Idan Gazit, GitHub, ETLS Vegas (29)
summary
Argues that adding an AI layer to the existing compile-and-build stack is inevitable but risky because compilers are deterministic and AI is not, so teams must add runtime checks and verification gates and design for nondeterminism. Predicts coding in natural language will push a shift to a tree of specs—small, testable modules—and shows how immediate feedback like the Solver live-sidebar model speeds iteration, making small janky apps acceptable during gradual rollout.
AI-written summary · about
related talks
46mPanel Discussion: AI's Impact on Developers
36mEngineering Intent: Solving the AI Productivity Paradox
36mHarnessing the Power of Nondeterminism for Software Engineering
45mThe Rebirth of Software Craftsmanship in the AI Era
27mThe new gravity of software engineering in the AI-first era
43mWill AI kill the developers job?
40mThe Developer RenAIssance: The Humans Powering the Age of AI
14mThe 4 Patterns of AI Native Development