Business interests starting to become visible across AI IDEs?Anthropic made Sonnet 4 available on launch (today) for Copilot and Cursor. But not for Windsurf!I will speculate this could be b/c of reporting that OpenAI (Anthropic’s rival) in the process of buying WindsurfVarun Mohan: Unfortunately, Anthropic did not provide our users direct access to Claude Sonnet 4 and Opus 4 on day one. We are actively working to find capacity elsewhere so we can continue to provide the most versatile and powerful AI assistance platform, period. That is our only focus.To

Common acronyms around AI I see a lot and what they mean:LLM - large language modelSOTA - state of the artRL - reinforcement learningRLHF - reinforcement learning from human feedbackNLP - natural language processingFLOPs - floating point operations

Re I mean WOWIt’s pretty much what GitHub Copilot aims to be! With the agentic addonsThe dev is still in charge (akin to the Chief Programmer in Brooks’ model)

There’s this saying of “never meet your heroes.” It most definitely does not apply to @shanselman who is as great of a human as you’d imagine he is.His article from 10+ years ago “your words are wasted” helped me back then and also now https://www.hanselman.com/blog/your-words-are-wasted

Every AI coding tools company doing demos with React + Typescript (and maybe React Native.)Microsoft meanwhile showing off Java and .NET case studies.This doesn’t look like an accident - but a very deliberate strategy.It’s easier to win when you have no competitors…

At one of my first jobs as a dev I had to build apps using Silverlight. It was a new technology with not that much written about it and one of the best blogs I found was written by @jeffwilcox. Same during my Windows Phone dev days.Very cool to meet in-person a ~decade later!Jeff Wilcox: Best part of Build so far: meeting @GergelyOrosz in person! So much shared Windows Phone and Microsoft development stack history. I’ve been a subscriber to The Pragmatic Engineer and appreciate the curation and perspectives. https://www.pragmaticengineer.com

I was today years old when I learned that Electron was created by...... the GitHub team(It was originally built for Atom)Back in 2013Talk about impact!!

If you work in a tech company with screens in meeting rooms: what is on those screens as screensavers?At Stripe: its realtime business metric graphs. Support satisfaction, fraud rate trends, revenue for product X.Just saw these - what a way to “expose” the business to devs!

Hanging out with @ALEngineeredWhat a cool space - and the amount of Amazon patent puzzle pieces really surpised me - 13 and counting!

This aged well.Web development is still web development.The marketing ploy to try and rename web development to “web2” (in order to make ‘web3’ somehow look superior when it’s just web dev + blockhain) never spread widely. Thankfully!Gergely Orosz: Can we please stop calling web development “web2”.It’s “web”.There is no “web3”. It’s embarrassing to associate unregulated securities with a splash of code with this term.

RT Giuseppe De GiorgiBuilding, launching, and scaling ChatGPT Imagesby @GergelyOrosz https://newsletter.pragmaticengineer.com/p/chatgpt-images?r=tf12u&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false

Talking to an energy expert, I’ve learned how a big difference between Europe and US electric utilities is how in Europe you can almost always choose your electricity provider (the utility company) while in the US it’s almost always based on where you live: no choice.This encourages more competition across energy providers in Europe - as switching for customers *is* possible.

Kubernetes is the second-largest open-source project in the world. What does it do—and why is it so widely adopted?Kat Cosgrove has led several Kubernetes releases and joined The Pragmatic Engineer podcast to share the history of Kubernetes, how it's built, when it is a good (or not so good) use case, and how you can contribute to it (and why it's a good thing to do for professional and career growth).Brought to you by•⁠ @WorkOS — The modern identity platform for B2B SaaS https://workos.com•⁠ @modal_labs — The cloud platform for building AI applications https://modal.com/pragmatic•⁠ @GetCortexApp — Your Portal to Engineering Excellence http://cortex.io/pragmaticWatch or listen:• YouTube: https://youtu.be/vBjonut1JMk• Spotify: https://open.spotify.com/episode/2zbYpj9hYwJtzGoYJGvdXN• Apple: https://podcasts.apple.com/us/podcast/how-kubernetes-is-built-with-kat-cosgrove/id1769051199?i=1000708482879• Summary and transcript: https://newsletter.pragmaticengineer.com/p/how-kubernetes-is-built-with-katThree of my takeaways from this great conversation:𝟭. 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 𝗵𝗲𝗹𝗽𝘀 𝘄𝗶𝘁𝗵 𝗺𝗮𝗻𝗮𝗴𝗶𝗻𝗴 𝗹𝗮𝗿𝗴𝗲-𝘀𝗰𝗮𝗹𝗲 𝗯𝗮𝗰𝗸𝗲𝗻𝗱 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀. Google originally built an in-house tool to manage the tens of thousands (then hundreds of thousands and later millions) of machines: this internal tool is called Borg. The roots of Kubernetes come from Borg: but Google has since donated the project to the Cloud Native Computing Foundation (CNCF) – and today, Kubernetes is the second largest open source project, after Linux. We previously did a deepdive on How Linux is built, and touched on Google’s SRE roots in What is Reliability Engineering?𝟮. 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 𝗶𝘀 𝗮 𝘃𝗲𝗿𝘆 𝘄𝗲𝗹𝗹-𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝗱 𝗮𝗻𝗱 𝗼𝗿𝗴𝗮𝗻𝗶𝘇𝗲𝗱 𝗼𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲 𝗽𝗿𝗼𝗷𝗲𝗰𝘁. The structure of the project and all processes are well documented. The project has around 150-200 maintainers, has a few dozen SIGs (Special Interest Groups) and releases run on a 14-16 week cycle.𝟯. 𝗧𝗵𝗲 “𝗹𝗲𝗮𝗱” 𝗮𝗻𝗱 “𝘀𝗵𝗮𝗱𝗼𝘄” 𝗰𝗼𝗻𝗰𝗲𝗽𝘁 𝗶𝘀 𝗮 𝗰𝗹𝗲𝘃𝗲𝗿 𝗼𝗻𝗲, 𝘂𝘁𝗶𝗹𝗶𝘇𝗲𝗱 𝗯𝘆 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀. The Release Team within Kubernetes owns releases, and the Release Team has about 20-30 people participating in each release. More than of the members on the Release Team are “shadows” who get to learn on the job how a release is done – and, hopefully, in a release or two, become leads themselves!

“AI Engineering is just software engineering with AI models thrown into the stack”- @abacaj in the book AI Engineering by @chiproTotally agree - and it’s why learning how to integrate LLMs is such a big win for devs!

Give it time for more people to understand thisGergely Orosz: “AI Engineering is just software engineering with AI models thrown into the stack”- @abacaj in the book AI Engineering by @chiproTotally agree - and it’s why learning how to integrate LLMs is such a big win for devs!

Mass layoffs at Microsoft under Satya Nadella (2014-2025): 3Under Steve Ballmer (2000-2014): 1Under Bill Gates (1975-2000): 0[I do think Steve Ballmer should probably get more credit than he does]

Oh yes I did this as wellI stopped adding “exclusive” to @Pragmatic_Eng articles because every one of them is “exclusive” - when doing deepdives with engineering teams I ask about stuff they’ve not yet talked about, or not in as much depthSimon Willison: Excellent scoop here: Gergely got OpenAI to share their core engineering principles, "Ship relentlessly", "Own the outcome" and "Follow through"Those are unreported anywhere else as far is I can tell...

While I am sympathetic to this idea, but…How have the price of games changed in the last 5 years, since it has become much more accessible for tiny dev teams to create good games and distribute them (thanks to eg Steam?)Did the best games get cheaper? (Same market forces!)Louie Bacaj: There is a real possibility the price of all software drops dramatically over time, because of LLMs.Competition will force this. Even complex SaaS, why pay some big company with large teams, when a small team or small time dev might offer it at a fraction of the cost.

From the outside, it's easy to assume that OpenAI suspects when a launch will be a massive success.From all I learned: this was not the case with ChatGPT Images. The scale of usage was unexpected even internally, to the point of some teams thinking the o11y system had some bug!The Pragmatic Engineer: OpenAI's biggest-ever launch happened on 25 March: releasing ChatGPT Images.100M new users in the first week700M images generated in 7 days1M new users/hour signing up at peekThe eng team told us how exactly they pulled it off. All the details: https://newsletter.pragmaticengineer.com/p/chatgpt-images

Data > vibesThe last months all I saw influencers talk about is how tech + software engineering is dead because of AI after someone found a chart trending downwards If anything, the industry is changing. PMs who know how to build AI products are in-demand. Seeing same w devsLenny Rachitsky: State of the product job market in 2025:1. The highest number of open PM and engineering roles in over 2.5 years (!!!)2. AI PM roles (and AI-related roles in general) are exploding3. Layoffs are slowing4. Capital investment is increasing5. The Bay Area continues to be the

My habits of finding a restaurant are changing quickly.Used to be places like Google Maps, or review sites.Now I go to Perplexity and ask "what are restaurants suitable for {even type} with {number of people} near {location of choice}" Works very well IME

What have I done 😅Gergely Orosz: A dev I talked to said his favorite tool for development recently is called:(I was expecting some AI IDE)(Drumroll… it wasn’t)Tuple!It’s a lair programming tool that allows “taking over” the other person’s screen/IDE. Apparently his whole team uses it daily & love it

Louder for people in the backI’m sick and tired of half-baked AI features released that make products worse and cannot be turned offGemini for Google Docs / Sheets is one of many examples (this reminds me daily of what a feature that should have not been released looks like)DHH: AI is awesome, but do you know what else is awesome? Not releasing AI-powered features until you've actually built something that's way better than what it was without. Not every feature you build or explore has to ship! (Apple used to know this).

A suggestion to everyone doing a desk job:Get a smart watch or band or step counter.Observe your mood on days based on how many steps you do per day.I never believed in this “10K steps per day” stuff until… I observed what happens when I walk about this much per dayPeter Steinberger: My gym coach pushes me to get at least 10.000 steps in daily, which forces me to go on walks, and that has been so good for mental clarity and happiness.5/5 do recommend.

I guess the cat’s out of the bag nowEpisode on how Windsurf is built coming on Wednesday in @Pragmatic_EngVarun Mohan: Big announcement tomorrow!

Somewhat ironic how I brought back this to Europe from the US… because it is not available for purchase in Europe (it’s just made there!)

Every monthly SaaS bill I have: X / month (where X is $10 or more)AWS monthly bills for a few prod services I run:I cannot express how amazing this is - and to this date *still* unconventional (that AWS does such granular billing - low usage means vey low cost!)

RT Gergely OroszJohn Ousterhout, the author of A Philosophy of Software Design (and a professor at Stanford, creator of the Tcl programming language), is currently busy with... contributing to the Linux kernel. He is adding support for the Homa Transport Protocol. What. A. Legend.

Re THIS is a great question to ask!And we will see the answer to it soon enough (perhaps even now)staysaasy: @GergelyOrosz The “write more code ship more bugs” POV is interesting with AI. AI lowers the cost of writing code dramatically. So you might prioritize a lot more or a lot less aggressively. But bugs still hurt your customers the same. I wonder if that mismatch will cause issues.

RT Giuseppe De GiorgiRobotics for software engineers: humanoid robotsby @GergelyOrosz https://newsletter.pragmaticengineer.com/p/humanoid-robots?r=tf12u&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false