An engineering leader with a background in highly regulated systems like aviation and fintech discusses his move into the AI space.
Agile principles were compromised by corporate pressures, leading to a similar rigid, plan-driven approach in modern AI agent development.
Current AI agents follow initial strategies without fundamental reassessment, lacking the human-like instinct to stop and rethink when new information emerges.
Moving beyond the black box model requires robust logging and chain-of-thought transparency to ensure accountability in high-stakes applications.
A new architecture allows agents to track assumptions versus knowledge, triggering a strategic reset to re-evaluate the entire plan when significant divergence occurs.
Unlike humans who are attached to their past efforts, AI agents can restart tasks without emotional cost, an advantage that should be built into their workflows.
AI tools are democratizing software development, shifting the focus from syntax memorization to architectural thinking and empowering senior engineers.
Jobs that call for the skills explored in this talk.