
Hallucinations and Vibe Coding

The Future of Code Review: Stop Reviewing Line-by-Line, Start Governing AI Agents - Itamar Friedman
Topic 2 of 11
Published May 4, 2026 · Clip 7:53–16:16 (8:23)
The host raises hallucination concerns about AI-generated code. Itamar explains that AI coding assistants are layered: a raw LLM hallucinates like a new developer with no context, but tools and platforms add planning, background verification agents, and opinionated review tuned to company standards, turning 'vibe coding' into 'viable' or grounded coding. They discuss whether linters, static analysis, and tests are still relevant: yes, but they lack the big picture of intent, requirements, and architecture, so they must become part of a governance platform. Semantic tribal knowledge like RCAs and architectural rules can now be codified thanks to LLMs.