Pet care software
Fast delivery and careful engineering in the same product, from day one.
12.5 hours from first commit to first production deploy.
Each figure links to how it was measured.
The starting point
Dog daycares, boarding kennels and trainers run on spreadsheets, paper vaccination records and a booking tool that knows nothing about the dog. Pawsitively is Aternex's first product: one home for the business and its customers, built with no client brief to follow. It had to be trustworthy from the first day, because it holds other people's customer records, and it had to keep moving fast without breaking what was already there.
The change
The first production deploy came 12.5 hours after the first commit. Over the first four days, 158 pull requests merged, none with a failing check, and 158 of 159 deploys succeeded, at a median of 63 s. The six production projects are held to 100% line and branch coverage by 3,186 tests, and the home page scores 98 to 100 in Lighthouse. The product is live at pawsitively.us, with a fictional business at demo.pawsitively.us, and is in evaluation: no one has paid for it yet.
Fast delivery and careful engineering in the same product, from day one.
Inside the work
A closer look at the decisions and changes behind the result.
Designed it as multi-tenant from the first commit: every business is isolated in the server and in PostgreSQL, and staff and pet owners sign in as different kinds of user, so a customer portal only ever shows one household.
Built it as one .NET 10 application with Blazor: public pages are plain server-rendered HTML with no script, and the staff workspace and customer portal run as a WebAssembly app.
Delivered it in small pull requests, each written test first, and let nothing merge without its checks: a build with warnings as errors, every test against a real PostgreSQL 18, 100% line and branch coverage, dependency and secret scans, and a code-health check.
Made each release prove itself before it goes out: the release check applies every migration to an empty database, starts the published app, signs up and signs in for real in a headless browser, keeps the download under a size budget, and restores a backup to compare row counts.
Worked with AI coding agents under those gates, so the speed of AI never meant skipping the checks a careful engineer would run.