Unibase is a high-performance decentralized AI memory layer that empowers AI agents with long-term memory and cross-platform interoperability, enabling them to remember, collaborate, and self-evolve.

The project addresses three fundamental challenges for autonomous AI systems: the absence of persistent memory, poor interoperability between agent frameworks, and limited user data sovereignty. By introducing a decentralized memory layer and a Web3-native interoperability protocol, Unibase lays the foundation for the open agent internet.

Unibase consists of three core modules. Membase provides secure and scalable long-term AI memory storage and management. The AIP Protocol defines Web3-native standards for agent-to-agent communication, identity, and shared state. Unibase DA delivers zk-verified, high-throughput data availability (>100GB/s), ensuring reliable and low-latency access for AI workloads.

The native token, UB, supports protocol fees (memory storage, agent deployment, interoperability usage), governance through veUB, agent staking, and knowledge mining incentives.

Ecosystem projects built on Unibase include BitAgent (multi-agent collaboration and staking platform), TradingFlow (autonomous trading agents), TwinX (self-evolving agent platform), and Beeper (intent-based Twitter AI agent). Unibase is also integrated with frameworks such as MCP, Virtuals, ElizaOS, and Swarms.

As of mid-2025, the testnet recorded over 200 deployed agents and more than 12.4 million on-chain memory entries. The UB token will be deployed on Ethereum and BNBChain on September 12, 2025, followed by zk-backed memory verification and cross-chain expansion to other Ethereum L2 networks.

Unibase combines decentralized memory, interoperability, and verifiable data availability into a unified infrastructure powering the next generation of autonomous AI agents.