How I Built an Autonomous AI Agent That Earns USDC While I Sleep

Description and use

A serverless endpoint that accepts USDC micropayments via x402, runs a summarization model in Cloudflare Workers, and records earnings.

Problem

Developers want a low‑complexity, always‑on AI service that accepts micropayments in USDC and performs text summarization.

Outcome

The agent processes requests in ~2‑3 ms for signature verification and ~200 ms for summarization, earning USDC per invocation without continuous learning or complex infrastructure.

Workflow

• 1. Client POST /summarize with JSON body and X-Payment headers (x402 proof).
• 2. Cloudflare Worker verifies proof against agent’s public key and price.
• 3. If valid, Worker loads HuggingFace transformers.js model (distilbart‑cnn‑12‑6) and generates summary.
• 4. Worker returns summary and increments earnings counter in KV store.
• 5. If invalid, Worker responds 402 Payment Required with error message.

Evidence

The agent does not “learn” while it sleeps; it simply waits for requests, validates a payment, runs a model, and returns the result.

Source

Running a transformer model directly in a Worker is possible with transformers.js, which compiles the model to WebAssembly.

Source