
Beyond Coding How to Think About Software Engineering (CTO's Perspective)
Feb 18, 2026
Joris Conijn, AWS CTO at Xebia and leader in automation and developer workflows. He discusses agentic tooling, system design, and why banning AI creates security risks. Short takes cover using AI to automate SDLC tasks, preserving engineering fundamentals, testing AI-written code, and checking AI agents into Git. The conversation spotlights the shift from just coding to designing resilient systems.
AI Snips
Chapters
Transcript
Episode notes
Avoid Banning AI At Work
- Don't ban AI outright; it pushes people to use shadow tools and increases data leakage risk.
- Offer approved AI solutions and guardrails instead of blocking usage entirely.
Embed Tech Debt Into Features
- Embed technical debt work into feature scope so developers can address infra while delivering product changes.
- Explain dependencies to stakeholders: fix the underlying issue first to speed future changes.
Senior Knowledge Enables Safe AI Coding
- AI coding assistants work today because senior developers understand underlying concepts and can validate changes.
- In the future, lack of foundational knowledge risks poor decisions when most code is generated by models.
