Integrating Agentic AI Into Internal Developer Platforms To Minimize Errors and Downtime
Talkat Platform Engineering Day North America 2025 · on Monday, 2025-11-10· at 09:10· Building B | Level 5 | Thomas Murphy Ballroom 2-3·
summary
A GitOps pipeline enhancement that uses kagent agents and an LLM to run semantic validation on PRs, invoked by a GitHub Action, that pulls and does Helm rendering, queries the live cluster (get secrets, node labels, quotas) and checks Kyverno-like policy conflicts. The agent posts human-readable PR reviews that flag CPU/resource quota oversubscribe, invalid nodeSelectors/tolerations, and missing secrets, proposes fixes, and in the demo the suggested PR is merged and Argo CD deploys the app successfully.
AI-written summary · about
chapters (6)
- 0:00Agentic AI in developer platforms
- 1:47K Agent and core components
- 4:56Semantic validation in GitOps
- 7:04Live demo: K Agent in action
- 19:52K Agent deep analysis
- 22:05Q&A on agentic AI
related talks
24mEmpowering the GitOps Future with AI
24mFix Production Rollouts on the Fly With Agentic AIOps
23mGitOps Meets AIOps: Building Reliable AI Pipelines With Argo
40mSecure by Default: Building an AI-Augmented, OSS-Powered Reusable CI/CD Pipeline
44mAgentic Delivery, Guardrailed: AI in CI/CD & Platform Engineering (Open Source Patterns)
31mAgents at Work: How GitLab Is Redefining AI Adoption for Engineering Teams
26mBeyond Argo Rollouts: Boosting Developer Experience With Intelligen...
36mAgentic CI/CD: from Pull Request to Production without the Paper Cuts