Posts
Essays on AI and platform engineering for technical executives, newest first.
-
When Accuracy Misleads: Cost Matrix Metrics and Thresholds
• UpdatedWhen does accuracy mislead? A sealed, pre-registered SECOM evaluation: the top model by accuracy catches zero defects until the threshold matches costs.
-
Verifying x402 Endpoints: The 5-Check Buyer Recipe
• UpdatedHow to verify an x402 endpoint before you pay: 5 free or cheap checks, 4 of 4 hostile-seller failure modes detected, 1 of 3 live endpoints diverged.
-
Pay-Per-Call APIs: $0.005 Calls, Measured Treasury Costs
• UpdatedSub-cent API calls clear the card fee floor and settle at request; measured USDC off-ramp sweeps show deterministic treasury costs for pay-per-call revenue.
-
x402 on Cloudflare Workers: AI Agents Pay Per Call in USDC
• UpdatedAI agents pay per call in USDC on Base and Solana; no accounts, no API keys. The full HTTP 402 handshake, documented from a production deployment.
-
When Should an AI Agent Enumerate, and When Delegate?
• UpdatedA sealed 240-call exhibit grounds the boundary between decision spaces an agent should enumerate exactly and those it should delegate to a typed judge.
-
Does WebMCP Pay Off for Content Sites?
• UpdatedAcross 135 agent runs on a production blog, shrinking a four-tool WebMCP surface yielded little median-token benefit; relevance ordering mattered more.