
exa.ai
Search API for AI applications
5.0•21 reviews•870 followers
Search API for AI applications
5.0•21 reviews•870 followers
Find perfect lists of leads, candidates or anything in the web. Websets’ semantic search lets you ask in plain English, it agentically sources results and enriches them with you ask for - like emails, tags, and more - all in the time it takes to brew a coffee.
This is the 4th launch from exa.ai. View more
Exa Websets V2
Launched this week
Websets v2 is Exa’s update to its tool for building custom datasets from live web data. Describe what you want in plain language and get rows with sources and reasoning for each match. Per Exa, v2 finds 20% more matches, runs 4x faster and costs about 2x less than v1. The monthly subscription is gone, usage draws from Exa credits, searches are $5 per 1,000 requests, and API access is open to all users on the usage plan.




Launch tags:Artificial Intelligence
Launch Team

AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted






Exa just released Websets v2, a major update to its tool for building custom datasets from live web data. You describe what you want in plain language, and it returns a list where each row comes with sources and reasoning for why it matches.
What changed from v1:
It finds about 20% more matches, with 20% better accuracy and recall
It runs 4x faster
It costs about 2x less
The monthly subscription is gone, and usage now draws from your existing Exa credits
Websets API access is open to all users on the usage plan
What it does:
Searches the web and checks each result against your requirements
Adds enrichment columns for new details, like a paper’s code repo or a company’s recent initiatives
Lets you review the sources for every result and export the dataset
Draws on Exa’s database of 1B+ people, 70M+ companies, 1.2B+ products and 500M+ research papers
Exa’s example datasets include healthcare companies mentioning AI in 10-K filings, UK publications on AI in ecommerce, and 1,000 YC startups with batch and industry data.
In Exa’s benchmark of 20 queries each, against Claude Code, Gemini 3.1 Pro and Parallel FindAll, Websets averaged 25.7 verified matches per company query at $0.024 per match, and 30.65 per people query at $0.0169 per match. These are Exa’s own tests on a small query set, so try it on your own searches.
Pricing is now usage-based, with searches at $5 per 1,000 requests. Existing Starter and Pro users can switch on the billing page, or stay on their current plan and keep their current pricing.
Good fit if you build lead lists, market maps or research datasets and want sourced rows without a monthly subscription.
Learn more: Exa Websets v2 announcement
Congrats on launch number four, nice work.
how fresh can a webset get, like could i ask for people who posted about needing a mobile dev in the last week
A rolling time window like that could make timely intent signals much more actionable. Can date filters be combined with role or location criteria?
The part that caught my eye is every row arriving with its sources and the reason it matched. I spend a lot of my time checking agent output by hand, so a dataset I can actually audit without re-running the search is the difference between using it and not. Curious how the reasoning holds up on the messier queries — the ones where two companies look nearly identical on paper.
Each row arriving with its sources and a reason it matched is what would let me hand a dataset to someone else without a second round of checking every row by hand. Does the reasoning stay attached when I export, or is it only visible inside the tool?
Dropping the monthly sub for usage-based makes sense. How do you handle rows where sources disagree, does the reasoning column show both or just the pick?