Services
We are a small studio, so we are deliberately narrow. These are the three things we do, and we would rather turn work down than take something we would be learning on your budget.
Software you will not have to rewrite
Backends, web applications and internal systems, built so the next person can safely change them.
The first version is rarely the hard part. The hard part is month eighteen, when the person who wrote it has gone and something has to change anyway. We build for that month, because on a long enough timeline you are the one standing in it.
- A deployed, working system - not a prototype that needs redoing
- Documentation written as it is built, not bolted on afterwards
- Tests where they stop real failures, not to reach a number
Know what it costs before you commit
One to two weeks of planning, at a fixed price, ending in a number you can put in a budget.
We model the data, set the service boundaries, choose the infrastructure and price the build. What you get is a document you could hand to any team in the world, including one that is not us.
- A written technical plan, yours whoever ends up building it
- What it costs to run every month, not only to build
- A straight answer if the thing should not be built at all
We will take on what is already breaking
Systems that are slow, fragile, undocumented, or written by somebody who has since left.
We start by making the current behaviour visible, because changing a system nobody can observe is guesswork with a deploy button on it. Then we fix things in the order of what they are costing you, which is rarely the order they are annoying you.
- Monitoring and alerts that name the thing that broke
- Your failures ranked by what each one actually costs
- A runbook, so the next incident is not an emergency
How we build
Three habits that decide whether your software gets cheaper or more expensive every year you own it.
Measure before changing anything
An optimisation with no measurement either side of it is a guess that cost money. We take the number first, change one thing, then check the number moved. Slower for a day, faster for a year.
Do the expensive work only when it is needed
Most slow systems are not badly written. They are cheerfully redoing work that nothing asked for. The first question about anything on a schedule is whether it has to run at all.
Write down what can reach what
Every service gets a documented answer to "what can reach this?". A security model that lives in one person’s head leaves the company on the day they do.
Commercial terms
Start with two weeks, not two years
A paid discovery first: one to two weeks, fixed price. You get the technical plan and a fixed quote for the build. Take that plan to another company and it is still yours - that is the thing you paid for.
A date and a price, not an hourly rate
We quote a defined piece of work with a delivery date, rather than selling you hours and letting them run. If the scope changes we re-quote in writing before anything is built, so no invoice ever arrives with a surprise inside it.
Nothing is ever held hostage
Code, infrastructure and accounts sit in your organisation from the first commit. There is no handover at the end because there is nothing left to hand over - and nothing that makes walking away from us expensive.
Who we are not for
We would rather turn work down than take something we would be learning on your budget. These are the ones we say no to.
- Staff augmentation
- We do not rent developers by the hour or the month. If you need somebody to fill a seat on your team, an outsourcing agency will serve you better than we will.
- Undefined scope
- We do not start building before knowing what has to be built. That is what the paid discovery is for, and skipping it is how fixed-price work turns into an argument.
- Large teams and 24/7 on-call
- If the work needs a dozen engineers at once, or somebody awake at three in the morning, we do not have that structure. We say so in the first conversation.
- The cheapest quote
- We do not bid against other companies on price. The number is fixed and it carries the risk of the delivery date, which a cheaper number usually does not.
- Starting next week
- Planning comes before the first line of code, and the planning takes one to two weeks. Anyone who can start on Monday has not understood the problem yet.