AstroFabric vs Exa Websets: the search engine that builds the list or the agents that finish it

Exa is a search engine built for AI rather than for people, and Websets is what happens when you point that engine at list-building: write the query, set the count, add the criteria every result must satisfy, and an agent checks each candidate against each criterion and shows its reasoning and references. It is the best answer available to the question "find me every company like this on the web, and prove it." AstroFabric asks the next question: once the list exists, who fills the fields with a source on each, verifies the emails, keeps watching for the signal that says it is time, builds the audience, and delivers the result into the CRM under an approval? The comparison is between a search product with verification built in and an agent platform that treats the list as the first step of a mission. Public information as of September 2026; verify current numbers on their pricing pages.

The short version

Websets is the strongest way to turn a description into a verified list from the open web, and it shows its work on every row. AstroFabric takes a list like that, or builds its own, and runs the rest of the job: fields with sources, verification, signals, audiences and governed delivery. A verified search result versus a standing data chain: that is the whole comparison.

What Exa Websets actually is

A Webset starts with a query in plain language and a target item count. You add up to five criteria that every result must satisfy, and the engine finds candidates and evaluates each one against each criterion, returning a yes, no or unclear with the reasoning and the references it used. Only matching results become items. Then enrichments fill extra columns per item - text, dates, numbers, options, emails, phones and URLs - again with reasoning and references attached. Monitors re-run a Webset on a schedule of at most once a day and either append or replace results, CSV imports let you enrich or exclude your own rows, and webhooks fire on eighteen event types. Exports are one-way today: people to Salesforce leads or HubSpot contacts, people to Instantly, Smartlead and lemlist, and any entity to Clay by webhook, plus CSV. Developers get a REST API, Python and JavaScript SDKs and a Websets MCP server with two dozen tools. Pricing, as published in Exa's docs in September 2026: a free tier with 1,000 credits and Websets capped at 25 results, Core at $49 a month with 8,000 credits, Pro at $449 a month with 100,000 credits, and enterprise on request, where ten credits buy one fully matched result and custom columns, contact details and alerts draw extra credits. The honest read: for finding entities on the web against precise criteria, with the evidence shown, Websets sets the bar, and its parent company has the index, the funding and the developer audience to keep raising it.

What AstroFabric actually is

AstroFabric treats the list as the beginning. The prospecting agent discovers accounts by firmographics, technology, hiring, funding, location or similarity to your customers, across a waterfall of sources rather than one index; the enrichment agent fills each field source by source with provenance recorded on every value; the verification agent runs email checks, dedupe and suppression before anything reaches a sequence; the signals agent keeps standing watches on hiring, funding, technology change and intent; and the audience agent turns the rows into matched audiences on LinkedIn, Meta and Reddit. Every row an agent adds to a list has to carry a source reference the platform observed, so a list cannot contain a record the model invented. Delivery goes into the CRM, sheets, outreach tools and ad accounts you already run, through the console or through REST, MCP and the CLI, and every external write passes an approval gate with an audit entry. Pricing is credits with no seats: Starter at $65 a month with 1,000 credits, Growth at $149 with 3,000, Scale by agreement, and a three-day trial with 200 credits. The trade against Websets is explicit: a narrower search surface than a web-scale index built for AI, in exchange for the whole chain after the search, the governance around it, and agents that run the motion every week.

Side by side

Choose Exa Websets if

Choose Websets if the job is finding entities on the open web against precise, checkable criteria, and you want the evidence on every match: research teams, recruiters, investors, and developers building their own pipeline on top of an API. Its index reaches documents and people a company database never held, the per-criterion reasoning is genuinely useful when a list has to survive scrutiny, and the SDKs and MCP server make it a natural component inside an agent you are writing yourself. If your team already owns the rest of the chain, Websets is a superb first stage.

Choose AstroFabric if

Choose AstroFabric if you want the finished outcome without building the pipeline: accounts discovered, fields filled with sources, contacts verified, signals watched, audiences matched and everything delivered into the systems your team already uses, with an approval on every write and a ceiling on every run. Teams building agents of their own sometimes call both through MCP: a web-scale search for discovery and AstroFabric's agents for everything after. For a GTM team, one governed chain is usually the simpler answer.

Frequently asked questions

Sources

• Exa - Websets FAQ (plans and credits)
• Exa - Websets overview
• Exa - how Websets work
• Exa - Websets dashboard integrations
• Exa - API pricing
• Exa - Series C announcement