The AI Native Dev - from Copilot today to AI Native Software Development tomorrow Why Faster AI Development Often Increases Rework | Cian Clarke
10 snips
Dec 23, 2025 Cian Clarke, Head of AI at NearForm, dives into the innovative BMAD approach, emphasizing clarity over rapid coding. He discusses the risks of 'vibe coding' and how BMAD helps in creating maintainable software. Cian also highlights the importance of understanding developer archetypes while tailoring spec-driven tools and the cultural shift required to scale AI-native software. As AI's role evolves, he debates how senior engineers can leverage their expertise more effectively in this changing landscape.
AI Snips
Chapters
Transcript
Episode notes
How NearForm Stumbled On BMAD
- NearForm discovered BMAD during internal Ignite discovery sessions that produced rich specs from meetings.
- A team member, James, found BMAD and it turned Cian from a vibe-coding skeptic into a spec-driven believer.
Ambiguity Drives Hallucination And Rework
- Ambiguity in requirements causes models to hallucinate features and create wasted work.
- Spec-driven context engineering reduces hallucination by grounding models in a clear "what" before coding.
Choose BMAD For Shippable Greenfield Work
- Use BMAD for greenfield projects and MVPs where you need shippable quality, not rapid throwaway prototypes.
- Avoid heavy spec tooling for fast prototypes; prefer lighter vibe-coding tools for throwaway experiments.