Datadog uses Codex for system-level code review

Source: OpenAI Date: 2026-01-09 URL: https://openai.com/index/datadog

Summary

Title-only: A case study featuring Datadog — the cloud monitoring and observability platform — using OpenAI’s Codex for system-level code review. “System-level” suggests Codex is reviewing infrastructure code, configuration management, or distributed systems code (rather than application-level code), which is a higher-stakes and more complex use case. Datadog’s massive codebase and infrastructure complexity make this a meaningful proof-of-concept for Codex at scale.

Implications

The infrastructure code review thread. System-level code review by AI is qualitatively harder than application code review — infrastructure code has complex failure modes, dependencies on runtime state, and security implications that require deep context. Datadog using Codex for this validates that GPT-5-class reasoning capability is sufficient for reasoning about complex distributed systems code, not just writing CRUD endpoints.

Observability company as AI pilot. Datadog building on Codex is strategically notable because observability companies are the ones that will instrument AI agent deployments. Datadog using Codex internally creates a feedback loop: they observe how AI coding agents behave in production, which informs their AI observability product development. The case study is Datadog demonstrating that Codex is mature enough to trust in their own engineering workflows.