What data do AI labs want from companies? The chain

The short answer

Labs want the chain from request to finished work across tools. In an undated forum talk, micro1's David Revland described a Slack request for a finance report and the report in Sheets. Troveo says AI companies pay more for work they can follow. Neither publishes a figure.

1A Slackmessage asks2Work happensin Sheets3The finishedreportmicro1's David Revland said in an undatedrecorded forum talk that labs want the wholechain, from the ask to the finished report.1A Slack message asks2Work happens in Sheets3The finished reportmicro1's David Revland saidin an undated recordedforum talk that labs wantthe whole chain, from theask to the finished report.

The sources agree on the shape of what buyers want. They don't agree on a price. None is public.

What is the request-to-deliverable chain?

micro1's CFO used a marketing budget request as the example in an undated recorded forum talk. The value the CFO described is the path from the request, through the analysis, to the approval, spread across tools.

A Slack message asks. A spreadsheet holds the work. An email carries the approval. Each step sits in a different system.

The CFO said in the same talk that micro1 does not want personal information, since the workflow is the product. In a micro1 video, an unnamed micro1 speaker named three most valuable artifacts (self-reported, about 0:41):

  • Company workflow and operational data.
  • Google Workspace data.
  • Internal communications.

Which tools does the chain run through?

micro1's data manifest form, as described in the undated recorded forum talk, asks about Google Workspace, Slack, project management tools and GitHub.

Mercor's data page lists Slack messages, Gmail, Salesforce accounts and opportunities, GitHub repositories, and Notion, Confluence and Drive content.

Troveo groups systems into eight named categories, from chat and email to finance and code. It adds a catch-all for internal and custom systems.

Our data manifest form page shows where that step sits in micro1's process.

Do outcomes matter?

Troveo's guide says data showing what happened next adds signal that activity logs lack. Examples are won or lost, and resolved or escalated. Fortune, in its report on micro1's Spirit offer, wrote that an agent needs to learn how to choose actions, use tools, respond to intermediate results and recover from mistakes. It said AI companies call the record of those steps a trajectory.

What does a lab do with the chain?

The sourced uses are on our page about what labs do with company data. The reasons real chains beat built ones are on why labs want real company data.

What labs say they look for in an offer, from their own job posts, partnership page and intake form, is on how a lab's data team judges an offer.

Where does a company start?

List the tools your team uses and who administers each. Our rights check asks who owns the records and who can sign. The data readiness checklist adds the questions a lawyer will ask.

micro1 may pay us a referral fee if a company we introduce goes ahead with it. How we are paid

Apply to micro1's data partnership (paid link)micro1's own data-partnerships pagePaid link. How we are paid

Vaultfish helps you run the process. We are not lawyers and this is not legal advice. If you need counsel who knows this area, we can point you to one. Ask for a referral