~/talk

How to Verify AI-Generated Code with Codex

on YouTube 48m
Talkat Sonar Summit 2026 · on Tuesday, 2026-03-03· [ai-assisted-development, security]

speakers

name:Vaibhav Srivastav
website:https://vaibhavs10.github.io
name:Tom Howlett
links:none on file
name:JOHN CLIFTON
links:none on file

resources

video:https://youtu.be/m3NFhJTTMdc
slides:none on file

summary

OpenAI uses Codex to review every pull request and assist across app, CLI and IDE surfaces; Codex 5.3 Spark runs ~60% faster than 5.2, spins up sub‑agents to parallelize testing, auto‑fixes CI failures and even bridges to Adobe Premiere Pro to edit videos. Sonar integrates via MCP so Sonar’s deterministic static analysis and taint/path checks run programmatically against agent output. Recommendations: sandbox agents, add agents.md, enforce MCP checks, run unit tests and nightly verifications before merging.

AI-written summary · about

chapters (28)

  1. 0:00Introduction
  2. 0:17Why AI Coding Agents Need Guardrails
  3. 1:37How OpenAI Ships Software Using Codex
  4. 2:10Codex Automatically Reviews Every Pull Request
  5. 3:13Designers and Product Managers Using AI Coding Agents
  6. 4:38Human Code Review Based on Risk Level
  7. 6:00Sharing Codex Agent Workflows Across Engineering Teams
  8. 7:44AI Agents Automatically Fixing CI Pipeline Failures
  9. 8:50Deterministic Checks to Build Trust in AI Development
  10. 11:17Balancing Development Speed with Production Safety
  11. 13:50Giving AI Agents Power with Safe Boundaries
  12. 15:40Enterprise Controls for AI Agent Actions
  13. 16:23Maintaining Engineering Standards with AI Agents
  14. 17:45Why New Developers Need Stronger Guardrails with AI
  15. 18:28Codex Designed for Every Type of Builder
  16. 19:41Repo-Wide Context Improves AI Code Review Quality
  17. 21:03Codex and Sonar Catch Different Types of Risks
  18. 22:46Verification Helps AI Coding Agents Learn and Improve
  19. 23:18Sub-Agents Splitting Work by Specialized Tasks
  20. 25:11Planning First to Scope Complex AI Coding Tasks
  21. 27:57Verifying Long-Running AI Agent Tasks Frequently
  22. 30:12Using Codex to Understand Legacy Codebases
  23. 33:00Why Automated Tests Still Matter in Legacy Systems
  24. 35:40How Enterprises Begin Safe AI Agent Adoption
  25. 37:19Safe Experimentation to Discover What Works
  26. 40:12Starting with agents.md and MCP Tools
  27. 44:33Practical First Steps for Teams Adopting AI Agents
  28. 47:19Final Takeaway: AI Development Speed Requires Strong Controls

related talks

~/search
type to search 6,249 conferences, 138,987 speakers and 185,172 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.