From vibe coding to spec-driven development
summary
Claims a move from ad hoc AI-driven prototyping to repeatable pipelines, demonstrating a tool called Kirro that generates requirements, design, and task artifacts from an intent and uses steering documents and MCP servers for context. Shows spec files versioned in the repo, an implementation plan broken into queued tasks, and code generation with diffs and approval controls. Reports fixes flow by editing specs and rerunning generation, yielding more consistent, reviewable outputs than vibe coding.
AI-written summary · about
chapters (13)
- 0:00Welcome
- 1:21AI coding tools evolution
- 3:12Vibe coding challenges
- 4:54Structured AI coding
- 8:01Spec-driven development intro
- 11:31Kirro for spec-driven dev
- 13:02Pie app with specs
- 23:30Code from specs in Kirro
- 28:18Testing generated app
- 31:09Adding new features
- 32:37Spec-driven on brownfield
- 35:17Spec-driven lifecycle
- 36:30Try Kirro and learn
related talks
25mFrom Vibe Coding to Spec-Driven Development
18mUsing Spec-Driven Development for Production Workflows
57mFrom Vibe Chaos to the Rise of Spec-Driven
45mSpec Driven Development with Kiro: Ensuring Quality and Traceability in the AI Era
28mHow To Use Spec-Driven Development for Production Workflows
25mFrom Vibe Coding to Spec-Driven Development
25mspec-driven development
46mRethinking Full Stack Application Development with Kiro