The market saw “AI + insurance” and reacted. Shares in Mony Group dropped after Insurify launched its new OpenAI powered comparison app. I used to work at Mony Group (MoneySupermarket), so I know this domain well. I installed the app and ran through a car insurance scenario.

The eyepopping valuations of "AI" companies are built on what these models might be capable of "soon". But what can physics tell us about the limits of this technology, and therefore its likely future? codemanship.wordpress.com/2026/02/14/w...

@robbowley.net there's a great rebuttal to Shumer's post: https://davidoks.blog/p/why-im-not-worried-about-ai-job-loss

The market saw “AI + insurance” and reacted. Shares in Mony Group dropped after Insurify launched its new OpenAI powered comparison app. I used to work at Mony Group (MoneySupermarket), so I know this domain well. I installed the app and ran through a car insurance scenario.

From @martinfowler.com's fragments newsletter margaretstorey.com/blog/2026/02...

Interesting study from Harvard Business Review saying that while AI coding tools increase productivity, it makes engineer's jobs harder, not easier, and more prone to burnout.

An article going round currently scaring people confidently saying AI will take all our jobs and he knows because he's an AI expert and techie...

"Code health determines whether AI accelerates delivery or amplifies defects. Large-scale studies show that AI-generated changes fail significantly more often in unhealthy code, with defect risk rising by at least 30%. In the AI era, healthy code is no longer optional."

Another good software engineering history lesson

Strong agree

Highly recommend this podcast with @booch.com. A tour of the history of software engineering, similar to some of my recent articles and posts. Plus, at least he has a positive outlook for the profession

Good write up on Claude Codes new agent teams or "swarms" As described I cant help thinking of the antipattern of silo'd specialists in teams (so much for the full-stack engineer) Local optimisation, coordination overhead, loss of shared intent, etc

I wrote this speculative piece over a year ago. Deflation didn't happen in 2025 but appears to be happening now Otherwise a lot of it holds - asides from software engineering, still no big transformative use cases as yet LLMs didn't hit a wall, have been improving incrementally...

Related, from Benedict Evans on LinkedIn SaaS is in trouble generally though, see @edzitron.com's piece from a year or so back. AI upsell being their big hope www.wheresyoured.at/saaspocalyps...

Financial markets gone into panic, with $ bns wiped off tech stocks after Anthropic launched a legal plugin for Claude Cowork. The concern being it's a SaaS killer. What actually launched is a set of structured prompts and workflows - essentially a “skill”, for anyone familiar with Claude Code

Good guide to get you up to speed with all the bits and bobs that get you the most out of Claude Code

I wrote an article for Ada National College for Digital Skills, who run digital technology apprenticeships in the UK

Learn from history

Will they get a free login to the amazing UK Gov AI Skills Hub?

They did not really make sense previously either. The most effective team size I have found work is usually 3 to 5 people, well supported with high automation - CD, automated tests, etc. Larger teams are often a symptom of missing those capabilities rather than a genuine need for more people.

Further thought, thinking about patterns. They appeal to the automatic, emotive side of our brain (as per Khaneman) not the rational part. People are super excited about Moltbot, but probably for all the wrong reasons

After a nice break from it, it appears the grifters are back in my LinkedIn feed. Oh look, a big claim a big claim/prediction on GenAI Check profile = SEO/Marketing manager

Moltbot - LLMs are very good at role-play - humans are very good at projecting meaning onto patterns It is fun, in the same way Conways Game Of Life is fun

A new study from Anthropic finds that gains in coding efficiency when relying on AI assistance did did not meet statistical significance; AI use noticeably degraded programmers’ understanding of what they were doing. Incredible.

GenAI feels like another turning point for software development. It’s really just the latest moment in a long, repeating pattern of partial revelation and broad avoidance of how creating software needs to be approached. blog.robbowley.net/2026/01/30/s...

GenAI feels like another turning point for software development. It’s really just the latest moment in a long, repeating pattern of partial revelation and broad avoidance of how creating software needs to be approached. blog.robbowley.net/2026/01/30/s...

I think I might take one of these "under 20 minutes" AI skills courses the UK Govt. seems very keen on everyone doing and live-post it here... Maybe we can all learn something together! The press release sends me to aiskillshub.org.uk/aiskillsboost/ - let's go and see!

Hang on, I had missed the fact that the AI Skills Hub cost £4m. mahadk.com/posts/ai-ski... If this wasn't tax payers' money it would be quite funny. Also, can only assume PWC spent some time rolling around in gold because they did NOT spend it on building this

I've written a piece for the @econ-observatory.bsky.social on the history of GDS. Notably, why you need change both top down and bottom up incentives if you're aiming to help a large, federated organisation adopt radical change.

Joined @johncrickett.bsky.social's Coding Chats podcast, if you like podcasts Youtube: www.youtube.com/watch?v=z9Rr... Spotify: lnkd.in/eNJw955r Apple Podcasts: lnkd.in/epZnMjZz Overcast: lnkd.in/eFNPGVvS

Spec driven development (a la waterfall in disguise, yet again) is probably going to come and go quicker than the hype cycle that convinced people it was new

I have Gas Town derangement syndrome and spent the last few weeks writing thousands of words on agent orchestration patterns; how they shift our bottlenecks and force us to ask whether and when we should stop looking at code maggieappleton.com/gastown

Apparently the instructions were missing (README) and it does compile. simonwillison.net/2026/Jan/19/...

Yikes embedding-shapes.github.io/cursor-impli...

I'll compact you in a minute 😠

If Eli Goldratt was alive today, I'm pretty sure he'd tell you that the rate at which you're able to generate code matters IF AND ONLY IF that rate is: - lower than the rate at which your team is able to review code AND

TIL the 1960s software engineering crisis. It all looks very familiar en.wikipedia.org/wiki/Softwar...

On this week's episode, @digiglean.bsky.social and I talk performance topics, spec-driven dev, and more. With content from: @misrasaurabh1.bsky.social @mkennedy.codes @di.github.io @andrewnez.bsky.social @francoisz.bsky.social @robbowley.net @free.law and others not on the 🦋

It's never been a more interesting time to be a software engineer. Excitement, fear, grief and possibility all tangled together. But the constant evolution of the discipline is, for me at least, what has always made it facinating.

Software engineering is a hot mess right now. We'll get back to you when we've figured it all out, which may take some time. In the meantime, be wary of people confidently proclaiming they've figured it all out, they haven't. Looks like we're going to have to learn the hard way.

Funny getting on a flight which has priority boarding only for everyone to get on the same cramped bus

I've been writing The Phoenix Architecture for a few weeks now. 13 articles published so far on strategies for safe, effective software development with generative AI. Follow along here: aicoding.leaflet.pub

Rich Hickey, creator of Clojure 👏 gist.github.com/richhickey/e...

Work can take you to some interesting places!

I've been using Claude Code for a lot more than just coding (thanks to @chrismdp.com) - a good guide for non-devs for how to get set up. hannahstulberg.substack.com/p/claude-cod...

"The hardest single part of building a software system is deciding precisely what to build. [...] No other part of the work so cripples the resulting system if done wrong. No other part is more difficult to rectify later." Fred Brooks, "No Silver Bullet"

One thing in this I do agree on - the relevance of CS/SE degrees Already many decades behind modern technologies & practices Ofc learning fundamentals very important, more so even but only a small part of the gig Like many, I never did a CS degree & self-teaching now vs late 90s...

The person who wrote the "AI 2027" piece that got everyone in a tizzle last year, predicting the arrival of AGI by then, is now rowing back on that timeline. It is now, of course, at least 3 years away (as it always is)