Raw data → task → grader → verified task.
Most data companies stop at step one. We start where they stop.
1
1
22
Every task we accept ships with proof:
→ reference solution scores 1
→ exploit attempts fail
→ two blind expert solves
→ sealed review, hashed before reveal
Grade the graders.
1
4
Why Applied Will exists:
The built world runs on numbers you can check. Takeoffs, load calcs, code requirements.
We turn that work into verified RL tasks, written and checked by people who've done it.
Tasks your agents haven't seen. Graders they can't game.
8
Hard tasks are scarce and need a domain expert who knows what good looks like.
Applied Will builds expert-verified RL tasks and graders, starting in construction and Polish.
Blind double solves. Sealed review. Experts paid in USDC on Solana.
1
4
138
Applied Will retweeted
I am looking to build out the missing domain expertise across construction for @Applied_will if you are investing in Data AI infrastructure companies, and are looking for an early stage company with domain expertise and ex-Scale AI DMs open
some takes after chatting with ppl at data companies:
- still lucrative. single deals can be millions, comp is generous, and acquisition by a frontier lab is a real exit
- two main categories: cs-centric environments (coding, ml, computer use; often synthetic tasks) vs. domain-specific data (finance, legal, science; usually built with domain experts)
- benchmarks are a sales tool: publish one to show expertise, build credibility, and turn attention into customers
- tech isn't the main moat. what matters more is connections, sales, and trust: winning deals is much harder than producing data itself
- capital is a flywheel (from @M0EGPT ): more capital -> more domain experts -> more and better data -> bigger deals -> more revenue -> reinvest in experts, infra, and research
- probably not a long-term business though. labs may buy less over time, so some data companies are also starting to train their own models
- the downside: much of the work seems engineering-heavy and repetitive, with limited room to grow technically
Applied Will retweeted
Bullish on frontier data
I asked people which data companies influence research direction at the frontier:
16 votes: Surge (@echen)
13 votes: Proximal (@calvinchen, @MatternJustus)
8 votes: Preference Model (@jennifgcrl, @Ning_Catsnail)
7 votes: Datacurve (@serenaa_ge, @charleyslee)
6 votes each:
- Mechanize (pre-acquisition) (@tamaybes, @EgeErdil2, @MatthewJBar)
- Mercor (@BrendanFoody, @adarsh_exe, @suryamidha)
5 votes: Fleet (@nicoup, @andrewthezhou)
4 votes each:
- AfterQuery (@CarlosGeorgescu, @spencermateega)
- Prime Intellect (@vincentweisser, @johannes_hage)
3 votes: Latch (@AlfredoAndere, @kenbwork, @kylegiff)
2 votes each:
- Abundant (@huyouare, @kehuangsf, @meji_abidoye)
- Artificial Analysis (@_micah_h, @grmcameron)
- Bespoke Labs (@madiator, @AlexGDimakis)
- micro1 (@aliansarinik)
- Terminal-Bench (@Mike_A_Merrill, @alexgshaw)
50 people voted. People couldn't pick a company they work at or founded. Disclosure: on this list, I'm a small investor in Abundant, AfterQuery, Arcanum, DLR, Fleet, Latch, Paper Instruments, Snorkel, and Vmax. I didn't vote.
Continued:
Applied Will retweeted
This is also what we are seeing @Applied_will based on discussions with frontier lab partners
The market for AI training data could grow 100x over the next five years.
Frontier labs need increasingly realistic data from actual businesses to train more capable models.
@BrendanFoody explains why this creates a new revenue opportunity for founders and companies.
Find out more at the link in the thread.
Great to be at the frontier of SI! 🇺🇸 🇵🇱
Our AI caught its own mistake.
Real construction drawing.
Our answer sheet said 6 footings.
Our machine pass flagged it.
Real answer: 8.
Caught. Fixed. Paid on Solana.
This is what @applied_will does:
Every training task ships with proof it was checked.
Demoed at @superteamPOL Build Station 🚀
1
6
96
Applied Will retweeted
Our AI caught its own mistake.
Real construction drawing.
Our answer sheet said 6 footings.
Our machine pass flagged it.
Real answer: 8.
Caught. Fixed. Paid on Solana.
This is what @applied_will does:
Every training task ships with proof it was checked.
Demoed at @superteamPOL Build Station 🚀
