Birgitta B

How far can we push AI autonomy in code generation? An experiment to test the limits of autonomous code generation by LLMs

We recently ran an experiment to explore how far GenAI can currently be pushed toward autonomously developing high-quality, up-to-date software without human intervention, and gather observations about where it breaks down.

martinfowler.com/articles/pus...

Das Individuum in der Maschine: Meredith Whittaker über die Rückgewinnung der Privatsphäre im Zeitalter der KI

Gemeinsam mit Publix und @algorithmwatch.org laden wir ein zum Gespräch mit @meredithmeredith.bsky.social, Präsidentin des Messengers Signal, über die Frage, wie wir Technologie wieder stärker an menschlichen Bedürfnissen ausrichten können – vor allem beim Schutz unserer Privatsphäre.
➡️ t.ly/lIzA3

I totally get it, I’m also tired of much of the public discourse. But here is one more argument for experienced devs like the author: If we want to guide & teach the „juniors“, we have a responsibility to know first hand what works and what doesn’t, because they are using LLMs if we like it or not.

I still care about the code Notes from my Thoughtworks colleagues on AI-assisted software delivery

I've seen a surge of discussions recently about large AI-generated change sets that are impossible to review by humans, paired with speculation if we still need to care about the code in the future. I expect to continue to care, especially if I'm on call for it martinfowler.com/articles/exp...

Learnings from two years of using AI tools for software engineering How to think about today’s AI tools, approaches that work well, and concerns about using them for development. Guest post by Birgitta Böckeler, Distinguished Engineer at Thoughtworks

I wrote a guest post on The Pragmatic Engineer newsletter.pragmaticengineer.com/p/two-years-... rounding up 2 years of using AI coding assistants - how they evolved; ways of working; impact I see on speed, quality and team flow; and some thoughts on the future

Comic. [block quote] “Far better an approximate answer to the *right* question, which is often vague, than an *exact* answer to the wrong question, which can always be made precise.” -John W. Tukey, The Future of Data Analysis (1962) [caption] Happy Approximate Birthday to John Tukey, author of my favorite statistics quote, who was born 110.000 years ago sometime this week.

Tukey

xkcd.com/3104/

Your CEO Should Not Be Slacking Your Coding Agent Autonomous coding agents may seem magical, but they generate drag on your team. Allowing stakeholders to directly ask an AI to do work leads to the same old disruptions as directly requesting work fro...

I blurghed some thoughts blog.thepete.net/blog/2025/06...

Starting to realize what makes me nervous about Devin et al. It becomes a new incantation of the perennial problem of letting a stakeholder get work done by shoulder-tapping their favorite young engineer

And I just thought about agents causing that, now @thepete.net is bringing up the possibility of non-coding managers doing it too, with AI‘s help…

In the past few weeks, a new surge of "background" coding agents came out. I wrote down an example of using OpenAI's Codex, this will hopefully help you understand better what they do under the hood, and which agent category they fall into.

martinfowler.com/articles/exp...

When I code with an AI agent, I revert to the last comfortable checkpoint as soon as I feel like losing control. I wonder what that will be like for the delivery managers & POs of the future, what will THEY do when things around them change so fast and erratically that they feel like losing control?

Good scope management is the linchpin of agile software delivery. Generative AI is like a scope chaos monkey, generating more code, more story details, more requirements than necessary

NEW POST

To work effectively with agentic coding assistants, Birgitta Böckeler found she needs to intervene, correct and steer all the time. She describes examples of these interventions indicating the skills we need to correct the tools' missteps

martinfowler.com/articles/exp...

My LLM codegen workflow atm A detailed walkthrough of my current workflow for using LLms to build software, from brainstorming through planning and execution.

Nice write-up by @harper.lol on his AI-assisted coding workflow. I personally prefer in-IDE tools, but the concepts are reusable. ❤️ this:"I really want someone to solve this problem in a way that makes coding with an LLM a multiplayer game. Not a solo hacker experience."

harper.blog/2025/02/16/m...

Exploring Generative AI Notes from my Thoughtworks colleagues on AI-assisted software delivery

I have thoughts and open questions about the role reasoning models might play or not play in coding assistance. A lot of stake is put into how reasoning models are a step change in coding assistance, but I don't see it - yet?

martinfowler.com/articles/exp...

Cursor's "Composer" does it, Codeium released a new editor called "Windsurf", and GitHub Copilot's new "Copilot Edit" feature makes this capability available to many of our clients at Thoughtworks, where Copilot currently remains the most widely adopted tool.

Multi-file editing capabilities have been available in open-source tools like Cline and Aider for some time, but in just the past few weeks some of the major commercial coding assistance products have released that capability.

Exploring Generative AI Notes from my Thoughtworks colleagues on AI-assisted software delivery

New GenAI memo: I wrote down my thoughts and observations about the multi-file editing features that are currently being released in lots of coding assistants: martinfowler.com/articles/exp...

Exploring Generative AI Notes from my Thoughtworks colleagues on AI-assisted software delivery

New "GenAI memo": In this one I explore the potential of AI assistance for tech stack migrations. I describe building an agent that changes the testing framework used in a test. As a side effect you can also gain a better understanding of how AI agents work: martinfowler.com/articles/exp...

YouTube video by GOTO Conferences AI Assistance Beyond Code: What Do We Need to Make it Work? • Birgitta Böckeler • GOTO 2024

My talk at GOTO Amsterdam is now live on YouTube: "AI Assistance Beyond Code: What Do We Need to Make it Work?" youtu.be/8jwiABwGC6c?...

Exploring Generative AI Notes from my Thoughtworks colleagues on AI-assisted software delivery

In my newest "GenAI memo", I explore how today's AI tools can assist with onboarding to existing, potentially messy codebases. I do this by trying to understand and solve an issue in a real life codebase: martinfowler.com/articles/exp...