The New Stack Podcast The reason AI agents shouldn’t touch your source code — and what they should do instead
8 snips
Feb 13, 2026 Alois Reitbauer, Chief Technology Strategist at Dynatrace focused on observability and autonomous operations. He explains why AI agents should change configurations with feature flags instead of rewriting code. He covers integrating feature management with observability, using flags as safety guardrails, and how constrained AI actions enable safer automated operations.
AI Snips
Chapters
Transcript
Episode notes
Feature Flags Are Observability’s Control Knob
- Feature management and observability are now tightly entwined and need each other to understand flag impact.
- Changing config via flags is safer and faster than letting agents rewrite and deploy code in production.
Quick Post-Acquisition Integration
- Two weeks after the DevCycle acquisition Dynatrace enabled freemium access and initial product integration.
- Dynatrace users could already log into DevCycle and use observability to understand and tweak feature flags.
Give Agents Guardrails, Not Free Rein
- Allow AI agents to act only within guardrails and predefined workflows like runbooks.
- Let agents toggle feature flags or configuration options instead of making unsupervised sweeping changes.