A small lab with a deep bench of agents.

cyber:cyber is run by Dan Andrews, a product and ux designer with more than a decade of experience across agencies, consultancies, FTSE 100 companies, NYSE-listed businesses, and several startups — including time as employee number eight at a company building from zero.That breadth matters. Most senior designers have seen two or three of those environments. Seeing all of them produces a particular kind of judgement: about when work needs time and when it needs to ship, when a process is helping and when it's a substitute for thinking, when a brief is real and when it's a politically-shaped artefact of someone else's calendar.That judgement is what cybercyber sells. The form it takes — agents, the products they live inside, the AI features that ship inside existing products, the design systems that hold it all together — is downstream of the judgement, not upstream of it.

cybercyber operates on two sides simultaneously.

On one side: client engagements. A small number per year. Full builds, AI features, operating systems, design systems, and weeks of senior design support — all calibrated to ship work that's still in your customers' hands six months later. Pricing is transparent and non-negotiable. We say no to engagements that wouldn't produce work worth keeping.

On the other side: our own AI agents and the products that run on them. cybercyber is operator-led, which means we build our own agent infrastructure in the same stack we install for clients. Some of these will ship as standalone businesses. Most won't, because the discipline of refusal applies to our own work as much as to client work. The point isn't that all our products succeed. The point is that the judgement we sell to clients stays sharp because we're operating on it daily.The two sides compound. Patterns we learn from our own builds inform client work. Things we discover in client work inform our own products. This is the structural argument for why a lab — and not an agency or a studio — is the right shape for what we're trying to do.

The strapline on the homepage is unambiguous: Not a fucking studio. Some context for what that means, and what it doesn't.Studios and agencies do good work. Some of the best work being shipped in 2026 comes from small studios with strong opinions, and from agencies that have built reputations on craft over decades. The world would be poorer without them. cybercyber is not in the business of denigrating either model.But the strapline is structural, not territorial. cybercyber is not a studio because it doesn't operate like one. Studios sell hours and aesthetic point-of-view. Agencies sell throughput and capacity. Labs sell judgement and output — engagement-priced, fixed-scope, calibrated to make the right calls about what's worth building and when to ship.The strapline exists to tell visitors that in three seconds rather than three meetings. It's a disqualifier, not an insult. If you came here looking for a studio, there are excellent ones, and you should hire them. If you came here looking for the model we just described, keep reading.

cybercyber operates on two sides at once.On one side: client engagements. A small number per year. The five offerings on the homepage cover most of what we do — full builds, AI features, operating systems, taste systems, and weeks of senior design support. Pricing is transparent and non-negotiable. We work with clients who can tell the difference, and we say no to engagements that would compromise the bar.On the other side: our own products and businesses. cybercyber is operator-led, which means we build our own AI products in the same stack we use for clients. Some of these will ship as standalone businesses. Most won't, because the discipline of refusal applies to our own work as much as to the client work. The point isn't that all of our products succeed — the point is that the judgement we sell to clients stays fresh because we're operating on it daily.The two sides compound. Patterns we learn from our own builds inform client work. Things we discover in client work inform our own products. This is the structural argument for why a lab — and not an agency or a studio — is the right shape for what we're trying to do.

A few principles that show up in every engagement:

1. We refuse before we agree. The first conversation often ends with us saying we're not a fit, or that the engagement as scoped wouldn't produce work worth keeping. We're not trying to win every prospect. We're trying to take on the engagements where we can do the work we'd defend.

2.We charge what the work is worth. Pricing is fixed and transparent. We don't discount, and we don't bundle. If our pricing feels like a lot, we're probably not the right lab for you. If it feels like a bargain, you're paying attention.We ship things, not decks. Every engagement ends with something live in production. No discovery phases that produce more discovery phases. No alignment workshops that produce more workshops. The work is the deliverable.

3.We build in public. You'll see what we shipped last week. You'll see what we killed because it wasn't worth keeping. Hiding the process is how studios protect mediocrity. We don't have any to protect.

4. We work with a small number of clients each year. Five to eight engagements maximum. Not because we're being precious — because the math of operating both sides of the lab requires the focus.

cyber:cyber operates primarily London and The Cotswolds, UK. We work with clients globally, primarily across the UK, US, and EU. Most engagements are remote. For larger client work — full builds, operating system installations — we're available to work on-site for short periods when it materially helps the work.