Guide
Is my firm big enough to license its records to AI?
Short answer: Often yes if you have about 20 to 30 or more staff and three or more years of records, since that's where most published minimums sit. Some buyers publish no minimum, and CAD/BIM archives can qualify at smaller sizes on quality alone. The bar has been rising: one commentator reports buyers moving from 10+ to mostly 50+ employees, and Mercor's CEO has posted a 100+ target.
Published size thresholds side by side, why desk staff and depth matter, and what smaller firms can do.
Last updated · Sources linked inline · Buyer figures as published, observed Oct 16, 2026 (micro1, Mercor and GDC re-checked Nov 3, 2026)
What size thresholds do AI data buyers publish?
| Buyer | Published size criterion | Source |
|---|---|---|
| Polyshares | No minimum (any sector); co-founder says CAD has no minimum head count As published · verify | register / X |
| SuperAnnotate | Any size, industry or location As published · verify | register |
| Mode | 6 (legal), 10 (accounting and finance), 30 (other) As published · verify | register |
| Handshake AI | 20+ full-time and 3+ years As published · verify | register |
| micro1 | 30+ full-time employees, primarily English As published · verify | checked Nov 3, 2026 |
| micro1 (defunct companies) | 50+ employees at peak As published · verify | checked Nov 3, 2026 |
| Scale AI | 40+ preferred As published · verify | register |
| Mercor | CEO's post targets 100+ employees As published · verify | X, Oct 23, 2026 |
Register entries observed Oct 16, 2026 via the Who Sells Data register; micro1 checked directly on its data partnerships page; Mercor from its CEO's post. As published, verify. Not offers.
Is the bar going up?
Commentators say so, though niche depth still sells.
One investor-commentator writes that "3 months ago they bought company internals with 10+ headcount, now they almost exclusively purchase from companies with 50 or more employees", while adding that "small companies in a esoteric domain … will still hold immense value" (post; opinion).
Does headcount or desk staff matter more?
Desk staff, according to some in the market, because they generate the records buyers want.
"A 300-person trucking company might have 25 people in email all day. A 60-person CPA firm has 60. When AI labs license company records, desk staff is what counts. Not headcount." (post; opinion). If most of your team works in the field or on a shop floor, enter your desk-staff count in the estimator; it uses the smaller number.
What can a smaller firm do?
Lean on depth and specialization, and prepare well.
- Specialized archives: engineering, architecture and manufacturing CAD/BIM projects are judged on quality and completeness (engineering, construction & manufacturing).
- Long history on consistent tools: years on the same systems show real cycles of work.
- A tight inventory and samples: see how to prepare your records.
- Buyers with no stated minimum: Polyshares and SuperAnnotate, as published.
Sources: eriknewsha (X) · DataDeskClub (X) · Polyshares co-founder on CAD (X) · Mercor CEO (X) · micro1 · Who Sells Data register
Part of our guide series. Start with the pillar: how to license your firm's records to AI.
What do smaller firms ask?
What is the minimum company size to license data to AI?
Does headcount or desk staff count?
We're a 15-person firm. Should we bother?
Does a long history make up for small size?
Check your firm against published minimums
Three answers give you a rough range based on what buyers publish. The rest helps a buyer take you seriously: every question after step 1 is optional. No files, logins or client names.