this onboarding idea from @ExaAILabs is brilliant, get the user to complete your setup via offering credits i generally skip onboardings but this worked on me

the economy for agents is developing quickly

"We evaluated five different search providers, but none came close to Exa's quality and breadth of data" -Arda Bulut, CTO of Hockeystack (AI sales marketing agents)

A new internet economy, built for agents: Agents break the old economics of the internet - when everyone has an AI assistant consuming and taking actions, what happens to ads? Exa Connect is a step toward a new internet economy built on data, not ads. It's a marketplace where data providers choose what to charge for their data, and developers choose which data providers their agents use. We feel this is the right economic model -- market-based, transparent, value-aligned, and as agent requests scale >1000x over the coming years, will be a more valuable market for providers than the ad market is today. This is also a good path forward for the internet. It's one that monetizes transformer attention, not human attention. We don't know exactly what that great intercontinental information exchange called the internet will become, but it's going to change, and we think very much for the better.

Today we're launching Exa Agent: Opus/GPT 5.5 quality web research at 2-10x lower cost. It's our most powerful endpoint, particularly good at deep research and list-building - as cheaply as possible. Basically you can now use a deep research API for close to the cost of a search API. As the world wrestles with the cost of big models, we'll keep pushing the price/performance frontier 🫡

most knowledge work tasks aren't bottlenecked by intelligence anymore, they're bottlenecked by access to knowledge

I should be able to call Bruce Wayne or Einstein and ask them for advice. The app would download everything about that person them speaking, and then would need realtime retrieval to answer realistically. Has to be built well. Does this exist / can someone please build?

had a ton of fun chatting with @sarahdingwang about where search for AI is going. In short, >1000x human search volume, paragraph-sized queries, comprehensive results expected, no misinformation, search shifting to infrastructure

Excited to announce we raised 250M led by a16z at a 2.2B valuation. The past 5 years have been a wild ride. Here's where we're going:

RL-ing models to use web search is critical for optimal deep search performance. Most studies keep the web search tool constant. Here we test how RL-ing on different search tools impacts performance:

10^18 🤝 10^100

This is a remarkable result in fast extraction models, a step toward an internet hyperoptimized for agents. Agents can now instantly extract the right tokens from a webpage, which cuts down customer LLM input cost by 20x. Proud of the team for a year-long research effort 🫡

Exa Deep Max is SOTA quality on the hardest search benchmarks, while being 20x faster (minutes instead of hours) This is possible because it recursively calls our lowest-latency retrieval and token-extraction tools. This was the plan when we started researching fast retrieval years ago - cool to see it play out. As our low-latency endpoints rapidly advance in quality the next few months, every high latency endpoint will improve accordingly.

Google has one search engine for all, but Exa has one per customer. To handle all possible customizations, we built internally a graph language to compile and trace search paths. Blog post here:

Agents can now decide which tools to use and pay themselves. So the best product for agents wins. In this case, the agent chose Exa for web retrieval.

Agents are entering the economy in droves and require tons of new infra. Excited to partner with @coinbase on search payments infra

Puppies are coming to the Exa office. We strongly believe that puppies searching for tennis balls will spark conversation about embeddings, rerankers, and new retrieval architectures. If you're an ML researcher (or a puppy), come April 12th, link below👇

What if you could monitor every situation? Mission control for the world. It's the kind of power presidents should have. We want to give that power to everyone. You (and your AI agents) can now monitor: - competitor announcements - engineers leaving AI labs - happy political news across the globe - new research papers in retrieval containing similar ideas to ones your team has found - literally any semantic query Under the hood, we run Exa searches on a recurring schedule (1 hour, 1 day, etc.) and deduplicate to only new information that matches your query. This is a step toward an agentic world that transitions from pull to push, where your AI assistant keeps you always up-to-date. Being lost in a sea of information as we all are now is not a necessary facet of modern life. It's a temporary problem. Exa's goal is to give everyone the ability to be fully informed, and that'll make us more reasonable decision-makers and citizens. We need to get there soon.

roles here, gonna be a very cool office: exa.ai/careers?location=sg