Datadog uses AI agents to test software written by AI
Datadog is betting on agentic testing through Propolis as more teams use AI to generate code. The company argues that fixed scripted tests cannot keep pace with software written by AI. The goal is for testing agents to adapt to code produced by coding agents.
Datadog is positioning Propolis around agentic testing as AI-generated code becomes more common in development workflows. The company contends that conventional scripted test suites struggle to match the pace of software produced by coding agents. Its aim is for testing agents to adjust to code written by AI agents, linking automated code generation with automated validation.
If such adaptive testing works as intended, it could affect developers, QA teams, and organizations relying on AI-generated code. Faster, more flexible validation may reduce some review burdens, but it could also create new dependencies on agent behavior and raise questions about trust, oversight, and accountability. The broader impact may depend on how reliably these systems explain failures and how well teams integrate them into existing development practices.