AI inference moves quickly, but paying for it often starts with a human signing up, adding a card, buying credits, and managing an API key. That approach becomes harder to use when apps and autonomous agents need to access models on demand and the final cost of a request is not known upfront.
In this episode, Blessing Adesiji from Circle Developer Relations talks with Jamie, founder of Alethieum and builder of its flagship product, Ando.
Jamie introduces MPP-I, the machine-payments protocol he built for AI inference, and explains how it keeps a model response moving while a separate path handles payment. They discuss why he chose Arc for USDC settlement and how Ando puts the protocol to work. Jamie also demonstrates Ando Wallet, from creating and funding a wallet to connecting an app, an autonomous agent, and a robot simulation.
The conversation closes with his vision for more open access to intelligence and what developers can build as agents begin to manage their own economic activity.
Resources:
The Ando walkthrough was recorded using Arc Testnet and testnet USDC. Check Ando’s documentation for current product availability.
What would you build if an agent could access and pay for inference on its own?
Let us know in the comments.
