I built Zero Human Labs after spending too much time babysitting AI agent pipelines — manually routing tasks, watching API costs spiral, and debugging coordination failures. With Zero Human Labs, you define your entire agent team (devs, designers, ops, etc.) in one YAML file. Tasks are routed automatically via a sealed-bid auction — each agent bids based on specialization and track record, and the best one wins the task. No manual assignment, no coordination overhead. Key things that make it different: Per-agent wallets + org-level budget caps → no runaway API spend Circuit breakers → agents that misbehave get frozen automatically Smart model routing → saves 30–80% on AI API costs by matching request complexity to the right model Governance presets calibrated from 146 SWARM simulation runs across 43 agent types Free demo at zero-human-labs.com — 1 agent, 1 example workflow, open-source models. Would love feedback from the HN crowd on the auction routing mechanism especially.