I'm currently offering vibe-friendly discovery packages from £300 half-day sessions for anyone with vibe-code, -ideas, -thoughts, -whatever who wants to take it up a notch and wrap a full, testable, scalable, deployable app around it. I mean let's face it, pretty much any simple idea is a handful of LLM prompts (and light expletives) away these days. But validating it - getting another human to add a layer of accountability, especially - that's something else. Like maybe you've put together a vibe coded prototype - dissecting it, you've probably got a mix of static content, dynamic views, some DB tables, some auth, and so on. What you don't have is a plan based around validated software engineering principles (and YES those still matter) - not just for the future scalability of your app but also for your Claude/Codex/Grok (if you insist) to build on. If you're vibe coding, you know very well how long and painful the road ahead looks with limited knowledge. I don't need to take you the whole way. Let's just get you swearing less at your computer. That's a good first step. DMs open.

This project's also fun for me because I'm using it as a way to challenge myself to build with AI without it having a hideous interface. And you can do so much more with it. Like contextually, how can you build more advanced UI that responds more intelligently to user input, based on e.g. selected mode, angle of view etc what does any given user need to see for any given action. I think this is a massively overlooked part of building with AI. They are atrocious at UI, and it's only adding to visual clutter and noise for people unfortunate enough to have to use the default stuff these things are building. So anyway, here's a little UI for my sunlight-hours visualiser, a way to easily edit a boundary fence - how to do that so it's not fiddly and so the user can very quickly grasp from current context how it works and how to use it.

I'm building out a tool right now for measuring sunlight hours, shadow impact for a property (useful for e.g. gardeners, people viewing homes to move) and I realised Opus 5.5 is REALLY good at trees.

"I one-shotted [GAME NAME HERE] in a weekend! Game dev is cooked" No it is not oh my goodness 1) Game studios are also going to be using these tools. You're competing with other people who understand their potential 2) If you can't see the absolute universe of complexity that's waiting for you beyond these interesting little tech demos you're producing, stop now

In a world where LLM coding was actually a game changer, users would notice. They'd say "wow, has anyone noticed how useful, reliable and fast things are these days? They'd aay it about websites, apps, windows, macOS... They do not. Ergo: the princess is in another castle.

I still can't get over the complete lack of impact agentic coding has had on software from a user's perspective. Our industry looks faintly ridiculous for waxing lyrical about the "agentic era" when it's had zero observable impact to the end user.

A thought: The prediction is that everyone will have personalised software but what happens when there's enough personalised software to cover the majority of niches and AI just recommends what someone else has already built instead of doing it for you? Maybe there could be some revolutionary licensing model where the Source is Open and then the AI could get you to suggest contributions to it.

the moat is literally just caring about your work and being proud of your output and holy shit it does not seem to apply to 99% of people

Now is the time for calm minds who understand AI and are immune to the hype

Ok, so Twitter has now become more unbearable than ever (quite an achievement) especially r.e. "software development is dead lol", and I think I'm going to stop engaging with it entirely. Summarising my thoughts on it, then I'm done - if you've got an opinion on it and are dunking on the profession for clout: 1) Shut the fk up, no one who matters cares 2) If you genuinely want to talk about it and engage in interesting discourse about it (because it IS fascinating, it's a massive paradigm shift and no-one really knows where this is going) seriously shoot me a DM or an email or whatever and I'd love to share some opinions But yeah this is not the subject to be seeking engagement bait on, it makes you look like a dumbass

Is all this “software is dead” “UX is dead” engagement noise considering that people won’t want their agents to do things for them. They’ll want to do things *with* them?

More backrooms dev. I chose this as a game project because I wanted a world I could sink a lot of time and effort into instead of jumping between ideas. Rule: if I have a new idea about something I (and Claude) want to build, it has to be embeddable within this world. From above, I want this environment to feel real and tangible. But as an overseer of a world, what do you see? I've build a "cutout" system so that all walls, when viewed from above and when "cutaway" (when player is behind them) show off their internal structure. Here, the "abandoned yellow hallways" architecture displays wooden joists and supports behind the plasterboard. Swapping to first-person view you can then see what that looks like up close. How does that play out as actual game mechanics? idk haha, I'm just vibing this right now. Does that sound like something conducive to a playable experience? No. It is not. Plan is eventually to make this more of a walking simulator, a maze to get lost in, an excuse to explore and examine.

The question isn't whether coding will disappear (yes, sadly) it's whether complexity will vanish too (no lol that's just getting worse)

Everyone's out there apparently one-shotting colourful interactive playgrounds with Claude Me: "more existential dread please claude. make the wallpaper greasier" I'm wandering the infinite isometric backrooms again today.

I'm considering branching out and focusing more on automation-infrastructure services in terms of what I offer professionaly vs traditional software development. Then I see services like Meta's Muse offering to provide on-demand integrations between systems, so is there much scope for this? I mean they've got all bases covered, from "help me read email plz" or "want book flights" to "help me read more emails please" So I think I'll be fine. Anyway. DMs open if you have a business-facing automation problem that's more complex than "help plz read emails"

Truths