Skip to content
Keegi

Pricing

Find out what your data can answer, before you buy anything.

A subscription is the thing that stays. Before that, a look at one export you already own, or a two-week audit over one source you connect. The numbers stay in the room. The model is here.

Export Read

You send one export you already own. A GitHub read-only token for repos you name, a Slack channel export, or a folder of documents. No OAuth grant, no standing access, nothing left connected in your tenant afterwards. GitHub is the usual first export: authorship is already resolved, and a VP of Engineering can issue the token alone.

You get the same five things the audit returns, over that export rather than a connected source.

The Expertise Audit

Two weeks, fixed price. You choose the scope: one department, one workspace, or a single folder tree. You connect one source, read-only. Nothing leaves your control and nothing is written back. The fee is credited in full against the first year if you deploy.

You get five things:

  1. How many of your people we can say something true about, and how many we cannot.
  2. Which of your sources actually carry the evidence, and where it is thin.
  3. Ten real questions we would have answered, each with the evidence behind the answer.
  4. Five we would have refused, and why.
  5. A straight recommendation: deploy, wait, or don’t.

If your corpus is too thin, we say so. That is a common outcome and it is the point of looking first.

Then, if you deploy it

The number comes out of a conversation, because it depends on your headcount and your sources. The model it comes from does not, so here it is.

Banded by headcount, metered on indexed people.
Everyone is indexed and everyone can ask. The price steps with the size of the organisation.
Per seat is wrong, and we will not sell it.
Per seat invites you to buy 200 seats for a 2,000-person company. That thins the corpus, and the corpus is the product. Keegi needs everyone indexed even though only some people ask.
Every connector is included.
A connector paywall is how you end up with a corpus one source deep and answers to match. The dashboard of questions your organisation could not answer is the upsell, not the connectors.
Fit depends on how the company is put together.
Where the teams sit matters more than how many people you have. Offices, countries, regions, time zones, and whether the work leaves a written trail. Three hundred people across five countries in Jira and Slack is a different picture from eight hundred in one building who decide everything in meetings. That is what a call is for.
Talk to us

Twenty minutes. No demo, no pitch. I want to know whether this problem is real where you work.

Five questions firstBack to the demo