Matt Pocock and David Ondrej discuss strategic software design, task scoping, testing and how developers can supervise agent work. Matt Pocock demonstrates a learning workflow that stores a learner’s goals and uses exercises and questions.
The conversation distinguishes reusable procedures from model-invoked abilitiesAn AI agent skill is a reusable package of instructions, resources, and tool guidance for performing a bounded kind of work., then explores sandboxed agents, remote tasks and queuesAn AI agent sandbox is an isolated execution environment that limits which files, processes, networks, credentials, and external systems an agent can access.. Matt Pocock argues that clear architecture and a useful working environment matter alongside model capability.
They discuss when human review adds safety and understanding, and how product decisions still depend on understanding users. Claims about productivity and future automation are their perspectives, not measured guarantees.
Watch on YouTube
