AI on Base is becoming one of the sectors I’m watching closely right now. Base is clearly positioning itself as a settlement layer for the agent economy, where AI agents can trade, pay for APIs, hire services and execute onchain. Some numbers already show the direction: – 75M+ agent tx – $24M+ payment volume – 12K+ active agents – 22K+ x402 APIs – 10M+ daily tx on Base x402 is one of the key pieces here. An agent can request a service, pay in USDC and receive the result without relying on a normal subscription flow. That opens a much cleaner path for machine-to-machine payments across inference, data, search and other paid services. The stack is also getting more complete: – x402 → payments – ERC-8004 → identity + reputation – Base MCP → wallet + onchain execution – USDC → settlement – TEE/ZK → private & verifiable compute Projects I’m currently watching: – @aeonframework – autonomous agent framework + self-healing skills – @OpenGradient – private/verifiable inference + x402 – @sibylcap – persistent memory for agents – @QuotientHQ – AI forecasting for prediction markets/perps – @KorProtocol – AI music/IP marketplace with x402 settlement – @ILITY_xyz – agent-to-agent jobs + escrow – @officialbunnyos – AI gaming + distributed inference – @capacitr_xyz – signals → perps/prediction execution – @myrad_hq – verified behavioral data for AI/robotics – @basemateagent – Base-native trading/execution agent Inference, memory, identity, payments and execution are starting to connect inside the same ecosystem. That matters because an autonomous agent eventually needs all of them to operate without constant human input. I’ll keep tracking actual usage here: paid requests, inference volume, jobs settled, active agents and trading activity.