12 May 2026
Why we still ship desktop apps
A browser tab is a promise that somebody else's server will be online, that the network between you and it will behave, and that a company you do not control will keep the lights on. Usually that promise is fine. On the days it is not, people who work outdoors, underground, or in a basement office find out the hard way.
A desktop app is a promise too, but a smaller one: install it once, and it works until the machine breaks. It starts in under a second, it does not reload when the wifi blinks, and it will still open the file in ten years without a maintenance contract.
That is the trade we make most of the time. It costs us the convenience of pushing an update without asking, and it gives the person using it something they can rely on. For tools used every day by a small team, that trade has been worth it every time.
3 February 2026
The cheapest backup is the one you actually test
Every company we have worked with claims to have backups. In three cases out of four, the first real restore test found something missing: a database that had been excluded, a key that lived only on the laptop of someone who left, a nightly job that had been failing silently since a version upgrade.
A backup you have not restored is a hope, not a backup. The test is simple and unpleasant: take a copy, put it on a clean machine, and try to work from it for an hour. Anything you cannot open is not backed up.
We now write the restore date on the wall of every project we hand over. It is the least glamorous line in the documentation and the only one anybody reads after the first incident.
18 November 2025
Spreadsheets are not the enemy
It has become a habit to blame spreadsheets for everything that goes wrong in a small business. We think that is unfair. A spreadsheet is a very good tool: it is immediate, it is flexible, and the person who owns the process can change it without asking anyone.
The trouble starts when a sheet outgrows the job it was built for — when six people edit it, when it becomes the only record of something, when a formula error costs a week. That is a signal about process, not about the tool.
Our usual answer is not to delete the spreadsheet. It is to move the part that must be correct into a small program, and leave the sheet for what it is good at: thinking out loud with numbers.