managed searxng + ai synthesis
Search like nobody's
watching.
VeilSearch queries 70+ engines through open metasearch, then an AI writes you one answer with citations. No tracking, no profiling, no server of your own to babysit.
3 free searches, no account. 50/month free with one.
70+ engines, zero fingerprints
Every query fans out across Google, Bing, Brave, DuckDuckGo, Wikipedia and dozens more through a managed SearXNG instance. The engines see our server, never you.
AI answers with receipts
A model reads the results and writes a synthesis with inline [1] citations. Click any citation, land on the source. No hallucinated links, no black box.
Zero DevOps
You get SearXNG without the server, the updates, the engine breakage or the rate-limit whack-a-mole. We run it, you search.
One query in, one cited answer out.
- 01 /you ask
Query leaves your browser over TLS. That's the last time it's yours alone.
- 02 /we fan out
Our SearXNG instance hits 70+ engines. They see us, not you.
- 03 /ai reads
The model reads the top results — and only the results.
- 04 /you decide
A synthesis with [n] citations. Click through, verify, done.
Free to try. ₹999/mo to go unlimited.
50 AI-summarized searches a month, free forever. Pro gets a private instance, unlimited AI answers and workspace export — for less than a VPS you'd have to maintain yourself.
Questions journalists ask us.
- How is this different from self-hosting SearXNG?
- It is SearXNG — kept updated, tuned and monitored for you. You skip the VPS, the broken-engine maintenance and the anti-bot cat-and-mouse, and gain an AI answer layer SearXNG doesn't have.
- What do you log?
- Not your queries and not your results. We store a monthly counter per account for quota enforcement — that's the entire data model. See the privacy policy for the exact schema.
- How is this different from Perplexity?
- Perplexity builds a profile of you; we can't — there's nothing to build it from. Queries hit open metasearch, the AI layer sees results only, and nothing is retained.
- Which AI model writes the answers?
- A frontier model routed through a gateway (currently Claude). Answers cite every claim, and result content is treated as untrusted input.