Overview
CapyWatcher is made by one person, and the About page says who. Every feature, page and picture starts with my decisions: what the app should do, how each screen behaves, what a page has to answer and what it must not claim. I use AI as a tool to turn those decisions into code, text and drawings. Neither the AI nor the checks have the final say. Nothing ships until I’ve tried it and approved it.
What this covers
- The app. Its features, its windows, its messages and the code behind them.
- The license server. The service behind the license check and the update check.
- The website. Its pages, layouts, titles and descriptions, the release notes and these legal pages.
- The pictures. The capybara character, the logo and the picture on the About page.
How a change is planned
Before any code or text exists, I write down what the change is for and decide how it should behave. Each change gets a written spec that I review and approve, then a step-by-step plan. AI writes the code and the text to that plan. Rules I keep for the project tell it what to avoid: claims the app can’t back, padding, and design patterns that would make it look like any other product. When I find a mistake that keeps coming back, I add a rule.
The legal pages follow the same path. Their facts come from what the app and the server actually do, and their legal points were researched against the law and the payment provider’s own terms. No lawyer reviewed them.
Automated checks
Every change goes through a fixed set of checks before it can ship:
- Tests. The app, the license server and the website each have their own automated tests.
- Accessibility. Every page is checked against WCAG 2.2 AA in the light, dark and High Contrast themes.
- Design. Pages may use only the app’s own colors, type and parts, and the pictures pass their own checks for contrast and file hygiene.
- Privacy. A test fails if a page other than the legal notice loads anything from another site, or if any page sets a cookie.
These checks catch mechanical problems. They can’t tell whether a feature is useful or a sentence is true. That part is my job.
My review
I try every change myself, in the running app or in the browser, before I approve it. I check that it does what the spec says, that each text says what the app really does, and that the page or the window reads well in every theme. I don’t read every line of code myself. I judge the code by what it does in the app and by its tests.
When something falls short, it goes back with notes: a behavior changed, a claim cut, a layout rearranged, a drawing redone. The new version goes through the same checks and comes back to me. Only a version I approve ships.
The character, the logo and the pictures
- The capybara character and the logo were drawn by AI as vector graphics, one draft at a time. I approved, changed or rejected each draft before the next one was drawn.
- The picture on the About page was made with an AI image tool from my description. It shows a capybara, not a real person.
- The app’s windows on the website aren’t screenshots. They’re built from the app’s own styles and words, with sample data from one example session.
What AI doesn’t do
- The app contains no AI and sends nothing to an AI service.
- Support messages never go to an AI tool. I read and answer every one myself.
- No review, quote or testimonial on this website is made up, by AI or anyone else.
Why I use AI
CapyWatcher is one person’s work. AI lets one person build and test much more than they could alone, so my time goes where it matters most: deciding what the app should do and checking that it does it. It’s never used to fill the website with pages for search engines.
Updates and corrections
A page’s effective date changes when what it tells you changes. A typo fix or a new look leaves the date alone. If you spot a mistake, write to support@capywatcher.com. I fix confirmed mistakes, and the fix goes through the same checks and the same review as anything new.