Web data should be a primitive,
not a project.
Brime is the umbrella API for agents that need to read the web. Search, extract, and research — one SDK, one credit balance, one bill.
Every agent eventually needs to read the web.
We kept watching teams reinvent the same plumbing. A scraper for one tenant, a PDF parser for another, a hand-rolled research loop stitched out of three SaaS APIs and a queue. Months of integration work for something that should be a line of code.
We thought reading the web should take a line of code, not a quarter.
So we built Brime as the umbrella. Three primitives behind one SDK, one credit balance, one bill. Search the open web, extract from URLs and documents, run multi-step research — with traces you can hand to your team, citations you can show your customers, and pricing that scales with what your agent actually consumes.
What we hold ourselves to.
Four rules we wrote down on day one. They show up in the product, the docs, the changelog, and the bill.
Primitives over products.
The agent decides what to do. We hand it cleaner verbs — search, extract, research — and stay out of the way.
Trace everything.
If you can't see the call, you can't trust the answer. Every response carries a trace ID, every source is cited, every step is replayable.
One credit, three jobs.
No per-seat surcharge, no usage gotchas, no add-ons. The same balance pays for search, extract, and research — at the same rate, on the same bill.
Public roadmap, public quality.
Catch-up gets shipped in view. The changelog is the source of truth — what landed, what's queued, what we got wrong and rolled back.
A small distributed team. Engineers, designers, and support, working across multiple timezones.
Logos, brand colors, and boilerplate copy. For working media, press@brime.dev gets a same-day reply.
- Logo zip· Coming
- Brand colors PDF· Coming
- Boilerplate.txt· Coming