Picture this

Who it is for

Kemi, a fashion stylist, ranks image generations in her spare time. Her accuracy versus peers earns USDC and a taste score that agencies can verify.

Two ways to own it

What you get

Base$229

$57 deposit to reserve

Ranking tasks UI, USDC payouts, peer agreement scoring, taste score badge

You choose: chain
You choose: name and logo
You choose: accent color
You choose: domain niche

Customfrom $1,500

Built around you

Reshape it entirely: new features, your flows, your brand end to end, integrations you need. We scope it with you.

Everything in Base
Custom features and design
Priority build

Why it is new

What exists today, and what this does differently

What is new

Paid preference ranking exists; a portable onchain taste reputation is uncommon.

Closest things out there

Sapien, Kled and Scale pay for labeling, and trust-and-teach prototyped onchain RLHF ranking, but a portable owned taste reputation was not found as a product.

Checked with web searches in October 2026. A search is not proof that nothing similar exists, and the market moves fast, so we recheck before every build.

What it unlocks

The human in it

Your eye becomes a credential and a paycheck.

The Regatta Board

Races it can enter

hackathon · Colosseum · rolling

Colosseum Eternal

On-demand, ongoing

Official page ↗

Existing projects allowed

Their rule on existing code: Designed for teams already building; existing projects allowed

grant · Base · rolling

Base Builder Grants

rolling, retroactive

Official page ↗

Shipped Base apps; existing OK

Their rule on existing code: Existing shipped projects OK (retroactive)

What you own

The whole app, not a starter kit.

The complete Taste Bounty app: experts rank pairs of AI outputs for labs, are paid in USDC for picks that agree with a trusted panel, and build a private taste score that agencies can verify, with plans and checkout built in.

Base price · $229

The user app, complete

Every buyer gets the same finished app. Basic branding and copy changes come through a quick form.

Three step onboardingHome

Shown once: two looks one brief, scored against real stylists, paid every night.

Monday drop and statsHome

Taste score, monthly earnings, panel agreement and the weekly drop line.

Pairwise rankingRank

A clear call, a subtle stylist only pair, and a check question; drawn looks for each brief.

Panel reveal after the pickRank

Panel and crowd answers only appear after picking, with the reason.

Nightly payout slipRank

Accepted pairs and USDC payout onchain; picks never sold with identity.

Taste score credentialScore

Badge, weekly history and attestation; public only if the ranker chooses.

Score share cardScore

Shareable score card for the ranker's niche.

Agency search by nicheAgency

Search opted in rankers by niche with scores. Gated behind Agency verify.

Verify and contactAgency

Check a score against its attestation and send a booking request. Gated behind Agency verify.

Niche, entry test and walletMe

Pick a niche, see the entry test result, payouts, and plans.

Open the full app demo

Add-on

Owner console

Your admin side: revenue, paywall editor, users, rules and the agent log.

Revenue and paywall editorRevenue

Lab fees, agency subscriptions, USDC payments, payment methods, live fee, price and gates.

RulesRules

Panel frequency, speed hold, entry pass mark, payout network.

Scoring log and roadmapAgent

What failed and the fix, running costs, and priced v2 and v3 builds.

Top-up

Growth and referrals

Referral and growth mechanics, switched on when you are ready to grow.

Growth loopsGrowth

Fashion week pop ups, score cards, niche data for labs and agency demand.

The money path, built in

How it earns for you

Rankers never pay. Labs fund pairs and pay a service fee on top of payouts, and agencies pay monthly to search and verify taste scores.

RankersFreeJoin, take the entry test, get paid per accepted pair. No fees taken from rankers.

Lab pay as you goPayouts plus 20%Labs fund pairs per batch and get cleaned preference data with panel agreement stats

Agency verify$29 a monthAgencies search and verify rankers' taste scores by niche to hire stylists, art directors and photographers

Lab fee on payouts set by the owner (default 20 percent). Nightly batch payouts cost a few cents of gas.

Prices are yours to set. Illustrative, not a promise of income.

Growth loops, built in

How it spreads

Top rankers share score cards that recruit peers in the same niche
Labs come for niche data the big platforms lack
Agencies checking scores make the credential worth earning
A Monday drop builds a weekly ranking habit

Roadmap · build the next phase with us

Where it goes next

v1 included

The full ranker app

Pairwise ranking with briefs and shuffled pairs
Panel workflow and check questions
Agreement scoring with timing checks
USDC escrow and nightly payouts
Weekly taste score attestation, plans

Included

Video and voice niches

Rank short video generations
Rank voice and music samples
Niche entry tests
Lab dashboards per niche

Add-on from $700

Taste credential network

Portable taste credential
Agency hiring flows
Ranker teams and studios
Revenue share for panel members

Add-on from $1,500

For investors and grant reviewers

The pitch, ready

AI labs pay experts to rank their outputs, and that taste becomes a verifiable credential.

Problem: Preference data from generic crowds misses what experts see, and experts' taste is invisible and unpaid.

Solution: Pairwise ranking scored against a trusted panel, nightly USDC payouts, and a weekly attested taste score agencies can verify.

Why now: Labs need niche preference data to tune image and video models, and cheap stablecoin payouts make micro tasks viable worldwide.

Market: Stylists, art directors and photographers, AI labs needing niche data, and agencies hiring creative talent.

The ask: A pilot batch with one image lab in fashion niches and three Lagos agencies.

Funding is never guaranteed. Check each program's rules.

For hackathons and demo days

Your 2 minute demo

Home: meet Kemi, a stylist in Yaba ranking AI looks in her spare time.
Rank: make the clear call, then the subtle gele pair the crowd gets wrong.
Rank: answer the check question and see tonight's payout slip.
Score: see the taste score, its history and the privacy toggle.
Agency: unlock Agency verify through the USDC checkout, search by niche and verify Kemi.
Owner Agent: read how majority scoring punished experts and how the panel fixed it.
Owner Revenue: change the lab fee and agency price, and preview the Plans sheet.
Owner Growth and Roadmap: run a fashion week pop up, then request video and voice niches.

Many hackathons require code written at the event. See the Regatta Boardfor each race's rule.

Make it entirely yours

Custom work from $1,500: your niches and panels, a lab data export format, agency integrations or a white label ranking studio. Scoped on a call.

Custom from $1,500

BLUEPRINT

Why it matters and how it really works

The need

AI image labs need human judgment in niches like West African fashion, where crowd raters often lack context. Experts with that eye rarely get paid for it, and their skill leaves no record they can show a client.

What exists and what is new

Paid preference ranking exists: Sapien, Kled and Scale pay people to label and rank AI outputs, and the trust-and-teach project prototyped onchain RLHF ranking. A portable, owned taste reputation that agencies can verify was not found as a product. Research verdict: rare, medium novelty. The ranking work is familiar; the new part is scoring against an expert panel and making that score the ranker's own credential.

The data it runs on

Pairwise tasks (two images and a brief) from labs, each ranker's picks and timing, gold answers from a paid panel of working professionals on a sample of pairs. Onchain: USDC payouts batched nightly and a periodic taste score attestation per ranker and niche. Images and picks stay in the lab's private dataset.

1.The mission

An expert eye should be paid like a skill and recorded like one.

2.Why people can't resist it

Right when the crowd is wrong A minority bonus pays more when you beat the majority and the panel agrees with you.

A score you can show Your taste score per niche lives onchain. Send the link to a client instead of a portfolio.

The Monday drop Twenty fresh pairs every Monday in your niche, a local leaderboard, and panel invitations for top scores.

3.How it works for real

Wallets: Privy embedded wallets on Base for rankers; labs fund a USDC escrow per batch. Solana as an option for labs that prefer it.

Task service: Serves shuffled pairs per ranker, records picks and time on task, never shows panel answers before a pick.

Panel and gold answers: Five paid professionals per niche answer about 1 in 20 pairs. Known check pairs catch rushing and bots.

Scoring agent: Weighs agreement with the panel, timing and consistency into a 0 to 100 score per niche. A small model tier (for example Claude Haiku via OpenRouter) flags odd patterns like copying rings for human review.

Payouts: A payout contract on Base pays USDC to many rankers in one nightly batch from the lab's escrow.

Taste attestation: Weekly EAS attestation on Base with ranker, niche, score and pair count. Rankers choose whether it is linkable publicly or only by shared link.

4.Cost per use and pricing

Per pair: no model cost for ranking itself; fraud checks on a small tier cost well under $0.001 per pair. Panel answers at about $0.50 each on 1 in 20 pairs add about $0.025 per pair. Base gas for a nightly batch payout is usually a few cents for the whole batch. Example month: 40,000 pairs at $0.10 is $4,000 to rankers, about $1,000 panel cost, $20 compute, $5 gas, $50 hosting; the operator charges the lab a 20% fee on top, $800.

Rankers Free

Join, take the entry test, get paid per accepted pair. No fees taken from rankers.

Lab pay as you go Payouts plus 20%

Labs fund pairs per batch and get the cleaned preference data with panel agreement stats.

Agency verify $29 a month

Agencies search and verify rankers' taste scores by niche to hire stylists, art directors and photographers.

5.Growth loop

Top rankers share their score card, which recruits peers in the same niche. Labs come for niche data the big platforms lack. Agencies checking scores make the credential worth earning, which brings more experts.

6.Risks and guardrails

Income expectations: Work depends on lab demand and varies week to week. Never present earnings as a salary or guaranteed.

Gaming and copying: Rings and bots will try. Shuffled pairs, check questions, timing holds and human review of flagged wallets.

Panel bias: A small panel can encode one taste. Rotate panelists, publish how they are chosen, and allow appeals on disputed pairs.

Content and rights: Labs must own or license the images; filter unsafe content before it reaches rankers.

Worker and tax rules: Paying people across countries in USDC has tax and labor implications. Provide earnings statements and follow local rules for contractors.

7.Build order

Pairwise ranking UI with briefs and shuffled pairs
Panel workflow and check questions per niche
Scoring by panel agreement with timing checks
USDC escrow and nightly batch payouts on Base
Weekly taste score attestation with EAS
Monday drop and shareable score card

8.Where this is going

Now
A stylist in Lagos ranks aso ebi looks and earns a score agencies can check.
Next
Panels for many niches: architecture, food photography, music mixing, each with its own experts.
Later
Labs route their hardest judgments to the highest scored humans automatically, paying for expertise, not clicks.
The world it builds
AI learns taste from the people who actually have it, and those people are credited and paid.

Costs and tools are estimates at the time of writing. General information only, not legal or financial advice.

OWNER PLAYBOOK

Own it, then chart your voyage

Nine routes from one vessel. Tap a port to open its chart.

Why own Taste Bounty

Taste Bounty sits between AI teams that need human judgment and people with a trained eye, and the owner of that marketplace sets the niche, the standards and the take rate. When you own it, every ranker's taste score is issued by your contract and every lab that buys rankings buys from your brand. You decide which experts belong, which is exactly the thing generic labeling platforms cannot do well.

What’s stopping you?

“I can't code. How do I run a data marketplace?”

The ranking task UI, USDC payouts, peer agreement scoring and the taste score badge ship ready. Your work is choosing a niche, recruiting rankers and selling tasks to buyers. Custom task types are paid options.

“Why would an AI lab buy from me instead of Scale or Sapien?”

Big platforms are general. You win with depth in one niche, like fashion imagery, African languages or medical illustration, where you can recruit real experts and prove their agreement scores. Start with small teams and indie model builders, not giant labs.

“What if rankers just click randomly for money?”

Peer agreement scoring catches most of that: rankers who disagree with trusted peers lose score and pay. Add a few known answer checks in each batch and remove repeat offenders.

“Is it legal to pay people worldwide in USDC?”

Rules differ by country and some places treat this as contract work or restrict stablecoins. Check your own local rules, keep records, and get local advice before scaling payouts.

“Is it really mine?”

Your wallet deploys it, your brand runs it, your take rate funds it, and the ranker community you build is yours to keep.

The million-dollar ladder

Buyers fund ranking batches; you keep a margin between what buyers pay per judgment and what rankers earn, plus a premium for verified expert pools. The ladder below is a target, not a promise.

$300

$4,000

$25,000

$85,000

First 10 rankers, one buyer

1 buyer

$300 per batch margin

Repeat buyers in your niche

8 buyers

$500 per month margin

Expert pools with verified scores

25 buyers