Database › Datadog
Datadog
No formal mandate reported, though Anthropic says all Datadog engineers use AI tools for production code.[1]
Tools
Metrics tracked
- All Datadog engineers use AI tools for production code, and Claude Code drives at least two-thirds of that use, according to Anthropic.[1]
- Codex reviews every PR in one major repository, used by more than 1,000 engineers, according to OpenAI.[2]
Guardrails
- Datadog replayed past incidents and found Codex review feedback would have helped in about 22% of them.[2]
- Datadog follows 'harness-first engineering': instead of reading every line of agent-generated code, it invests in automated checks and production telemetry that verify the code is correct.[3]
Hiring, reviews & process changes
Nothing verified from a public source yet. Know a source?
Reported results
- An internal tool (Helix) was built in days with Claude Code.[1]
- BitsEvolve, an LLM-guided optimizer, shipped 10x speedups on key ingestion functions, verified against live traffic.[3]
Metric tags: Adoption rate, Code review, Quality / defects
Editor's note: Adoption and review figures are vendor-reported (Anthropic and OpenAI customer stories). openai.com returns a bot challenge to automated checks but loads for readers.
Sources
- Claude: How Datadog built a universal machine tool for Claude Code · 2026-07-21 · vendor source
- OpenAI: Datadog customer story · 2026-01-09 · vendor source
- Datadog blog: Harness-first agents · 2026-03-09 · company source
Last verified 2026-10-25. Industry and size are editorial bands; see methodology. Spot an error? Submit a correction.