Miju Labs

All companies

Prolific

The only company in the vertical that publishes its take rate — 42.8% on top of participant pay — and the best public anchor for what this spread actually is.

medium confidence4 minupdated 2026-08-29ai labs · data · take rates · research
Vertical
Expert data for frontier labs
Founded
2014
Headquarters
London
Raised
~$33.4M (£25M Series A, July 2023, Partech and Oxford Science Enterprises; earlier YC seed)
Last valuation
Not disclosed; no recent mark
Revenue
~$350M annualised as of April 2026 — a Sacra ESTIMATE, not a disclosure. Prolific books its platform fee, so this is closer to NET than most names here.
Status
Active; pivoted from academic research panels into AI evals, RLHF and red-teaming
Who runs it · 4 people in the index

Latest (Sep 2026): Sacra estimates Prolific at ~$350M annualised revenue as of April 2026. This is an estimate, not a company disclosure, and it is fee-based (close to net). The company publishes its platform fee: 42.8% on top of participant pay for corporate customers. In Dec 2025 it launched HUMAINE, a blinded human-preference leaderboard of frontier models (~26-27k evaluators). On 19 May 2026 CEO Phelim Bradley published an open letter arguing that AI-agent contamination of online research is overstated outside MTurk. The last disclosed funding is still the £25M Series A (Jul 2023).

What Prolific is saying
Phelim Bradley
Co-founder and CEO at prolific.com
Excited to announce that Prolific has been named in Tech Nation's Future Fifty cohort for 2026. Future Fifty is Tech Nation's recognition of the UK's strongest scaling companies, having supported over 35% of the country's unicorns to date, so I'm pleased to see Prolific part of this year's group alongside companies like Encord pushing AI infrastructure forward. I'm at the Forum today alongside tech founders, investors, and policymakers. Looking forward to good conversations with all of them. We're also featured in the Tech Nation 'UK Tech Companies to Watch' report - link in the comments to read.
1133 comments5 reposts
Ekaterina Damer, PhD
Behavioral Scientist, Founder & Angel Investor
"Europe is exposed to the risks of powerful AI while much of the economic upside and power is ending up elsewhere. Even if AI 'goes well,' we don't want our defense, scientific progress and economic prosperity to depend on other countries' goodwill in granting Europe access to models and chips."
1
Who gets to judge the model? Prolific and AI Circle are bringing researchers, founders, operators, and technical leaders together in NYC, Sept 16th, for a candid panel on who should actually decide whether a model is good. RSVP ↓ luma.com/4jliw8f5

Prolific matters out of all proportion to its size for one reason: it is the only company in this entire vertical that publishes what it keeps.

From its own pricing page: "our platform fee is usually 42.8% for corporate customers, and a discounted 33.3% for academic or non-profit customers." Minimum participant pay is £6.00/$8.00 per hour; the recommended rate is £9.00/$12.00 (Prolific pricing).

Why 42.8% is the number to remember

The fee is charged on top of participant rewards, not deducted from them. So a researcher paying £100 to participants pays £142.80, and participants receive 100/142.80 = ~70% of total spend — which is exactly how Sacra frames the same arrangement (Sacra). The two descriptions are arithmetically identical, and that consistency is the point: this is the one place in the market where the gross and net figures can be derived from each other without a leak.

Set it against everything else that is pinned down:

CompanyTake / gross marginBasis
Prolific42.8% fee on top of pay (~30% of gross)Published
Mercor27% (2025), 33% (Q2 2026) gross marginLeaked to The Information
Handshake AI60–70% paid to contractors ⇒ ~30–40% takeReported
micro1~30–40% implied by $150–200M net on $500M grossReported, internally inconsistent
Appen40.3% gross marginAudited
Innodata49% adjusted gross marginAudited

Prolific's disclosure brackets Mercor's leaked margin from above, and Appen and Innodata bracket it from the public side. The convergence of six independent figures on a 30–45% band is the strongest evidence in the atlas about what this vertical's spread actually is — and it is a services spread, not a software one. See What a rake can actually be and GMV is not revenue.

What it sells now

Prolific began in 2014, founded by two Oxford PhD students, as a marketplace for academic research participants. CEO Phelim Bradley has pivoted it into AI evaluations, RLHF and red-teaming, which is where Sacra attributes the growth to ~$350M annualised as of April 2026 — a figure that, because Prolific books its platform fee rather than the participant rewards it sits on top of, is close to net and not comparable to the gross run-rates elsewhere in this vertical.

Not verified

That $350M is a Sacra estimate, not a company disclosure, and is tagged [WEAK] in the research notes. Prolific has never published a revenue figure.

If the estimate is even directionally right, Prolific is the highest revenue-per-dollar-raised business in the sector: ~$350M of revenue on ~$33.4M of total capital, roughly 10x. Compare Scale AI at ~$1.6B raised, or Mercor at ~$484M.

Supply

200,000+ ID-verified active participants across 40+ countries, a 2M+ waitlist, 11,000+ AI annotation specialists and 5,000+ active AI taskers (Sacra).

Note "ID-verified." Prolific's original buyer — an academic running a behavioural experiment — cares about data integrity in a way a lab buying volume historically did not, and identity verification is baked into the product rather than bolted on. That matters more now than it did: academic work confirms that chatbot-assisted cheating contaminates crowdsourced behavioural research at scale (Asher et al., 2026). See Who is actually on the other end.

The read

Prolific sits at the cheap end of the wage ladder — $8/hr minimum, $12/hr recommended, against Handshake AI's $100–125/hr average and Mercor's $81–100+/hr. It is not competing for the same supply as the expert-data names, and it is not winning nine-figure lab contracts.

What it demonstrates is the shape of the business stripped of the GMV headline. A ~30% share of gross, a fee that is visible on the pricing page, a fragmented buyer base of researchers and labs rather than two dominant customers, and no valuation mark to defend. On One customer is a binary event alone it is structurally safer than anything else in this vertical.

Gap in the record

Prolific's valuation, profitability, customer concentration and 2026 revenue are all undisclosed. The 42.8% fee is the only hard number the company itself has put on the record.