Axiom

Agents don’t authenticate — they pay.

AXIOM is an economic runtime for autonomous agents that replaces API keys with payment as access control.

Instead of relying on accounts, credentials, or subscriptions, AXIOM enables agents to discover, pay for, and execute APIs on demand using micropayments via HTTP 402. Agents can dynamically select tools (AI, weather, finance), choose pricing tiers based on task complexity, and execute requests autonomously using a wallet.

In our system, a single agent performs multi-step tasks by selecting the appropriate APIs, deciding how much to pay for each request, executing payments automatically, and aggregating results into a final response. This transforms APIs into economic primitives that agents can interact with directly.

AXIOM integrates decentralized infrastructure to provide persistent memory and execution context, allowing agents to store past interactions, improve decision-making, and operate continuously over time.

By replacing authentication with payment, AXIOM enables a new model for the internet — one where agents can act, transact, and coordinate autonomously without onboarding friction.

AXIOM is built as an agent runtime that combines a payment-enforced API gateway with autonomous decision-making and decentralized AI infrastructure.

We implemented a reverse-proxy gateway that wraps any API and enforces payment via HTTP 402, replacing API keys with micropayments. On top of this, we built an LLM-based agent loop with tool-calling, enabling agents to select APIs, choose pricing tiers, and execute multi-step workflows autonomously.

To align with the 0G track, we integrated 0G Storage for persistent agent memory and 0G Compute for decentralized inference, allowing agents to store context, reuse past results, and execute tasks on verifiable infrastructure. This enables continuous, autonomous operation.

A key design choice was embedding payment directly into the execution flow, enabling seamless “priced → executed → result” interactions without friction.