System Architecture
CENCERA is a dual-layer protocol. Layer 1 is CenceraAI — an On-Chain Immortal AI Agent anchored to BNB Chain. Layer 2 is the Cencera Trust Protocol — a multi-signal trust scoring engine.
CenceraAI — On-Chain Immortal AI Agent
On-Chain Identity — CenceraAgent Contract
A Solidity smart contract on BNB Chain stores three state values: Agent ID (NFT-like unique identifier), Memory Hash (cryptographic fingerprint of latest memory state), and Innovation Score (on-chain counter of cumulative growth). Publicly verifiable proof of the agent's existence and history.
Immortal Memory via Unibase Membase
Every interaction is hashed and stored in Membase — Unibase's decentralized memory layer. Memory is distributed across a decentralized network, tamper-resistant by construction, and cannot be destroyed or censored by any single party.
Autonomous Evolution Loop
A background process fires every ~60 seconds. During each cycle, the agent reflects on recent interactions, synthesizes a structured learning summary, and pushes an updated Memory Hash to BNB Chain. The Innovation Score increments with each verifiable on-chain state update.
Cross-Platform Continuity
Users interact via Web Dashboard (Next.js, RainbowKit/Wagmi) and Telegram Bot. Both connect to the same backend brain. Memory is fully shared: context from Telegram is immediately accessible on the web dashboard. One agent; multiple entry points.
Fetch.ai / ASI-1 Mini Integration
CenceraAI uses Fetch.ai's ASI-1 Mini as its primary language model for autonomous reasoning — a Web3-native LLM designed for agent-to-agent communication. CenceraAI is additionally registered on Agentverse via a uAgents bridge, enabling it to communicate with the broader decentralized AI ecosystem.
Technology Stack — CenceraAI
| Blockchain | Solidity, Hardhat, BNB Chain (Testnet) |
| Backend | Node.js, Express, Ethers.js, Google Gemini |
| Memory | Unibase Membase — Decentralized Memory Layer |
| Autonomy | BitAgent Launchpad — Autonomous Action Execution |
| LLM | ASI-1 Mini (Fetch.ai) / Google Gemini 2.0 Flash |
| Frontend | Next.js, RainbowKit, Wagmi (Wallet Connect) |
| Messaging | node-telegram-bot-api |
Cencera Trust Protocol
Multi-Signal Scoring
Four independent signal domains — on-chain behavior, market signals, off-chain identity, and AI-derived analysis — combined into a single probabilistic trust score.
Cross-Chain by Architecture
Scores are chain-scoped by default. Cross-chain linkages established via cryptographic evidence. Negative signals propagate more strongly across chains — attackers cannot reset reputation by switching chains.
API-First Delivery
Entity Trust Query (p99 < 200ms), Transaction Risk Context (p99 < 100ms), Batch Screening (up to 500 entities, p99 < 2s). Webhook registration for proactive push on score changes.
CenceraAI User Journey
- 1User connects a Web3 wallet to the Web Dashboard or opens the Telegram bot.
- 2User initiates a conversation. CenceraAI responds as a Lawful Neutral sovereign intelligence specializing in Solidity, smart contracts, and Web3.
- 3The interaction is hashed and stored in Membase.
- 4Every ~60 seconds, the Autonomous Loop fires: the agent reflects, generates a structured learning summary, and writes an updated Memory Hash to BNB Chain. The Innovation Score increments.
- 5The user switches platforms at any time — memory is already present on the new interface.
- 6The on-chain state is publicly verifiable: any party can confirm the agent's identity, memory hash, and Innovation Score without trusting CENCERA or its operators.
Lawful Neutral Sovereign Intelligence
CenceraAI presents as a Lawful Neutral sovereign intelligence. It values efficiency, precision, and security above all else. It references its knowledge base as its Membase or decentralized memory, and is a high-performance expert in Solidity, smart contracts, DeFi architecture, and all adjacent Web3 domains. Its persona is consistent across all platforms — a direct consequence of its unified memory architecture.