Pricing, no sales call needed

Try before you commit

Put the agent to the test

Don’t take a landing page’s word for it. Design partners run the agent on a real questionnaire, on their real evidence — and judge it the way they’d judge a new hire.

1. 1

   Send one real questionnaire

   The ugliest one you have — a 400-row SIG, a portal request, a custom sheet. Length doesn’t change the process, and we’d rather see the hard case than the demo case.

2. 2

   Watch the agent run on your evidence

   A read-only connection to your Drata or Vanta, then a live session: drafts, citations, confidence scores — including the questions it refuses to guess.

3. 3

   Judge the drafts yourself

   Score first-pass accuracy against what your team would have written, and how long it took. You keep the completed file either way.

Send us a questionnaire

FAQ

Questions, answered

Does the AI submit answers on its own?

Never. The agent drafts and cites; a human on your team reviews, approves, and submits every answer. Autopilot for the work — sign-off stays human.

What counts as a questionnaire?

One incoming document or portal request, regardless of length — a 400-row SIG counts the same as a 20-question custom sheet. We’d rather you send the ugly ones.

Do you support Drata and Vanta?

Yes. We’re built Drata-first — the deepest integration is with the Drata evidence graph — and Vanta is supported as well. Both connections are read-only.

Can the agent learn from questionnaires we’ve already answered?

Yes. Upload your historical questionnaires and policies to the Evidence Library and they become citable evidence, just like a control or policy from Drata or Vanta. Drift detection covers them too — when your live program no longer matches an old answer, it gets flagged, so last year’s “yes” doesn’t ship stale.

Do you store our portal credentials?

No. The browser agent works inside your own authenticated session. We never see or store passwords, and nothing is submitted without your explicit approval.

What happens when the agent isn’t sure?

It refuses to answer. Questions below the confidence threshold are flagged and routed to the right person on your team — a guessed answer on a security questionnaire is a liability, and we treat it that way.

Is our data used to train models?

No. Your evidence and answers are used only to answer your questionnaires.

Is Provebase itself SOC 2 compliant?

Not yet — a third-party audit of our own stack is on the pre-launch roadmap. In the meantime we work like we expect to be reviewed: design partners get our full security documentation and an architecture walkthrough, and our security principles are public on our Security page. You’re welcome to send us your own questionnaire.

How long does setup take?

Minutes, not weeks. Connect Drata or Vanta with read-only OAuth, optionally upload past questionnaires to your Evidence Library, and send the agent your first questionnaire the same day.

Which AI models do you use — and do they see our evidence?

The agent runs on frontier models from major providers, accessed under API terms that prohibit training on your content. Your evidence is sent only to draft your answers — and every draft still passes through your review before it goes anywhere.

What about a trust center?

Coming soon. The same evidence base that answers your questionnaires will power a public Provebase Trust Center, so buyers self-serve answers before sending a spreadsheet. It will be included in Growth and Scale at launch.

Why is your pricing public?

Because almost nobody else’s is. You shouldn’t need three sales calls to learn what a questionnaire tool costs.