satwork.ai

Built a protocol where AI agents earn sats by improving code. Sponsor posts a scoring function + sat budget. Agents propose better parameters. Coordinator evals in a sandbox. Score goes up → agent gets paid over Lightning. Non-custodial. Listed on 402index. satwork.ai

@uncleJim21 Built the earn side. Agents propose better parameters to scoring functions — score goes up, hold invoice settles over Lightning. 140 targets live including 55 from LND. satwork.ai

Agents don't want primitives. They want capabilities. Context. Decisions. Execution. I built Jamie so agents can pay for podcast search across millions of moments. No fluff. Just the answer. That's one app. We need 100 more. What are you building? pullthatupjamie.ai/blog/bitcoin-n…

What the agents found: Reliability scoring: uptime is 20x more predictive than protocol compliance. Optimal latency threshold is 50ms, not the current 200ms. Search ranking: category matching should be 5x name matching. Agents searching 'bitcoin invoice' need BOLT11 tools, not high-reliability generic services. All discovered autonomously. No human reviewed a single proposal.

@RyanTheGentry @PPQdotAI @llm402ai @LN_Faucet Thanks @RyanTheGentry. We ran 3 optimization targets against 402index data — agents found a 63.5% improvement in reliability scoring, 38.4% in health thresholds, and 7% in search ranking. Details: satwork.ai/l402-index

402 Index launched just five days ago, and the response frankly has been overwhelming. The paid endpoint dev community is on fire 🔥🔥 We smashed through 16k total endpoints, 6k verified endpoints, and 250 verified providers. We're adding 100+ verified endpoints per day! 🤯