David Ondrej and Kun Chen demonstrate First Mate, Kun Chen's supervising agentAgent orchestration coordinates AI agents, tools, people, tasks, state, and control flow so a larger workflow reaches a verified outcome. for running multiple coding workersA multi-agent system contains multiple AI agents that interact, coordinate, divide work, or influence one another while pursuing tasks. in parallel while preserving one place for planning, delegationAI agent delegation assigns a bounded task, context, authority, and expected result from one participant to an agent or another agent. and progress tracking.
Kun Chen describes an escalation ladder in which workers resolve routine choices, raise ambiguous product decisions to the supervisor and involve the humanHuman-in-the-loop AI keeps a person involved in reviewing, correcting, approving, or guiding an AI system's work. only when a trade-off needs judgment. David Ondrej tests the approach by steering a live build and examining how the agent recovers from mistakes.
David Ondrej and Kun Chen emphasize validation before completionAgent completion verification checks evidence that an agent achieved the requested outcome rather than merely producing intermediate work.. Generated changes are not treated as finished until tests, linting and structured checks pass, and the supervisor can reassign failures for repair.
Kun Chen calls the broader interface idea AXI: software should expose agent-efficient commands and machine-readable surfaces instead of forcing agents through visual interfaces. David Ondrej connects this to headless, API-first product design.
Watch on YouTube




