Introducing Exa Connect: connecting agents to data beyond the public web.
Available today with ZoomInfo, Crunchbase, Similarweb, and many other leading data providers.
https://t.co/MlEhWoCIiF
Introducing Exa Agent: frontier web research at less than half the cost of GPT 5.5 and Opus.
/agent orchestrates a mixture of cost-effective models to complete any web research task, from simple data enrichments to building gigantic lists.
Exa raised $250M at a $2.2B valuation, led by a16z, to continue organizing the web for agents:
- Exa now serves search to Cursor, Cognition, Openrouter, 5000+ other companies, 500k+ developers
- Weโre SOTA in many important verticals (code, companies, people, news, more very soon)
- We make agents smarter and cheaper by returning 90% less text with little to no tradeoff in RAG quality
- Weโre building out web agents that are Pareto optimal on price x performance x latency, possible because we own our search stack fully end to end
We used to tell candidates that without innovation in search, we may live in a world where we have both AGI and fake news. Funnily enough, I think that weโve now been living in such a world for quite some time. With this funding, we should be able to dramatically improve the state of information in society.
๐๐๐
go to bed
right now
i know the build is almost finished
the eval can wait til morning
the agent will still be failing tomorrow
you won't figure out why it's hallucinating
yes your coworker ships on 4 hrs of sleep
they also hallucinate a lot
off you go
Exa now reduces input tokens for web agents by 96%.
We trained a text extraction "Highlights" model to dynamically select only the most relevant tokens from a webpage, for a given query. Only 500 tokens of highlights are needed to match the RAG performance of a full 10K token webpage.
We find Highlights to be particularly useful for frontier models like GPT 5.5, where content density enables longer horizon tasks by preventing context bloat. You can use it today by using the "highlights" content type in our API!
i really want something like conductor that actually works and isnt the most buggiest piece of crap there is, literally half the time making a worktree doesnt work and claude performs so much worse on it, anyone know if anything better
A search engine has many moving parts, and they constantly change with scale of index and traffic.
In this blog post, we dive into the code-level abstractions we've built to develop search infrastructure that is both performant and a joy to work with.
https://t.co/vf7i7NM1gC
@Steve_Yegge Maybe tell your buddy to do some actual work and to stop spreading absolute nonsense. This post is completely false and just pure clickbait.
@Steve_Yegge Maybe tell your buddy to do some actual work and to stop spreading absolute nonsense. This post is completely false and just pure clickbait.