The AI-native SDLC: agents across planning, coding, testing and deployment

Pavan_Belagatti · x · 2026-09-12

Pavan Belagatti describes the AI-native SDLC: AI agents participate across planning, design, coding, testing, deployment and maintenance. Developers can enter anywhere — capturing intent, defining requirements, building reusable skills, running subagents, evaluating outputs, reviewing PRs, triggering CI/CD — all connected by continuous feedback loops rather than a rigid linear process, with production learnings flowing back into planning. He argues agents can't work reliably on code alone: they need organizational context like service ownership, engineering standards, dependencies, deployment history and real-time ops data, which an Agentic SDLC platform must provide.

Original post →

More from coding & agent

coding & agent channel →