Byte-exact engines — not approximations
Every calculation is a step-by-step implementation of the rules in the official guides — the income test, the 7-band cost percentage, the recognised participation cap (formerly called the activity test, renamed 5 Jan 2026), the supplements, the taper rates. Each engine is hand-traced against the published formula before it's allowed to ship.
198 regression tests on every deploy
We've locked in 198 hand-checked scenarios — real-shape families, with the answer reconciled to within 50 cents of the rulebook. Every code change runs the full test suite. If a single one drifts, the deploy doesn't go out.
A schema-guard gate stops drift between code and database
A build-time check reconciles the columns the app writes against the live database before any deploy ships. A code change that would write a field the database can't accept fails the build automatically — your data never lands in an inconsistent state.
Rate updates within weeks of the Government publishing them
Every rate, threshold, and tax bracket lives in one registry, pinned by financial year. When Services Australia or the ATO publishes new figures (CPI indexation each March/September, ATO each July), we update the registry, re-run all 198 tests, and ship.