Desktop software · quoted per project
Software that runs on your machines, not someone else's cloud.
Some work does not belong in a browser. Files that never should leave the building, tasks that must run whether the internet is up or not, jobs measured in gigabytes rather than page views.
Tools for the work that does not fit a website.
Usually this replaces a person spending three hours a week doing something a machine should have been doing all along.
- Desktop applications in C#. Proper Windows software with a real interface, for staff who need a tool open all day rather than a browser tab among thirty.
- Data and automation work in Python. Scraping, cleaning, matching, converting and reporting — the jobs that are tedious by hand and instant once written down.
- Scheduled jobs that just run. Overnight imports, nightly reports, folder watchers. Set up so nobody has to remember to press anything.
- Excel and CSV handled properly. Reading messy real-world spreadsheets, and writing ones that are formatted the way your team already expects to receive them.
- It runs offline. No dependency on a connection or a third-party service that can change its pricing or disappear.
- Errors that explain themselves. A log that says what went wrong in a sentence, so the first response is not to ring me.
- A real installer. Something your team can install without a developer standing over them, and update the same way.
- Source code, handed over. Yours outright, documented, buildable by anyone competent. No licence server, no phone-home, no seat count.
Three hours a week is a hundred and fifty hours a year.
The tasks worth automating are almost never the dramatic ones. They are the small, dull, every-week jobs nobody thinks to mention.
- Repetitive manual work is where mistakes come from, and the mistakes cost more than the hours
- Data that must not leave your building genuinely should not be uploaded to a third-party service
- A tool bought per seat gets more expensive as you grow; one built once does not
- Nightly jobs run at 3am without anyone being awake to start them
Scoped properly before anything is quoted.
Same as web applications: a price before the edge cases are understood is a price that gets revised.
You describe the job
What is being done by hand now, how often, and by whom.
We scope it together
Inputs, outputs, and every awkward exception written down. The exceptions are where the real work hides.
A fixed price, in writing
One number for a defined tool, with the maths on what it saves you.
You test it on real data
Run it alongside the manual process until you trust it, then stop doing it by hand.
The ones that come up every time.
C# when people need to sit in front of an application all day on Windows. Python when the job is data moving from one shape to another and nobody needs to watch it happen. Plenty of projects use both, and this is my decision to make, not a question you should have to answer.
Python does, anywhere. C# can, though desktop applications with a full interface are usually simpler to keep Windows-only unless you genuinely need otherwise. Tell me what your team actually uses and that decides it.
It depends entirely on what you are collecting and from where, and I will give you a straight answer rather than a convenient one. Public data for your own analysis is generally fine. Anything behind a login, personal data, or a site whose terms forbid it is a conversation before it is a build.
Quoted per project, because a report generator and a full desktop application are not the same job. Small automation tools frequently land under $1,000. I will also tell you when a scheduled script would do rather than an application.
You own the source, so anyone can. Practically, these tools need very little once they are working — they are not exposed to the internet and nothing forces updates on them. If something breaks I would rather fix it than have you fighting it.
Tell me what's being done by hand.
If it happens every week and follows rules you can describe, it can probably be written down once and never done manually again.