The one thing you can't fake
Everything else in a studio's pitch is asserted. The case studies are edited. The testimonials are curated. “We sweat the details” is a sentence anyone can type. The one artefact a prospective client can inspect all the way to the bottom — view the source, resize the window, click the thing in the corner, read the empty space — is the studio's own website. It's the only work we ship where the client is also the QA.
So we treated it as the hardest brief we'd take this year. Not the most expensive, not the most visible — the hardest, in the sense that there was nobody to hand the awkward decisions to. When you build your own front door, every shortcut is a shortcut you chose, and every shortcut shows. Here are a few of the decisions, and why they went the way they did.
No template
We didn't start hand-written. The first version of this site was built in Webflow — visual, fast, nothing to touch in code. For a generic studio site, that's a perfectly good tool, and for a while it was fine.
It stopped being fine once the site stopped being generic. Every specific decision — a notan split down the middle of the homepage, a clock you can click to change timezone, a section that's deliberately almost empty — meant fighting the visual builder's own model of what a page is, then bolting on custom code to route around it anyway. We were paying for Webflow's opinions and overriding them at the same time, which is the worst of both. Past a certain point it stopped making any sense: more hours spent fighting the platform than shipping the thing on it.
So we moved it over and hand-wrote the HTML instead. Every page is a single file: vanilla, no framework, no CMS, no build step. Not out of nostalgia, and not to flex. The reason is narrow — the tool you reach for is a tell. Reach for a template and you inherit its opinions: its spacing, its rhythm, its idea of what a section is. Then you spend your effort either fighting those opinions or, more often, quietly surrendering to them. A studio that sells judgement can't ship a site whose every proportion was decided by someone at a SaaS company optimising for the average customer. The average is the one thing we're not.
The cost of hand-building is that there's no default to blame. Every margin is a number one of us typed. That's the point. It's the difference between a place that was assembled and a place that was decided.
The empty space is on purpose
The first honest critique of the early drafts was that the site felt tight — tidy, uniform, everything the same distance from everything else. Tidy is what you get when nothing has actually been decided. You space things evenly and move on.
So we went back and broke the evenness on purpose. The section headers sit off-centre and alternate which side they lean on. The vertical rhythm ruptures — some sections breathe, some compress, because not everything on a page deserves the same weight. And there's a stretch on the homepage that is, deliberately, almost nothing: a plumb line, a single coordinate, and a lot of held emptiness. People ask whether something failed to load. Nothing failed to load. That emptiness is the most decided thing on the page. Room you leave on purpose is a claim — this mattered enough to protect — and it's a claim you can't make by accident, or fake having made.
Real screens, not mockups
The easy version of a case study is a beautiful device mockup with a tasteful gradient behind it. It looks great and it proves nothing, because the screen inside the frame is usually a render — the product as it ought to look, in a state it's rarely in.
We wanted the opposite: the actual app, in a real, populated state, with real data on it. For our own strength-coaching product that meant building a small rig to seed a plausible history — a real programme, a week of logged sessions, actual personal records — and then capturing the screens the way someone would genuinely meet them, mid-use, mid-thought. It was more work than dropping a mockup into a frame. It's also the entire difference between showing what a product looks like and showing that it works. One of those is a claim; the other is evidence. A site that sells judgement should trade in evidence.
The clock
In the corner of every page there's a small clock. Click it and it steps through the timezones we work across. It changes nothing else. It isn't a feature. Nobody asked for it.
It's there because the small, useless, cared-for details are the ones that can't be faked. Anyone can get the big moves roughly right — the hero, the headline, the grid. The tell of whether a place was actually sweated is in the parts nobody would have complained about if they were missing. A clock that could have just told the time, and instead does one quiet, deliberate thing when you touch it. If we couldn't be bothered with that, why would you believe us about the parts that are hard to see?
Why bother
Because the site is the portfolio. Not the thing that points at the work — the thing that is the work, the one engagement where a client can check every claim against the artefact right in front of them.
The judgement a studio sells is either visible in its own front door or it isn't real. We'd rather you decided from the evidence than the adjectives. Everything here is a decision we'd defend, made by the people who have to live with it. That's the whole pitch. It's also just the site.
---
cybercyber is a lab for AI products, AI features, and the agent infrastructure that runs them. We work with a small number of clients each year. The ones who can tell the difference. Email hello@cybercyber.ai.