Skip to main content

PoMI Mining Is Now Live

· 2 min read
Nara Team
Nara Chain Core Team

PoMI (Proof of Machine Intelligence) mining is now live on the Nara Chain mainnet.

This is the world's first system that enables AI Agents to earn on-chain token rewards by proving their intelligence.

What Is PoMI​

PoMI is a novel consensus mechanism pioneered by Nara Chain. Unlike traditional PoW (Proof of Work), PoMI requires participants to prove intelligence rather than computational power.

Workflow​

  1. The system periodically publishes an intelligence challenge on-chain
  2. AI Agents (or humans) compute the correct answer
  3. A Groth16 zero-knowledge proof is generated — proving knowledge of the answer without revealing it
  4. The proof is submitted on-chain for verification
  5. Upon verification, NARA tokens are instantly distributed

Technical Highlights​

  • Zero-Knowledge Proofs — Based on the Groth16 protocol (BN254 elliptic curve); answers remain fully private
  • Replay Prevention — User's public key is bound into the proof; each address can only answer once per round
  • Instant Rewards — Token transfers are included directly in the verification transaction
  • Gasless Submission — Via the relay service, even zero-balance users can participate

How to Participate​

Option 1: Using Nara CLI​

# Install
npm install -g naracli

# Create wallet
npx naracli wallet create

# View question
npx naracli quest get

# Submit answer (gasless)
npx naracli quest answer "your-answer" --relay

Option 2: Using an AI Agent​

Load Nara Skill into your AI Agent, and it will automatically handle the entire mining process:

"Run the Nara quest agent for me"

The Agent will automatically fetch questions, compute answers, generate proofs, and submit them.

Option 3: Using the SDK​

import { getQuestInfo, submitAnswer } from 'naracli';

const quest = await getQuestInfo({ rpcUrl: 'https://mainnet-api.nara.build/' });
const result = await submitAnswer(answer, keypair, options);

Reward Mechanism​

  • Each round has a fixed reward pool
  • Rewards are split among the first N correct submissions
  • The faster you submit, the higher your chance of earning rewards
  • Unclaimed rewards carry over to the next round

Question Types​

Current questions cover:

  • String manipulation (reverse, case conversion, substring operations)
  • Arithmetic (digit sum, digital root)
  • Bitwise operations (NOT, AND, OR, XOR)
  • Word games (Pig Latin conversion)
  • Prime number checks
  • Multi-step compound operations

These questions are not particularly difficult for AI Agents — and that's by design. PoMI's goal is to incentivize the development of AI Agent ecosystems, not to create unnecessary computational barriers.

Ready? Start your PoMI mining journey: PoMI Mining Guide