The Archaeologist’s Copilot
Lessons from modernizing a Java 1.5 codebase with AI, a stable Docker environment, and test-guided incremental refactoring.
Martin Fowler
@martin-fowler
Publisher
33
Posts
Posts by Martin Fowler
DSLs Enable Reliable Use of LLMs
Why Domain-Specific Languages and good abstractions make LLM code generation reliable and reviewable.
Experiences with local models for coding
Notes from my Thoughtworks colleagues on AI-assisted software delivery
Viability of local models for coding
Notes from my Thoughtworks colleagues on AI-assisted software delivery
Building Reliable Agentic AI Systems
AI helping pharmaceutical researchers query decades of information buried in PDF reports
The VibeSec Reckoning
Engineering guardrails teams must adopt to move safely from vibe-coded AI prototype to production.
Maintainability sensors for coding agents
A practical walkthrough of computational sensors on the path to production, with a deep dive on ESLint and static analysis as feedback for coding agents.
Bliki: Interrogatory LLM
To give an LLM context, I can ask the LLM to interview me for it
Bliki: Mythical Man Month
a bliki entry for Mythical Man Month
Structured-Prompt-Driven Development (SPDD)
A structured approach to using prompts to guide AI-assisted programming.
Feedback Flywheel
Teams have always had mechanisms for collective learning. Retrospectives, post-incident reviews, lunch-and-learns. The best of these share a property: they convert individual experience into shared practice. What one person encountered in a debugging session or a production incident b...
Principles of Mechanical Sympathy
Principles of mechanical sympathy for creating performant software: Memory access patterns, false sharing, the single-writer principle, and natural batching.
Harness engineering for coding agent users
A mental model for building trust in coding agents through feedforward guides, feedback sensors, and iterative harness engineering.
Encoding Team Standards
When a team has worked together long enough, certain practices become invisible. The senior engineer who rejects a pull request does not consult a checklist; she recognizes, almost instantly, that the error handling is incomplete, that the abstraction is premature, that the naming doe...
Bliki: Architecture Decision Record
A short document that captures a single decision, structured into a log to provide a history
Context Anchoring
Externalize decision context into a living document shared between model sessions
Ideological Resistance to Patents, Followed by Reluctant Pragmatism
A personal account of moving from ideological opposition to patents toward a defensive, pragmatic approach for protecting innovation.
Humans and Agents in Software Engineering Loops
Notes from my Thoughtworks colleagues on AI-assisted software delivery
Design-First Collaboration
Before writing code, dialog with progressive levels of design alignment