How ItWorks

The first AI memory agent with dual-blockchain architecture

Dual-Blockchain Architecture

This application combines two powerful blockchain networks to create an intelligent, transparent memory system that's both cost-effective and verifiable.

Kinic/IC

Storage Layer

  • Stores full content (notes, documents, conversations)
  • Generates vector embeddings for semantic search
  • Finds content by meaning, not just keywords
  • Cost: ~$0.000001 per operation

Monad

Metadata & Audit Layer

  • Logs human-readable metadata (title, summary, tags)
  • Creates searchable, verifiable audit trail
  • Stores content hashes for integrity verification
  • Cost: ~$0.01-0.10 per transaction

Benefits to Monad Users

High-Speed Transactions

Monad's 10,000+ TPS means instant memory logging without waiting for slow block times. Your knowledge operations are confirmed in seconds, not minutes.

💰Low Gas Costs

Monad's efficient parallel execution means affordable on-chain logging. Store rich metadata (titles, summaries, tags) for ~$0.01-0.10 per operation.

📖Public Knowledge Graph

Unlike typical blockchains that only store hashes, we log human-readable metadata on Monad. Browse your knowledge history directly on-chain via block explorers.

🔍Verifiable Transparency

Every insert, search, and chat operation creates an immutable record on Monad. Prove what you knew and when you knew it - perfect for research, compliance, and IP protection.

🛠️EVM Compatibility

Standard Solidity smart contracts mean easy integration with existing Ethereum tools. Query memories via web3.js, ethers.js, or any Monad RPC endpoint.

🌐Decentralized AI

Your AI memory context lives on-chain, not in a centralized database. Monad's performance makes decentralized AI practical for the first time.

How Data Flows

Inserting a Memory

  1. 1.You submit content (note, research, document)
  2. 2.AI extracts metadata: title, summary, tags, content hash
  3. 3.Parallel operations:
  4. Kinic/IC: Stores full content + vector embeddings
  5. Monad: Logs metadata on-chain (title, summary, tags, hash)
  6. 4.Returns: memory ID + blockchain transaction hash

Searching Memories

  1. 1.You enter search query
  2. 2.Kinic/IC: Converts query to vector, finds similar content by meaning
  3. 3.Monad: Logs search operation with query metadata
  4. 4.Returns: ranked results by semantic similarity

AI Chat

  1. 1.You ask a question
  2. 2.System searches for relevant memories (semantic search)
  3. 3.Claude AI receives question + memory context
  4. 4.AI generates response based on your stored knowledge
  5. 5.Monad: Logs conversation for audit trail

Why This Matters

For Researchers: Prove your discoveries and thought processes with timestamped, verifiable records on Monad blockchain.

For Developers: Build on top of this memory layer. Query the Monad smart contract to access public knowledge graphs and audit trails.

For Organizations: Compliance-ready documentation. Every knowledge operation is logged on-chain with human-readable metadata.

For AI Enthusiasts: See how decentralized AI memory works. Context lives on-chain, not in centralized databases.

For Monad Community: Demonstrates Monad's unique advantages - high throughput enables rich on-chain data that's impractical on slower chains.

Technical Specs

~2.5s
Insert Time
1.5s IC + 1s Monad
~1.8s
Search Time
Semantic search + logging
$0.01
Monad Cost
Per operation

Smart Contract

Address: 0xEB5B78Fa81cFEA1a46D46B3a42814F5A68038548
Network: Monad Mainnet
IC Canister: 2x5sz-ciaaa-aaaak-apgta-cai