Martin

Comic of Tintin talking to Captain Haddock, who looks exhausted. Haddock says ‘What a week, huh?’ and Tintin replies, ‘Captain, it’s only a major AWS outage on a Monday.’

How’s your week going?

The new BIBE environments is a simple shift which has made a big difference for me this last month: No big overhaul — just one tiny adjustment to my workflow with great results. Learn more here: lnkd.in/dDCRcS8f

This new project from Dan Hollick makes me want to share with you all the content I've been gathering as great "teaching".

These usually involve visuals and tailored interactive components to help explain complex topics.

#1 makingsoftware.co by Dan Hollick

Your 20s: "Commitment, effort, involvement"
Your 30s: "I'm delivering this shit as it is and going straight to the gym"
#LifeGoals

Good point @jeremias.codes! I was assuming teams that already know what debt to tackle but can't prioritize it over features—the classic "we'll fix it next sprint" that never happens. 😅

However, AI development creates new technical debt too. Especially when using multiple agents for feature development, they tend to solve problems through duplication rather than integration—generating similar solutions independently instead of reusing existing patterns, or creating new abstractions

Managing technical debt is fundamentally changing: AI agents are quite good at understanding legacy code and executing directed refactoring faster than humans, making previously impossible debt reduction viable in resource-constrained teams.

Have you ever wanted to work 4 days a week with full pay? Fully remote. Fully transparent (even salaries), values-driven company? This role pays between 141K-163K USD. You can work from anywhere in the world. We're looking for someone to join our Marketing Team at Buffer. Read more buff.ly/gU6JHk7

Most people misunderstand how AI Agents are designed to assist us in our everyday activities. There's a common fear that AI will take over human jobs.

You don't always have to review GenAI code - it could be the other way around!

Ask AI to review your own code instead. Use Copilot as a reviewer on GitHub, or ask Claude Code to review your PR content. It can give you really good insights and catch those small, subtle bugs you could easily miss.

It happened a similar thing to me with Airmail, but I found the experience with Mimestream way superior. Without super fancy features or AI everywhere. It just works.

When I worked as a sysadmin 10+ years ago, Mutt was my go-to email client. After moving away from it, I never felt comfortable with anything else... until I found Mimestream 2 years ago. Been using it since beta and it's simply the best: elegant, fast, and feels like it truly belongs on the Mac. 🙌

Due to the Coriolis effect, loading spinners in the Southern hemisphere spin the other way.

resolution - CSS: Cascading Style Sheets | MDN The resolution CSS media feature can be used to test the pixel density of the output device.

Did you know there are resolution media queries?

developer.mozilla.org/en-US/docs/W...

#CSS

This is my favourite performance-enhancing JS feature ⚡

It allows ES modules to achieve laziness similar to inline require() in CJS modules, and adds safe handling of Top-Level Await 👍

It's an easy way to speed up existing codebases without introducing the async virality of dynamic import() 🦠

I've been recommending this pattern to my teammates. It's been helpful for those tedious migration tasks.
Curious if others have found similar patterns when working with AI tools?

It wasn't perfect (multiple bash scripts instead of a unified codemod), but is fascinating that it came up with the pattern by itself.

One interesting observation, while testing Claude Code with a similar migration task, it actually switched strategies mid-way through. After making a few individual changes, it started building migration scripts on its own.

The key insight: for some tasks, asking AI to build tools that help you is often better than asking it to do the task directly.

The remaining 5% were edge cases and some legacy legacy tooltip variants I wasn't aware of. Still, much more efficient than manual updates or having AI attempt each change individually.

This is something I've already used at @buffer.com, we were migrating tooltip components from our old design system to a new one. Rather than having AI update each instance, I showed it one manual migration and had it write a codemod.

~95% of components migrated perfectly in one go.

Instead of asking AI to perform the migration directly, I do this:
- Manually migrate one instance first
- Get the diff of my changes
- Share the old/new component/function signatures with AI
- Ask it to build a codemod that automates the rest

🧵 Want to share a technique I've been using with generative AI for coding that's been really helpful

When you ask AI to make large changes across a codebase (like migrations), it often struggles - uses lots of tokens, makes mistakes, takes forever. But there's a better approach

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...

It’s not new but I just discovered this @theoatmeal.bsky.social post and I think it may well be my favorite of his.

“Sometimes going down the wrong path isn’t a mistake — it’s a construction line”

Amen to that!

theoatmeal.com/comics/creat...

I want my code to be understandable to idiots with limited context, because that's also future-me.

Line chart showing Palabreja.com revenue growth from September 2024 to May 2025. The blue line starts low and trends upward over 9 months, with some fluctuations including a notable dip in January 2025, but overall demonstrates consistent growth with the highest peak in April 2025.

9 months of Palabreja.com growth captured in one chart. The ups and downs tell the real story, but that upward trend? That's what happens when you keep showing up and iterating. Slow and steady feels pretty good.

$22,384,596 may seem like a random number, but it's oh so meaningful for us at @buffer. This number marks the transition from ARR recovery, which is essentially a continual look backward at our previous all-time high, to a bright and beautiful launch forward into *everything* that is to come.

async await async await async await async await

in the function, the mighty function, the child thread sleeps tonight 🎵

Wow, we did it! We have officially achieved a new all-time-high ARR at Buffer 🎉

Our previous all-time-high was $22,384,598 on February 6, 2020. Today, our ARR stands at $22,395,427.

This milestone has been 5 years in the making for us, it's going to take a little time to sink in for me personally!