Why We Are Building GUVNR Inside a Real UK Contractor

There is a particular kind of software demo that every contractor has sat through. Everything is clean. The example job has one line item. The invoice goes out and comes back paid. Nobody mentions retention, nobody mentions CIS, and the word "valuation" does not appear once.

It is not that those products are bad. It is that they were designed by people who have never had to get an application for payment in by the 25th, with half the labour hours still sitting in a WhatsApp thread and a quantity surveyor who will happily knock off anything you cannot evidence.

So when we started building GUVNR, we made one decision that has shaped everything since: we would build it inside a working contracting business and use it for real work, before anybody else touched it.

Who is actually building this

GUVNR is built by Alpha Controls LDN Ltd, a BMS and controls contractor working across Kent and London. We install, commission and maintain building management systems on commercial buildings — usually as a subcontractor inside a larger M&E or main contract. Trend, Distech, that sort of work.

Which means we are not a software company that researched a market. We are a contractor that got fed up. The applications for payment in GUVNR exist because someone had to submit one. The CIS handling exists because the numbers had to be right for HMRC. The shift reports exist because we needed to know what happened on site without ringing four people.

Alpha Controls is tenant number one, and — as of today — the only tenant. That is not a soft launch position we are dressing up. There is genuinely one company on GUVNR, and it is ours.

What building it this way gets right

You cannot ship a feature that does not work

If the invoice module gets the reverse charge wrong, we send a wrong invoice to a main contractor we want more work from. That is a very effective quality gate. There is no product manager to convince and no user-research session to schedule — the feedback arrives as a phone call from a QS.

It also kills a whole category of plausible-but-useless feature. Every contractor tool has a screen somebody thought was a good idea and nobody has ever opened. Building inside the business means the cost of a useless screen is immediate: it is a screen we have to walk past every day while the actual job is still not done.

The edge cases arrive first, not last

In most software, the awkward cases turn up months after launch, when a customer emails to say the numbers do not add up. In ours they turn up on day one, because the awkward cases are the job. A payment application that has to reconcile against the previous one. Labour that is CIS-deductible and materials that are not. A job where retention is 5% but the contract says 3% after practical completion. Those are not edge cases in this trade. They are Tuesday.

You find out what people actually do

Engineers do not fill in forms. They will take a photo and type six words. Any system that requires more than that gets quietly abandoned, and you find out three weeks later when you need the record and it is not there. We know that because we built a thing that asked for too much and watched it fail.

What it costs us — honestly

This approach is not free, and it would be dishonest to write a founder story that made it sound like it was.

It is slow

The people building the product are also the people running the jobs. When a site goes wrong, the software waits. Nothing gets built in the week of a big commissioning push, and we have shipped a lot of features at eleven at night. A funded team of six would have got further by now, and we have no interest in pretending otherwise.

It is narrow, by construction

Everything in GUVNR is shaped by one company's way of working. Our project structure, our approval habits, our tolerance for paperwork. That is exactly the risk of dogfooding: you can build something that fits one business beautifully and no other business at all. This is the single biggest reason the Founding 10 exists, and why it is ten companies and not one.

It makes you conservative about automation

When it is your own money and your own reputation on the line, you do not let software send things on your behalf. That has become a design principle — anything commercial that leaves the business stays behind a human — but it started as simple self-preservation. Someone building this at arm's length would probably have shipped autonomous chasing a year ago. We have not, and we are not going to.

The line we hold on what we claim

The uncomfortable side effect of building in public while running a real business is that you have a strong incentive to describe the product as further along than it is. So we set a rule early and we are keeping to it.

Every product statement we publish falls into one of four buckets: available now, currently being tested, under active development, or planned. Something moves up a bucket when there is real work going through it — not when the code merges, and not when it demos well.

Concretely: invoicing with CIS is available now, because the invoices we raise ourselves carry a CIS position and the VAT treatment we set on them. Applications for payment are being tested, because the feature is built but we are still doing this month's valuation the old way while we prove it. Those two sentences would be a lot more attractive if they were both the first one. They are not, so they are not.

You will find the full breakdown in what GUVNR is and who it is being built for, and we keep it current rather than letting it drift.

Why the Founding 10 is small on purpose

The obvious failure mode of building software inside your own business is building software that only works in your own business. The only real fix is other people's businesses, early, in enough depth that they will tell you when something is wrong rather than politely stopping using it.

So: we are selecting our Founding 10 — ten hand-selected UK companies who will help shape and test GUVNR against their real workflows. Not ten logos for a website. Ten firms whose applications for payment, CIS position and site paperwork are different enough from ours to break the assumptions we have baked in without noticing.

After that comes Early Access 30, a controlled cohort of thirty companies, and wider release comes after that. Joining the waitlist is how you follow the build and apply for one of those places. We are not going to promise everyone a place, because thirty is thirty.

Why bother at all

Fair question. There is software in this category already, some of it decent, and building your own is a spectacularly inefficient way to solve your own admin problem.

The honest answer is that we kept hitting the same wall. The tools that understood construction were priced and shaped for main contractors. The tools priced for a firm our size were built for domestic service work and could not model a valuation. And every one of them was a filing cabinet with a nicer front end — a place to put the admin, not a thing that would do any of it.

We think there is a third option, and we would rather find out by building it than by sitting through another demo where the invoice comes back paid.