Home / Business automation
Most businesses have jobs that come round every week and eat time. Someone copies figures between systems. Someone rebuilds the same report by hand. I build the software that does those jobs instead, and I hand it over documented so your people can run it.
Fig. 01Who this is for
Smaller and growing businesses
Larger and regulated organisations
Fig. 02Where this usually starts
They come because something takes too long. So I measure it first. If the numbers do not justify building anything, I will say so.
Someone copies figures from one system into another. Someone rebuilds last month's report. It works. It simply costs you, every week.
11 hrs
a week on admin and finance tasks, against 3.6 days a month on sales and growth.
SME Business Barometer, 2026 — 1,000 UK business owners
Plenty of people have had a go with ChatGPT, liked it for a fortnight, then drifted back to the old way. Usually because nothing was built around it.
47%
have never automated a single admin task — and for most, cost is not the reason.
UK Admin Drain Report, 2026
You know the job is slow. You probably do not know what it costs you over a year. Most organisations do not.
83%
have never calculated what admin time costs their business per year.
UK Admin Drain Report, 2026
Sorting it out takes time you have not got, so it stays on the list. Next quarter, every quarter.
54%
say paperwork gets in the way of running their business.
SME Business Barometer, 2026
Fig. 03Measurement
Pick one job that repeats. Put in roughly how long it takes each week and what an hour of that time is worth. It is a rough number, not an exact one. It is usually enough to tell you whether this is worth a conversation.
83% of UK small business owners have never run this calculation. UK Admin Drain Report, 2026
Hours a year
312
Cost a year
£9,360
That's 7.8 working weeks a year, on one task.
Fig. 04Scope of work
This is not a product you subscribe to. It is built around the way your organisation actually works. You get a licence to run it, and the documentation that goes with it.
01
The route an enquiry takes to become a record, then a quote, then an invoice, without anyone copying it between systems by hand.
02
Getting the tools you already pay for to talk to each other, and telling someone when they stop.
03
Staff records, training histories, sign-offs and renewal dates held in one place rather than six spreadsheets, with reminders before anything lapses. Built so the record holds up when someone asks to see it.
04
The small piece of software that replaces the shared spreadsheet nobody wants to be the one to break, and the reporting that sits on top of it.
05
Web apps your team opens in a browser, built to work properly on a phone for people who are out on site rather than at a desk. Built when a form or a spreadsheet is demonstrably not enough, and not before.
06
Proposals, certificates, compliance packs and monthly reports, produced from your data in your format.
07
Reading documents, drafting from templates, sorting and routing. Used where it measurably beats a simpler approach, and a person still makes the decisions that matter.
Fig. 05Method
Five stages, always in this order. Most projects that go wrong went wrong before anyone wrote any code, usually because nobody agreed what finished meant.
We work through the jobs as they are genuinely done, and establish which are worth automating. Some won't be. Too rare, too varied, or cheaper left alone. I will tell you which. We also agree here what success looks like as a number.
Output A shortlist, with an estimate against each one.
I write down what the system will do, what it will not do, what data it touches, what happens when a step fails, and what it has to pass before it counts as finished. The fixed price comes out of that document.
Output A written scope with acceptance criteria, approved by you.
Built to the spec, with logging and error handling in from the start rather than bolted on at the end. You see it as it goes. No long silence followed by a reveal.
Output A working system, running on test data.
I test it against every criterion we agreed, including the awkward ones. The malformed file. The missing field. The duplicate record. The run that gets interrupted halfway. Then your people test it. It is not finished because I say it is.
Output Written test results, and your sign-off.
Documentation written for whoever picks it up next, a walkthrough with the people using it, and a check-in a month later. It should carry on working whether or not I am still involved.
Output Documentation, training and a 30-day review.
Fig. 06Governance & data
Working with one person should mean less admin, not less rigour. These are my standing positions. They apply whether the job runs two weeks or a year.
Data protection
A signed data-processing agreement before I touch any personal data. The system only gets the data it needs to do the job. If you need a DPIA, I will support it.
AI data handling
Live systems call commercial APIs, ideally on your own account so the data relationship is yours rather than mine. Never a consumer chat tool. Development runs on dummy or anonymised data.
Intellectual property
I keep ownership of the software and the reusable parts behind it. You get a perpetual licence to run it, with full documentation, for as long as you want. If you need ownership transferred outright, that is available and priced separately. Either way it is agreed in the scope before I start.
Provenance
Everything is built from the spec, for you. Nothing from another client or engagement gets carried in, and there is no proprietary component that locks you to me.
Assurance
Professional indemnity and cyber cover in place before any engagement starts, under written terms, a scope letter and a data-processing agreement.
Continuity
Standard tooling and documentation written so another developer could pick it up. That is deliberate. It is also the honest answer to the risk of working with one person.
Fig. 07Engagement
Fixed price against a written scope, agreed before I start. If the scope changes, we agree the change and what it costs before anything gets built.
Small and growing businesses
Sized so the first commitment is small and you see something working before you commit to more.
Larger and regulated organisations
Priced per programme after a paid discovery stage, then delivered in stages with a sign-off at each one. You only commit to the next stage once the last one has landed.
The small-business rates are founding rates. BrightDragon is new. You would be taking a chance on someone without a long public record, and the price reflects that. What I ask in return is permission to write up what we did, anonymised if you would rather. The rates go up once I have that record.
Fig. 08Questions
Fair question, and it is the real trade-off of working with one person rather than a firm. Two things help. Everything is documented so another developer can take it on, which is a deliberate goal rather than an afterthought. And I build with standard, widely used tools, so you are never stuck with a component that only I understand. On larger programmes we can write continuity arrangements into the contract.
For some programmes, no. I will tell you that rather than take the work and let you down. What one person does well is depth on a defined problem. A records system, a reporting pipeline, an integration between two systems that will not talk. Work is staged so you can judge me on the first stage before committing to the next. If the scope is more than one person should reasonably carry, you need a larger supplier, and I will say so.
I do. You get a perpetual licence to run it, plus the full documentation, for as long as you want it. I keep ownership of the software and the reusable parts underneath it, which is what stops me rebuilding the same foundations from scratch for every client and charging you for it. If you want ownership transferred outright, that is available and priced separately. Whichever it is gets agreed in the scope before I start, not afterwards.
Development runs on dummy or anonymised data wherever I can manage it. Where a live system uses AI it calls a commercial API, ideally on your own account so the data relationship is yours. Never a consumer chat tool. Anything involving personal data is covered by a data-processing agreement signed before work starts. If you have specific regulatory requirements, tell me early and I will design around them.
Then I will say so. Often the better answer is simpler and cheaper anyway. A fixed rule, a tidier process, or a tool you already pay for and are not using properly. Recommending something that makes me money and does nothing for you is a bad trade for both of us. Some of the most useful results of a diagnosis are the jobs we decide to leave alone.
Usually the problem was that nobody wrote down what done looked like, so afterwards nobody could say whether it had worked. Every job here starts with a written scope and acceptance criteria you approve, and ends with a test against each one. If it does not meet them, it is not finished. That is also what the fixed price is tied to. You are paying for a result, not for my time.
ContactNext step
Thirty minutes, nothing to prepare. Tell me what you are trying to sort out and I will give you an honest view on whether it is worth building anything.
Or contact directly
hello@brightdragonsystems.com