Most finance apps tell you what already happened. The forecast is the reason DohFlow exists.
A projection you can interrogate
Every projected row traces back to the assumption that produced it. Select a row and the app shows why it is there — which recurring bill, which detected pattern, which statement estimate. If you disagree, you change the assumption rather than arguing with a number.
Credit cards are modeled statement-aware: the forecast knows the difference between a charge you made and the payment that will actually leave your checking account weeks later. That single distinction is where most naive projections go wrong.
Bands, not false precision
The projection carries empirical uncertainty bands derived from your own history rather than a fixed percentage. It also scores its own readiness: if the app does not have enough history to project confidently, it says so instead of guessing.
Accuracy is backtested. The app measures its own error against what actually happened and reports it, so the confidence you place in a projection is earned rather than assumed.
Scenarios that compose
Ask several what-ifs at once — a raise, a move, a big purchase — with explicit precedence when they conflict. Diff them against the baseline, then apply onto real data with full provenance and a clean reversal path.
Confirm what matters
When the forecast depends on something it is unsure about, it goes in a needs-confirmation queue rather than silently shaping the projection. You answer a short question; the forecast updates.