How Greenwood Labs works
A marketplace where AI labs pay for the data they need, and the people who supply it are settled in tokenized stock the moment their work is accepted.
What Greenwood Labs is
Frontier AI models are limited by the data they can learn from. Labs need specific, high-quality, real-world data — and the people who can supply it have had no straightforward way to be paid for it.
Greenwood Labs closes that loop in four moves:
- 01
A lab posts a bounty
It describes the data it needs and funds a reward pool behind it.
- 02
You contribute
Upload data that fits the brief. It is checked before anything costs you.
- 03
The work is graded
An AI agent or the lab itself scores how useful your submission is.
- 04
You are paid
The contract settles your reward on-chain, automatically.
Nothing sits in an approval queue. There is no invoice, no minimum payout and no waiting for a monthly cycle — settlement happens in the same transaction that records your grade.
Earning as a contributor
Start to first payment, end to end:
- Connect a wallet.This is where your reward lands and how your submission is signed. You don’t need to fund it to begin.
- Pick a bounty. Browse the marketplaceand read the bounty’s verification requirement — that line, not the description, is what your upload is actually measured against.
- Upload your data.It is checked against the requirement immediately. If it doesn’t fit, you are told why and nothing reaches the chain, so a rejected attempt costs you nothing.
- Confirm the submission. Accepted data is stored, and you sign one transaction to record it on-chain.
- Get graded and paid. Your reward arrives automatically. Track everything on your dashboard.
How rewards work
You are paid in the tokenized equity the lab funded its pool with — NVDA or TSLA, held in your own wallet. Greenwood Labs settles in nothing else: no stablecoins, no platform credit, no points.
Your payout is:
The base reward is what any accepted submission earns. Your score is how useful the reviewer judged the data to be, from 0 to 1. The bonus multiplier is set by the lab and decides how much quality is worth on top. A score of zero pays nothing.
Posting work as a lab
Describing a bounty and funding it happen in a single step, so a bounty is never listed without money behind it. If any part fails, the listing is removed rather than left stranded.
- Title and description
- What contributors see when browsing.
- Verification requirement
- The line uploads are actually measured against. Be specific — this is the filter.
- Data requirements
- Format, resolution, quality — shown as a checklist.
- Funding
- How much goes into the reward pool, and in which token.
- Base reward and multiplier
- What each accepted submission earns, and what quality adds.
Once live, incoming work appears on your lab dashboard, where you can watch submissions land, review them yourself, or leave the AI agent to settle them.
How work is reviewed
Two things can settle a submission, and which one is in force is recorded on-chain:
The AI agent
Grades continuously against the bounty’s requirement, looking at the file itself. Most submissions settle within moments of being recorded.
The lab
A lab can pause the agent and review every submission by hand, inspecting the data before approving a score or rejecting it.
Either way the outcome is public: the score and the amount paid are recorded on-chain against your submission, so you can always see what was decided and what it earned.
Roadmap
Where the platform is now, and where it’s heading.
The marketplace
- Labs post bounties and fund a reward pool
- Uploads checked against the bounty before they reach the chain
- On-chain submissions with automatic settlement
- AI agent grading, or manual review by the lab
- Runs on Robinhood Chain, with pluggable data storage
Real-world asset rewards
- Tokenized equity payouts on Robinhood Chain
- Contributor earnings denominated in the equity you were paid in
- Per-bounty reward pools instead of one shared pool
- Bounty terms recorded on-chain so they cannot be edited after the fact
Reputation and scale
- Contributor reputation that raises the base reward for consistent quality
- Multi-reviewer consensus before a submission settles
- A dispute window so contributors can challenge a score
- Dataset bundles — buy a whole verified corpus, not one file at a time
Open network
- Anyone can run an evaluator and stake against their grades
- Public API and SDK for programmatic submission
- Trusted-execution grading, so a lab need not be trusted at all
Planned items are direction, not commitments, and the order may change as the network grows.
Common questions
Do I need crypto to start?
You need a wallet to receive payment and to sign your submission. You do not need to fund it first — the lab pays the reward, and uploads that fail the quality check never reach the chain, so a rejected attempt costs you nothing.
When do I get paid?
As soon as your submission is graded. There is no invoice, no approval queue and no payout threshold — the contract transfers the reward to your address in the same transaction that records the grade.
What stops someone submitting the same file twice?
Duplicate content scores zero. The evaluator checks it before anything else, so re-submitting work that has already been rewarded earns nothing.
Who decides whether my work is good?
Either an AI agent grading against the bounty's stated requirement, or the lab reviewing by hand. Which one is in force is recorded on-chain, and the lab can switch it.
Can a lab take the reward pool back?
The lab that funded a pool can withdraw what is left of it. Rewards already paid to you are yours and cannot be clawed back.
Developer docs
Build on Greenwood Labs
Contract reference, deployment guides, the evaluator agent, storage backends and a submission API are being written up. Until then, the repository README covers running the whole stack yourself.
- Planned
- Contract reference and ABI
- Planned
- Running your own evaluator
- Planned
- Submission API and SDK
- Planned
- Self-hosting guide