Human-in-the-loop for AI

Label Data. Earn Rewards. Power Better AI.

SupaFi is a decentralized platform where humans validate datasets and earn on-chain rewards through consensus-driven labeling. No annotation team. Just a global community of expert labelers.

No KYC required · Web + PWA

Consensus Engine

Every label is validated by multiple humans with trust scores.

Instant Rewards

Earn tokens for both labeling and validation accuracy.

For Any Modality

Images, text, audio, and more with task-specific UIs.

Live Task Demo

Crop Disease Labeling

Which crop disease is this?

A · Healthy62% consensus
B · Rust28% votes
C · Blight10% votes

✔ Consensus reached

3 of 3 labelers agree (98.3% trust score)

Reward

+5 tokens

Jump into the app

Pick the flow that matches what you want to do today—set up your profile, publish a task, label data, or validate submissions.

Step 1

Complete profile

Connect your wallet details with a name and email so we can track your rewards.

Go to complete profile
For AI teams

Create a task

Spin up a new labeling campaign with custom labels, rewards, and dataset type.

Go to task creation
For labelers

Submit work

Choose a task, upload your data, pick a label, and earn rewards for quality work.

Go to submissions
For validators

Validate labels

Swipe through the feed, vote yes/no on submissions, and boost your reputation.

Go to validator feed

The Problem

AI Needs Clean Data. Most Companies Don't Have It.

Models are only as good as their datasets. Today, labeling is slow, expensive, and often wrong. Static annotation teams can't keep up with real-world data.

Noisy Datasets

Mislabels and inconsistent guidelines silently poison your models, increasing hallucinations and decreasing trust.

Slow Manual Labeling

Traditional workflows struggle to keep pace with ever-changing domains like crops, medical images, or safety content.

Expensive Annotation Teams

Centralized vendors lock you in with opaque quality metrics and rigid pricing structures.

The Solution

A Decentralized, Gamified Labeling Network.

SupaFi routes datasets to a global pool of vetted labelers. Consensus, trust scores, and AI checks work together to guarantee data quality—while contributors are rewarded in real time.

Flow Overview
  1. 1

    Company uploads dataset

    Images, text, or events come in with task instructions.

  2. 2

    Platform distributes tasks

    Labelers see missions tailored to their expertise & trust score.

  3. 3

    Consensus validation

    Multiple independent labels, conflict resolution, and AI checks.

  4. 4

    Earn + analytics

    Contributors earn tokens; companies see real-time dataset health.

How It Works

Gamified missions that feel like a game, not a form.

Step 1

Pick a Task

Choose from image, text, or review missions that match your interests.

Step 2

Label the Dataset

Follow clear instructions with task-specific UIs and keyboard shortcuts.

Step 3

Validate With Community

Agree or disagree with others, flag edge cases, and resolve conflicts.

Step 4

Earn Rewards

Get paid in tokens for labels, validations, and streak bonuses.

Live Demo

Try an example task in 10 seconds.

This is a simplified preview of how a labeling mission feels on desktop, mobile, or as a PWA.

Image Classification

Which crop disease is this?

Optimized for lightweight PWA · Replace with real task image
Pick an answer to see how quick a mission feels.

Rewards

Earn tokens for skill, not clicks.

SupaFi rewards expertise, streaks, and consensus accuracy—so the best labelers naturally rise to the top.

Per-task rewards

Dataset label reward: 2 tokens Validation bonus: 1 token Accuracy bonus: +20% Leaderboard boosts each week

Progression & streaks

Daily missions, streak multipliers, and seasonal events keep labeling sessions short, focused, and rewarding.

Transparent scoring

Every contributor has a visible trust score, accuracy stats, and category strengths to help companies route the hardest tasks.

Community

Top labelers this week.

A global leaderboard showcases consistent, high-quality contributors across different domains.

Top LabelersWeekly reset · Anti-cheat enabled
1Alex
1200 labels
2Priya
980 labels
3Sam
870 labels

For Companies

Ship production-ready datasets without building an annotation team.

Plug your datasets into a curated, global pool of labelers with transparent quality metrics and real-time analytics.

  • Faster labeling via parallel, global missions.
  • Consensus-based validation and anomaly detection built in.
  • Cheaper and more transparent than traditional annotation vendors.
AI Dataset Insights
  • • Detect label conflicts across annotators and time.
  • • Highlight suspicious labelers and unusual behavior.
  • • Summarize dataset quality, coverage, and drift.
  • • Generate shareable reports for internal stakeholders.

Ready to Launch?

Start earning by labeling data—or upload your first dataset—today.

Join the community, complete your first mission in under 60 seconds, and see real-time impact on AI models.