What does the future of software engineering look like?
Deer Valley, Utah | Feb. 1-3, 2026Twenty-five years ago, in the mountains of Utah, a small group of technologists gathered to rethink how software is built. Their ideas ignited what would become the agile movement, setting a new direction for the industry.
In February 2026, we returned, not to memorialize the past, but to confront a new inflection point: the shift to AI-native software development. Hosted by Martin Fowler and Thoughtworks, the event brought together a small group of practitioners, researchers and enterprise leaders to ask what responsible and effective software development looks like in an era defined by AI.
Below is a report on the findings and insights from the conversations that took place.
Key insights and themes from the event
The range of topics and ideas discussed at the event was wide-ranging. However, a number of key themes emerged that form the foundations for future exploration, experimentation and reflection:
- The rigor has to go somewhere. As AI agents produce more and more code, the engineering discipline doesn't disappear but instead moves elsewhere. Chad Fowler's framing was widely regarded as invaluable: if we stop caring about the code, our rigor has to move somewhere else.
- Supervisory engineering: A new middle loop. Conversations identified a type of work that doesn't yet have a name. Neither writing code nor release management, it sits between them: directing agents, evaluating their output, calibrating trust, encoding standards and defining the constraints within which agents can safely operate.
- Engineering is becoming risk management. If code production is becoming faster and cheaper, understanding where risk lives, how changes propagate through a system and what requires human attention is critical.
- Self-healing systems and the prerequisites that matter. What helps agents also helps humans. Investing in better incident response processes, clearer documentation and stronger observability isn't just an "AI readiness" play, it will ultimately make systems more operable for everyone. Looking further into the future, the event also discussed the notion of an "agent subsconcious" — a knowledge graph built from years of post-mortems.
- Teams, roles and organizational design. The relationship between functions — like product and engineering — and even within teams, across generations and levels of experience remain open issues that demand continued attention from the industry.
- Semantic layers, knowledge graphs and the data foundation. Semantic layers and knowledge graphs are rapidly becoming the agent interaction layer; they are the way agents understand your business domain, validate business logic and maintain a source of truth that transcends any individual microservice or application.
- Programming languages, source code and what comes next. Conversations explored whether source code will eventually disappear — or at least stop being the primary artifact humans interact with. A future language designed for AI agents would be terse and type-safe, rely on tooling for verification and provide "projections" that let humans view the underlying representation in whatever form is most useful.
- Security: The uncomfortable gap. A small but worried group noted that security consistently gets deprioritized in AI adoption. Granting agents broad tool access — especially to email, which can enable password resets and account takeovers — was flagged as a specific and immediate risk.
- The enterprise reality check. A healthy tension existed between what's technically possible and what enterprises will actually adopt. AI-native platforms sound compelling in theory, but most sessions acknowledged the gap between small-scale experiments and deployment across tens of thousands of engineers.
We kept asking the same question in every room: if AI handles the code, where does the engineering actually go? Nobody had the same answer. But everybody agreed the question is urgent.
We kept asking the same question in every room: if AI handles the code, where does the engineering actually go? Nobody had the same answer. But everybody agreed the question is urgent.