How we actually use AI (our honest AI policy)

We use AI in the studio, openly and every day, as a tool that speeds up parts of the work. We do not use it to replace the judgement, the design decisions, the security thinking, or the human responsibility for what we ship. This is our honest policy on it: where it helps, where it does not belong, and why we would rather tell you than pretend either that we avoid it or that it does everything.

As a studio repositioning around AI and apps, I think we owe clients clarity on this rather than either breathless hype or defensive silence. So here it is, plainly.

Where AI earns its keep

There is real, unglamorous work that AI genuinely speeds up. It helps us write and check boilerplate code faster, draft and refine technical configuration, summarise research, produce first drafts of text that a human then rewrites, and explore options quickly. Used this way it is a power tool: it removes drudgery and frees up time for the parts that actually need a person. We are not precious about using it, and we are not going to charge you for hours of work a good tool can do in minutes.

Where humans stay firmly in charge

The limits matter more than the uses. The architecture of a site, the design decisions, the security review, the strategic judgement about what your business actually needs, all of that stays with a person, every time. AI is confidently wrong often enough that nothing it produces ships without a human checking it. It does not understand your business, it cannot take responsibility, and it has no judgement, only patterns. So we treat it as an assistant that drafts and suggests, never as the thing that decides. When you hire us, you are paying for that judgement and that accountability, not for raw machine output.

The energy cost, honestly

AI is not free, and not only in money. The data centres behind these tools draw real and growing amounts of electricity, which is part of why we retired our carbon neutral label rather than pretend otherwise. So we use AI where it genuinely improves the work, and not as a reflex. Burning compute on something that does not make the result better sits badly with us, partly on principle and partly because it is simply waste. Using it deliberately rather than by default is both the greener and the better-quality choice.

Being open about it

We would rather disclose than hide. If an AI tool drafted the first version of something, or generated an image, we are happy to say so. The norms and the rules are moving this way anyway, with new transparency requirements coming into force for AI-generated content, but we would do it regardless, because the alternative is letting clients believe a person did something a machine did. Honesty about the tools is part of the same rule that governs everything else here: do not claim what is not true.

The bottom line

We use AI as a serious tool, kept on a short lead, with a person responsible for everything that leaves the studio. That balance, not blind automation and not pretend purity, is how we think it should be used. This piece is part of how we run the studio, and it reflects how we approach the work we build for every client.