Outcomes
- Run a fleet of production agents with monitoring, evals, and incident logs
- Design handoffs, retries, and escalation between agents
- Ship an agent fleet that survives a 72-hour unattended run
- Operate agents the way an EM runs a team — briefs, reviews, postmortems
Who it's for
Builders who have already shipped with LLM APIs and want production-grade agent operations. Application required — this is not for first-time agent builders.
Syllabus
Week 1Agent architectures that survive contact with production
Week 2Tooling, memory, and context budgets
Week 3Multi-agent orchestration — handoffs, retries, escalation
Week 4Evals and monitoring for autonomous systems
Week 5Fleet week — deploy and instrument your fleet
Week 6The 72-hour unattended run + incident review
Capstone
An agent fleet that survives a 72-hour unattended production run — with the incident log to prove it.