Architecture

A few weeks back I saw someone type /explodedview /xray into an image prompt and get back a Flutter app pulled apart like a Haynes repair manual. My first thought was that I had missed a product launch. I had not. These are not official commands. No AI tool ships a /blueprint button, and there is no secret prompt language. They are visual shorthand. Words like exploded view, cutaway and isometric already name well-known drawing styles, and putting a slash in front simply makes them short, memorable and easy to stack. The model reads them as plain English. The slash is for you, not for it. ...

There is a new comment doing the rounds in review meetings, Slack threads and document margins. It is rarely written this bluntly, but it means the same thing every time. “This looks AI-generated.” Notice it is not “this is wrong”. It is not even “I disagree”. It is a comment about where the work came from, and somehow it lands harder than any technical objection. Once it is said, the document is no longer being read. It is being suspected. ...

Three weeks. That is all it took. In early September Meta launched Muse, a personal agent that runs on its own little Linux machine in the cloud. Then on 29 September, at DevDay, OpenAI announced Dots, “always-on agents” that also get their own cloud computer and keep working after you close the chat. I read both announcements on my phone, standing in the kitchen, and my first thought was not “wow”. My first thought was: finally, somebody to use all that plumbing. ...

You have used ChatGPT, Claude or Gemini. You know roughly how an LLM works. And this week your feed has been full of something called Jev, with people calling it 200 times faster than an LLM, “the end of chatbots”, or just “a fancy classifier”, depending on who you follow. So you searched. Welcome, sit down, I will try to be the explanation I was looking for last week. The short answer first. Jev is not a better LLM and it is not a replacement for one. It is a different kind of model that does one job, making structured decisions inside software, and does it very fast and very cheap. It cannot write a sentence. The most useful way to think about it is as the fast half of a system in which the LLM remains the slow, thoughtful half. ...

A recruiter, a candidate and a large language model join a Teams call. Nobody on the call is a person, and the call still starts eight minutes late. That is the joke. Here is the news. Companies have started putting AI in the first round of hiring. Not just the CV filter, which has been a machine for years, but the actual interview. An avatar with a name and a haircut asks you to tell it a story about a time you showed initiative, and something behind it scores your answer. Semafor and The Next Web have both covered the next turn of the screw, which is that candidates are now sending avatars of their own to sit those interviews, and recruiters are spending five minutes working out that the very polished person on screen is not, strictly speaking, a person. ...

The MCP 2026-07-28 release candidate drops sessions, promotes extensions, and deprecates three core features. Here is the architectural read, the impact on existing implementations, and what tech leaders should plan for.

उलझे नहीं तो कैसे सुलझोगे, बिगड़े नहीं तो कैसे संभलोगे If you never get tangled, how will you learn to untangle? If things never go wrong, how will you learn to set them right? I was part of three AI trials inside a New Zealand government agency. All three gave us invaluable lessons. Out of those three, one progressed quite a lot, through four stages of development, and has shown genuine promise. ...