WorkOne company and one website. The company is the one that matters, and most of what it will be does not exist yet.
Fusion Development GroupBuildingI read Nick Bostrom’s Superintelligence and could not get past what it implied about timing. Not the far scenarios — the near ones. The argument that made it land was not about machines waking up. It was about how quickly a capability stops being scarce once it stops being hard.My speciality is software development and frontend design. It is what I am good at, and what people have paid me for, and it is plainly one of the things a machine is about to do well. The instinct when your craft starts commoditising is to defend it — to specialise harder, to insist on the parts a model cannot reach yet. I would rather hand it over.What actually stops people Not talent. I think nearly everyone has more capability than their circumstances let them spend. What is scarce is leverage on your own time — and good design was expensive, gatekept, and taught badly, so the distance between having a vision and shipping it stayed wide for anyone without money or a decade.AI collapses part of that distance. Fusion is my attempt at the rest. Helix is a design system, which sounds like a narrow thing to care about, and it is the same question I was writing tutorials about in 2020: how does a person put what is in their head onto the internet without asking permission or raising a budget.Individuals first, then businesses. After that are the verticals I have committed to and not started — energy, capital, defense, civic, living. Where it actually goes may change, and I would rather say that than draw a roadmap I cannot honour.What I am confident of is that every one of those is about to meet the same headwind, and that things built to be left will be worth having when it arrives. Studio is the only vertical shipping today. Services pay for the rest, which is the unglamorous arrangement underneath everything above.What it has cost, and what I do not know Time and energy, which is the boring answer and the true one. The constraint is expensive in a specific way: it rules out the business model everyone recommends, and it means shipping the part that makes us least necessary before the part that would make us rich.The known unknown is AI itself — whether what it builds is closer to a utopia or a dystopia. I do not know, and I distrust anyone who says they do. I am building for the first one. If I am wrong about which is coming, I would still rather have spent the time making people harder to trap. rosssheppard.comBuildingIf Helix is meant to make design something a person can simply have, this is the first honest test of whether it does. It is the first consumer of Helix outside the monorepo that produced it — installed from the published packages, with no special access and no shortcuts back to the source.The palette is four seeds run through the theme engine rather than a set of hex values I liked, so every colour on the page derives from one input and stays in contrast when it moves. That is the part I would have paid a designer for, and it took four lines.It is also legible without a rendering engine. Every page has a markdown twin generated from the same source the page renders from, never scraped from the output — because a site built out of web components serves the words without the structure, and the structure is most of the meaning./work → /work.md
/writing → /writing.mdA site whose meaning is legible only through a rendering engine is itself a small dependency nobody can exit, which makes this the one place the constraint applies to my own weekend.