other use cases
An internal-docs answer engine
Point SearxNG at your internal indices and turn matches into cited chunks for a private assistant your employees can trust.
Read →A cited research assistant
Ground every generated paragraph in real source URLs, because the chunk-to-URL mapping is enforced by the response shape.
Read →A content-processor test harness
A stable HTTP harness that exercises your content-processor service against real-world URLs drawn from arbitrary queries.
Read →