For twenty years the honest answer to a good idea was often no. Not because it could not be built, but because finding out took weeks of discovery, and building it cost more than the idea was worth. Ours was one of the industries that answered that way. Briefs got trimmed to what was safe, the interesting half went in the bin, and everyone called it being realistic.
That arithmetic has changed, and it has changed more in the last two years than in the twenty before them. Working software now arrives in the time a scoping document used to take. So the constraint has moved. The question is no longer whether a thing can be built, or what the research would cost before anyone found out. It is whether it is the right thing to build.
We have rearranged the work around that. Less time producing documents about software, more time with the software itself in front of you.
How a project runs.
You see the thing, not a proposal
Most agencies open with a document. We open with a working, branded prototype of your actual system, running on your real data wherever we can get it. Days, not months. You decide what is worth funding by using it, which is a far better test than reading about it.
AI writes much of it. Engineers answer for all of it.
Claude Code sits inside our workflow the way a compiler does: drafting, refactoring and testing faster than any team could by hand. Every line still passes a person who is accountable for it in production. That is the whole of the trick. The pace has changed; the standard it is held to has not.
Scope stays open far longer than it used to
The old process front-loaded every decision because changing your mind later cost weeks. It does not any more. We keep the shape of the thing negotiable well into the build, so what you get is the version you worked out you needed, not the one you guessed at in a kick-off workshop.
Tests are written with the code, never after it
Every change arrives with the tests that prove it, and the suite runs before anything reaches your users. On a system like The Global Health Network, twelve years of code and still growing, that suite is exactly what lets us move quickly without breaking the parts nobody has touched in years.
Accessibility, security and data protection are constraints, not a phase
WCAG, Cyber Essentials and UK GDPR are designed in from the first prototype, because retro-fitting them is where budgets go to die. We are Cyber Essentials certified and a B Corp, and most of what we build is for organisations audited on all three.
Built to be handed over
You own the code, the infrastructure and the documentation, and we train your people to run it. We stay on afterwards if you want us, never because you are stuck with us.
Small team. Institutional scale.
The scale we work at has not shrunk. We support platforms for the University of Oxford and the NHS, built the licensing system that a whole sector runs on, and stood up the infrastructure behind a national launch. What has shrunk is the number of people it takes, and the time between deciding to build something and watching it work.
That means the answer to "can you build..." is now almost always yes. The interesting questions are the ones after it. Is it worth building? What is the smallest version that would prove it? Those are the conversations we would rather be having with you.