I actually love @habitkit by @SebastianRoehl the UI is truly beautiful. This year for me, there has been a lot of focus on health and gaining control of my weight gain. I'm now 17kg down and a big part of that is tracking my habits and getting better 1% daily. One of the tasks I have is to log weight daily, and I wanted an app as beautiful as HabitKit to track my weight so I built WeightKit for myself. I have the app working on ios and android but won't release without the blessing of @SebastianRoehl. Did I cook? Do I have permission? I think functionality is different enough, but I genuinely feel the need to work on my own unique brand and launch that way instead! Massive shout out to @SebastianRoehl once again for the inspiration in amazing design. Until either blessing or I figure out branding, its exclusively for me only. 😎

State of linked in. Sigh. I never posted a post about losing an AI Engineer to export controls..... this noise is annoying.

The shiny thing and the soul of it, Is design dead? A Follow up on yesterdays post. It is a question I have been asked more than once this month. On the surface, the case looks strong. AI tools can now produce wireframes, slide decks and entire brand identities in minutes. Image and video models turn out campaign creative on demand, and there are even models that generate clean vector graphics and SVGs. If a machine can make it look good, the argument goes, what exactly is the designer for? I dont think design is dead. But I do think something important is being lost, and I want to explain what. Hundreds of apps cross my desk. Many of them have a genuinely good premise, and a surprising number of them look perfectly fine at first glance. Then you start to use them, and the cracks appear almost immediately. Buttons sit where nobody would look for them. Flows that should take three steps take seven. Screens are full of things that are there because they could be, not because anyone needed them. What is missing isnt polish. It is thinking. The understanding of how people actually behave in front of a screen, the discipline of human-computer interaction, the craft of shaping an experience before a single pixel is placed. That work used to be done by designers, and it still should be. A good, experienced designer does something no prompt can quite replicate. They tell the story of why a design should exist. They can explain why this button, why this spacing, why this order of things, and why the obvious alternative is wrong. For the money, they will often take a product further than any amount of weekend building. As a developer, I feel their absence most sharply at the handover. The best designers I have worked with gave me more than a file. They gave me the reasoning behind it, the edge cases, the nuances, the small decisions that make the difference between something that works and something people love. When that handover is done well, it is close to flawless. When it is missing, you are left guessing, and guessing shows. Over the past two decades I have had the pleasure of working alongside and seeing some some brilliant designers action, people like Tom Good, Resh Sidhu and Mehmud Ali, to name just a few. What they share is not a style. It is judgement. To be clear, AI has made design faster, and that is a good thing. Iteration that once took weeks can now take an afternoon. My worry is about control. When anyone can produce something shiny in a weekend, the designers voice is easily sidelined, right up until someone has to use the thing. The current crop of AI-built experiences is often good enough. But good enough is not the same as good, and much of it is missing something harder to name. Call it soul. Design isnt dead. It just needs designers to stay in the room. Love to hear from my network how they feel design is evolving, especially from those in the space!

And we don't stop.

What...

Damn I just swapped out for Gemini 😭😂😂😂😱

The weekend app and the forever bill! Last week I found myself around a table with friends and colleagues, and the conversation turned, as it so often does now, to whether SaaS is dead. The verdict came quickly and was almost unanimous. AI is here, they said, and anyone can now build the software they need, so why keep paying for someone else's? I stayed quiet. That wasn't because I agreed. I didn't, and I wanted to think it through properly before saying so. These are those thoughts. The argument, as I understood it, goes something like this. Why should we keep paying for software when we can simply build our own over a weekend? It is a fair question, and the honest answer is that yes, you probably can. The tools are good now. You describe what you want, nudge the AI along, and by Sunday evening you may well have something that looks and behaves like the product you have been paying for every month. But building software was never the expensive part. Keeping it alive is. Ask anyone who has run a piece of software in the real world. The costs arrive quietly, and they never stop. A security flaw nobody noticed. An integration that breaks because another company changed its API on a Tuesday. A bug that only appears for users in a certain browser, in a certain country, at the end of the month. The colleague who built the thing moves on, and suddenly nobody quite understands how it works. That is the job. It is unglamorous, relentless and expensive, and it is precisely what a subscription pays for. When you buy good software, you are not really buying code. You are paying for a team whose entire working life is devoted to keeping that code working, improving it and keeping you safe. That leaves your people free to get on with the work you actually hired them to do. Vibe coding does not make that bill disappear. It simply moves it onto your desk, and it tends to arrive long after the excitement of the weekend has faded. None of this lets the software industry off the hook, however. Something real has changed. When a competent person can knock up a passable alternative in a few days, passable is no longer a business. The floor has risen, and companies that coasted on being merely adequate should be worried. The ones that survive will be those that are unmistakably better than anything a customer could build for themselves, and that improve at a pace customers can feel. The moat is no longer how hard the software is to write. It is the depth, the reliability and the speed of everything that happens afterwards. So SaaS is not dead. The mediocre version of it may be. And for those of us who buy software, that is rather good news.

Shout out to everyone in my x-circle!

I've updated mersive.so to use voice generated by Googles 3.8 Flash TTS models and the results are incredible!

Don't sleep on 3.8 flash tts. It's a beast.

OMG Google Cooked. Testing google text to voice generation via their apis. Comparing what my app originally from ElevenLabs to what Google produced today. OMGGGGGG!!!! 🔥🔥🔥 Listen below! Sound on! 👇

Final moat left for development agents, beyond just MCP. Is this a 100 million dollar idea??? Anyone working on agent registrations and logins for platforms, I know people are looking at agentic payments? I just want to let the ai go sign up for what it needs, setup the platforms for what it needs, and crack on with 0 or near minimal interaction from myside. Imagine agent registering its own email, registering to gcp, setting up and provisioning dev, stagin prod servers, buying the domain, and connecting all together, and just serving the final result.... AUTOMATE THIS NOW!!!!

Anyone else feel like this after Opus 5.5 released?

What time to be alive.

Claude Opus 5.5 x Higgsfield. First attempt with a vauge prompt. Script can be improved but crazy how good this is with first attempt. Sign up at mersive.so to keep up to date!