Martin Fowler
@martinfowler.com
Software Design Loudmouth. Works for Thoughtworks. Also hikes, watches theater, and plays modern board games. He/him.
host of https://martinfowler.com
host of https://martinfowler.com
Unmesh Joshi finds LLMs to be a useful tool, but explains why their help becomes illusory if they shortcut the essential learning loop of our professional practice.
martinfowler.com/articles/llm...
martinfowler.com/articles/llm...
The Learning Loop and LLMs
How continuous learning shapes effective software development with LLMs
martinfowler.com
November 4, 2025 at 2:19 PM
Unmesh Joshi finds LLMs to be a useful tool, but explains why their help becomes illusory if they shortcut the essential learning loop of our professional practice.
martinfowler.com/articles/llm...
martinfowler.com/articles/llm...
Concerns about LLM-enabled browsers, the different perspectives on AI-Coding, and looking back at the loss of an iconic helmet.
martinfowler.com/articles/202...
martinfowler.com/articles/202...
Fragments Nov 3
a short post
martinfowler.com
November 4, 2025 at 12:45 AM
Concerns about LLM-enabled browsers, the different perspectives on AI-Coding, and looking back at the loss of an iconic helmet.
martinfowler.com/articles/202...
martinfowler.com/articles/202...
Reposted by Martin Fowler
There's got to be a massive chance of public backlash against all this AI assistant stuff
It's totally being forced down people's throats and it barely works yet
Those pushing it have a massive megaphone
But one killerapp of capitalism is people vote with their wallets
It's totally being forced down people's throats and it barely works yet
Those pushing it have a massive megaphone
But one killerapp of capitalism is people vote with their wallets
October 28, 2025 at 6:24 PM
There's got to be a massive chance of public backlash against all this AI assistant stuff
It's totally being forced down people's throats and it barely works yet
Those pushing it have a massive megaphone
But one killerapp of capitalism is people vote with their wallets
It's totally being forced down people's throats and it barely works yet
Those pushing it have a massive megaphone
But one killerapp of capitalism is people vote with their wallets
NEW POST
Agentic AI systems are amazing, but so are the security issues they raise. Korny has written a great primer on the problem - and what we can do to mitigate it.
Agentic AI systems are amazing, but so are the security issues they raise. Korny has written a great primer on the problem - and what we can do to mitigate it.
I've shared an updated version of my Agentic AI and Security article on @martinfowler.com's website:
martinfowler.com/articles/age...
(This is based on all the great research being done by others especially @simonwillison.net - my aim is to provide an accessible overview for others)
martinfowler.com/articles/age...
(This is based on all the great research being done by others especially @simonwillison.net - my aim is to provide an accessible overview for others)
October 28, 2025 at 1:27 PM
NEW POST
Agentic AI systems are amazing, but so are the security issues they raise. Korny has written a great primer on the problem - and what we can do to mitigate it.
Agentic AI systems are amazing, but so are the security issues they raise. Korny has written a great primer on the problem - and what we can do to mitigate it.
Reposted by Martin Fowler
The #1 criteria for my browser is that I trust it, and it’s secure. It’s logged into my email, Stripe account, stores my passwords and credit cards.
A browser is useful if it does all these.
These AI browsers are interesting but… with prompt injections, no way I trust them…
A browser is useful if it does all these.
These AI browsers are interesting but… with prompt injections, no way I trust them…
October 22, 2025 at 7:16 PM
The #1 criteria for my browser is that I trust it, and it’s secure. It’s logged into my email, Stripe account, stores my passwords and credit cards.
A browser is useful if it does all these.
These AI browsers are interesting but… with prompt injections, no way I trust them…
A browser is useful if it does all these.
These AI browsers are interesting but… with prompt injections, no way I trust them…
Reposted by Martin Fowler
The primary and most pervasive form of speech suppression that this Administration employs is chill; it's pervasive but much harder to quantify than outright censorship. But stories like this must be happening everywhere.
www.politico.com/news/magazin...
www.politico.com/news/magazin...
Opinion | My Bosses Were Afraid of Crossing Trump. So, I Quit.
A veteran journalist opens up on the censorship that he says led to his resignation.
www.politico.com
October 22, 2025 at 4:55 PM
The primary and most pervasive form of speech suppression that this Administration employs is chill; it's pervasive but much harder to quantify than outright censorship. But stories like this must be happening everywhere.
www.politico.com/news/magazin...
www.politico.com/news/magazin...
👀Domains and Bounded Contexts in DDD, tech industry focusing on the wrong kind of AI, story of Direct File, and U.S. government crippling its cybersecurity
martinfowler.com/articles/202...
martinfowler.com/articles/202...
Fragments and Links
a short post
martinfowler.com
October 21, 2025 at 3:05 PM
👀Domains and Bounded Contexts in DDD, tech industry focusing on the wrong kind of AI, story of Direct File, and U.S. government crippling its cybersecurity
martinfowler.com/articles/202...
martinfowler.com/articles/202...
Reposted by Martin Fowler
„…the tech industry should stop focusing so heavily on these one-size-fits-all tools, and instead concentrate on narrow, specialized A.I. tools engineered for particular problems. Because, frankly, they’re often more effective.“ www.nytimes.com/2025/10/16/o...
Opinion | Silicon Valley Is Investing in the Wrong A.I.
www.nytimes.com
October 17, 2025 at 9:18 AM
„…the tech industry should stop focusing so heavily on these one-size-fits-all tools, and instead concentrate on narrow, specialized A.I. tools engineered for particular problems. Because, frankly, they’re often more effective.“ www.nytimes.com/2025/10/16/o...
Reposted by Martin Fowler
This is so good
October 16, 2025 at 1:47 PM
This is so good
Reposted by Martin Fowler
I tried to make sense of "spec-driven development" by looking at 3 tools: Amazon's Kiro, GitHub's spec-kit, and the Tessl Framework
martinfowler.com/articles/exp...
martinfowler.com/articles/exp...
Understanding Spec-Driven-Development: Kiro, spec-kit, and Tessl
Notes from my Thoughtworks colleagues on AI-assisted software delivery
martinfowler.com
October 15, 2025 at 1:11 PM
I tried to make sense of "spec-driven development" by looking at 3 tools: Amazon's Kiro, GitHub's spec-kit, and the Tessl Framework
martinfowler.com/articles/exp...
martinfowler.com/articles/exp...
Can we use AI to help services update with their templates?
One of the challenges with service templates is that once a team instantiated a service with a template, it’s tedious to feed template updates back to those services. I wonder if anchoring AI agents to a template or reference application could help make that easier? martinfowler.com/articles/exp...
September 25, 2025 at 12:42 PM
Can we use AI to help services update with their templates?
Reposted by Martin Fowler
Don't want to say I called it, but... you know... sometimes I know what I'm talking about.
www.techdirt.com/2024/11/27/b...
www.techdirt.com/2024/11/27/b...
Brendan Carr Makes It Clear That He’s Eager To Be America’s Top Censor
When Donald Trump announced that he was appointing current FCC Commissioner Brendan Carr to be the next chair of the FCC, it was no surprise. Nor was it a surprise that Trump tried to play up that …
www.techdirt.com
September 18, 2025 at 6:05 AM
Don't want to say I called it, but... you know... sometimes I know what I'm talking about.
www.techdirt.com/2024/11/27/b...
www.techdirt.com/2024/11/27/b...
Reposted by Martin Fowler
Beth certainly will!
When I am elected, I will read the code DOGE wrote. I will find out how they stole your data and what they have done with it.
And then I will tell you
And then I will tell you
September 9, 2025 at 7:58 AM
Beth certainly will!
NEW POST
I'm about to head off on travels for the next few weeks. Before I go, here's a bunch of scattered thoughts about Gen-AI and software development
martinfowler.com/articles/202...
I'm about to head off on travels for the next few weeks. Before I go, here's a bunch of scattered thoughts about Gen-AI and software development
martinfowler.com/articles/202...
Some thoughts on LLMs and Software Development
a short post
martinfowler.com
August 28, 2025 at 2:13 PM
NEW POST
I'm about to head off on travels for the next few weeks. Before I go, here's a bunch of scattered thoughts about Gen-AI and software development
martinfowler.com/articles/202...
I'm about to head off on travels for the next few weeks. Before I go, here's a bunch of scattered thoughts about Gen-AI and software development
martinfowler.com/articles/202...
NEW POST
Legacy Modernization is hard, especially when we've lost the source code. Thiyagu Palanisamy and Chandirasekar Thiagarajan describe how AI helped them reverse engineer.
martinfowler.com/articles/bla...
Legacy Modernization is hard, especially when we've lost the source code. Thiyagu Palanisamy and Chandirasekar Thiagarajan describe how AI helped them reverse engineer.
martinfowler.com/articles/bla...
From Black Box to Blueprint
Lost source code? Legacy systems don’t have to stay black boxes—AI can turn them into blueprints
martinfowler.com
August 28, 2025 at 11:37 AM
NEW POST
Legacy Modernization is hard, especially when we've lost the source code. Thiyagu Palanisamy and Chandirasekar Thiagarajan describe how AI helped them reverse engineer.
martinfowler.com/articles/bla...
Legacy Modernization is hard, especially when we've lost the source code. Thiyagu Palanisamy and Chandirasekar Thiagarajan describe how AI helped them reverse engineer.
martinfowler.com/articles/bla...
NEW POST
The Bahmni open-source hospital management system was built with an AngularJS front-end over OpenMRS REST API. Rahul Ramesh is using AI tools to convert it to React over FHIR.
martinfowler.com/articles/res...
The Bahmni open-source hospital management system was built with an AngularJS front-end over OpenMRS REST API. Rahul Ramesh is using AI tools to convert it to React over FHIR.
martinfowler.com/articles/res...
Research, Review, Rebuild
Using an LLM to change an existing system
martinfowler.com
August 27, 2025 at 2:21 PM
NEW POST
The Bahmni open-source hospital management system was built with an AngularJS front-end over OpenMRS REST API. Rahul Ramesh is using AI tools to convert it to React over FHIR.
martinfowler.com/articles/res...
The Bahmni open-source hospital management system was built with an AngularJS front-end over OpenMRS REST API. Rahul Ramesh is using AI tools to convert it to React over FHIR.
martinfowler.com/articles/res...
NEW POST
CLI coding agents are powerful tools but don't understand your particular environment. Ben O'Mahony shows how to assemble your own.
martinfowler.com/articles/bui...
CLI coding agents are powerful tools but don't understand your particular environment. Ben O'Mahony shows how to assemble your own.
martinfowler.com/articles/bui...
Building your own CLI Coding Agent with Pydantic-AI
How to build a CLI coding agent
martinfowler.com
August 27, 2025 at 2:09 PM
NEW POST
CLI coding agents are powerful tools but don't understand your particular environment. Ben O'Mahony shows how to assemble your own.
martinfowler.com/articles/bui...
CLI coding agents are powerful tools but don't understand your particular environment. Ben O'Mahony shows how to assemble your own.
martinfowler.com/articles/bui...
NEW POST
Unmesh Joshi and I had an interesting email conversation about how when programming with an LLM he likes to grow a language of abstractions.
martinfowler.com/articles/con...
Unmesh Joshi and I had an interesting email conversation about how when programming with an LLM he likes to grow a language of abstractions.
martinfowler.com/articles/con...
Conversation: LLMs and Building Abstractions
How should we work with LLMs when growing abstractions?
martinfowler.com
August 26, 2025 at 1:34 PM
NEW POST
Unmesh Joshi and I had an interesting email conversation about how when programming with an LLM he likes to grow a language of abstractions.
martinfowler.com/articles/con...
Unmesh Joshi and I had an interesting email conversation about how when programming with an LLM he likes to grow a language of abstractions.
martinfowler.com/articles/con...
NEW POST
When I used to give talks, I was very good at keeping them to the allotted time, even with a last-minute change. Key tool: Expansion Joints.
martinfowler.com/bliki/Expans...
When I used to give talks, I was very good at keeping them to the allotted time, even with a last-minute change. Key tool: Expansion Joints.
martinfowler.com/bliki/Expans...
bliki: Expansion Joints
a bliki entry for Expansion Joints
martinfowler.com
August 18, 2025 at 3:19 PM
NEW POST
When I used to give talks, I was very good at keeping them to the allotted time, even with a last-minute change. Key tool: Expansion Joints.
martinfowler.com/bliki/Expans...
When I used to give talks, I was very good at keeping them to the allotted time, even with a last-minute change. Key tool: Expansion Joints.
martinfowler.com/bliki/Expans...
NEW POST
OKRs are popular but often set in a top-down cascade. Paulo Caroli describes a path for teams to create their own, in a collaborative process to align with broader strategy.
martinfowler.com/articles/tea...
OKRs are popular but often set in a top-down cascade. Paulo Caroli describes a path for teams to create their own, in a collaborative process to align with broader strategy.
martinfowler.com/articles/tea...
Team OKRs in Action
Rather than cascade OKRs, use collaborative alignment.
martinfowler.com
August 13, 2025 at 2:22 PM
NEW POST
OKRs are popular but often set in a top-down cascade. Paulo Caroli describes a path for teams to create their own, in a collaborative process to align with broader strategy.
martinfowler.com/articles/tea...
OKRs are popular but often set in a top-down cascade. Paulo Caroli describes a path for teams to create their own, in a collaborative process to align with broader strategy.
martinfowler.com/articles/tea...
FINAL §
@srny.bsky.social concludes his article in impact intelligence by addressing five common objections to this activity, including slowing down, lack of agility and collaboration, and the unpredictability of innovation.
martinfowler.com/articles/imp...
@srny.bsky.social concludes his article in impact intelligence by addressing five common objections to this activity, including slowing down, lack of agility and collaboration, and the unpredictability of innovation.
martinfowler.com/articles/imp...
The Reformist CTO’s Guide to Impact Intelligence
A guide for technology leaders and their teams to shift the organization
martinfowler.com
August 12, 2025 at 1:13 PM
FINAL §
@srny.bsky.social concludes his article in impact intelligence by addressing five common objections to this activity, including slowing down, lack of agility and collaboration, and the unpredictability of innovation.
martinfowler.com/articles/imp...
@srny.bsky.social concludes his article in impact intelligence by addressing five common objections to this activity, including slowing down, lack of agility and collaboration, and the unpredictability of innovation.
martinfowler.com/articles/imp...
Reposted by Martin Fowler
This is why the whole thing is so hard to trust and if you have any critical thinking, you stay away from such rankings. The same is true for Gartner, where many vendors also pay very heavy sums to Gartner to have Gartner evaulate them
August 9, 2025 at 7:46 AM
This is why the whole thing is so hard to trust and if you have any critical thinking, you stay away from such rankings. The same is true for Gartner, where many vendors also pay very heavy sums to Gartner to have Gartner evaulate them
Reposted by Martin Fowler
So to be impressed by a machine going way beyond the question asked feels like the technical implementation of this exact anti-pattern. Worse, all that generated output creates significant gravity/path dependence, your brain turns off all that much faster. Why do we assume this is good?
August 7, 2025 at 7:54 PM
So to be impressed by a machine going way beyond the question asked feels like the technical implementation of this exact anti-pattern. Worse, all that generated output creates significant gravity/path dependence, your brain turns off all that much faster. Why do we assume this is good?