# fowler-thoughtworks-retreat-llm-software-development-2026-02-13

## Veille

Thoughtworks retreat on the future of software development with LLMs — reflections on organizational impact, cognitive debt, and supervised programming

## Titre Article

Fragments: February 13

## Date

2026-02-13

## URL

https://martinfowler.com/fragments/2026-02-13.html

## Keywords

LLM, software development, AI agents, cognitive debt, developer experience, work intensification, supervised programming, software architecture, Thoughtworks, team size, pair programming, refactoring, IDE, burnout

## Authors

Martin Fowler

## Ton

**Profile**: First-person narrative perspective, reflective professional register, intermediate-to-advanced technical level.

**Description**: The tone is that of a seasoned, respected observer of the software industry, reporting on and synthesizing discussions from a high-level peer retreat. The writing is characteristic of Fowler: sober, nuanced, avoiding hyperbole. It alternates between factual accounts of other experts' contributions and his own measured reflections. The style is fragmentary and deliberately inconclusive — it raises questions rather than imposing answers. Authority comes from experience and network (the cited speakers are all recognized figures). The target audience consists of senior developers, architects, and technical leaders seeking to understand the ongoing transformations without succumbing to blind enthusiasm or outright rejection.

## Pense-betes

- Senior developers are largely optimistic about LLMs, especially when they focus on architecture — they compare managing LLM agents to managing junior developers
- A third of initially resistant seniors change their minds after hands-on exercises with LLMs
- Mid-level developers are the most vulnerable: careers built before LLMs, but without the senior expertise to steer agents
- Key concept from Margaret-Anne Storey: "cognitive debt" — when a team can no longer explain its design decisions, it can no longer evolve the code
- Important distinction between "cruft" (degradation through ignorance) and "debt" (a deliberate, conscious cost)
- Laura Tacho: the Venn diagram between developer experience and agent experience is a circle — what helps developers helps agents
- Critical observation: leaders make accommodations for LLMs that they refused to make for humans
- HBR research (Ranganathan & Ye): AI adoption causes work intensification and burnout, with initial gains giving way to a decline in quality
- Camille Fournier: "everyone becomes a manager" — supervised programming generates context-switching fatigue
- "Two-pizza" teams keep the same size but produce more — the question of pair programming with agents remains open
- IDEs are evolving toward a hybrid model: non-deterministic tasks (LLMs) + deterministic tasks (refactoring), with new orchestration possibilities

## RésuméDe400mots

Martin Fowler shares his reflections from the Thoughtworks retreat on the future of software development, an event bringing together industry experts to examine the impact of LLMs on development practices.

First observation: senior developers are largely optimistic about LLMs. Their approach is to focus on architecture and treat AI agents as junior developers to be supervised. Notably, a third of initially skeptical seniors change their minds after hands-on exercises. Mid-level developers, by contrast, find themselves in a difficult position: their careers were built before the LLM era, yet they do not yet have the senior expertise needed to effectively orchestrate these tools.

Margaret-Anne Storey introduces the concept of "cognitive debt," describing the situation where a team becomes unable to modify its code because it can no longer explain the underlying design decisions. Fowler draws a distinction between "cruft" — unintentional degradation through ignorance — and true technical debt, which involves a conscious choice and a calculated cost.

Laura Tacho offers a striking observation: the Venn diagram between developer experience and agent experience is a perfect circle. Everything that makes work easier for human developers also makes it easier for agents. A telling paradox: leaders are willing to make accommodations for LLMs (documentation, code clarity, clean environments) that they stubbornly refused to make for their human teams.

On the IDE front, the trend is moving toward a hybrid model combining non-deterministic tasks handled by LLMs and deterministic tasks like refactoring, opening up new orchestration possibilities.

On team size, the consensus is that "two-pizza" teams will keep their size but increase their productivity. The question of pair programming with agents remains open and promising.

Research published in the Harvard Business Review by Ranganathan and Ye offers an important counterpoint: AI adoption leads to work intensification and burnout. Initial productivity gains give way to quality degradation over the medium term.

Camille Fournier sums up this tension with the phrase "everyone becomes a manager": supervised programming turns every developer into an agent manager, generating fatigue from constant context switching. This new paradigm demands supervisory skills more than direct execution skills.

## GrapheDeConnaissance

- Martin Fowler —fait_partie_de→ Thoughtworks Future of Software Development Retreat (EVENEMENT, 0.98)
- Thoughtworks —a_créé→ Thoughtworks Future of Software Development Retreat (EVENEMENT, 0.95)
- Margaret-Anne Storey —a_créé→ dette cognitive (CONCEPT, 0.95)
- Laura Tacho —affirme_que→ expérience développeur équivaut à expérience agent (AFFIRMATION, 0.93)
- Camille Fournier —affirme_que→ tout le monde devient manager (AFFIRMATION, 0.92)
- Ranganathan & Ye —publie→ recherche intensification travail IA (DOCUMENT, 0.9)
- LLM —s_applique_à→ développement logiciel (CONCEPT, 0.95)
- Développeurs seniors —affirme_que→ les agents LLM sont comparables à des développeurs juniors (AFFIRMATION, 0.88)
- Martin Fowler —affirme_que→ les développeurs intermédiaires sont les plus vulnérables face aux LLM (AFFIRMATION, 0.85)
- Ranganathan & Ye —affirme_que→ l'adoption de l'IA provoque intensification du travail et épuisement professionnel (AFFIRMATION, 0.88)
- IDE —converge_avec→ modèle hybride déterministe/non-déterministe (CONCEPT, 0.82)
- Programmation supervisée —permet→ fatigue de changement de contexte (CONCEPT, 0.85)
- Dette cognitive —réduit→ capacité d'évolution du code (CONCEPT, 0.9)

---
Canonical: https://www.thekb.eu/en/fiches/fowler-thoughtworks-retreat-llm-software-development-2026-02-13/