The product
It agrees your year to its evidence.
Seven questions in, a record out. Every dollar gets a class and the evidence that class needs. Every owner gets a determined U.S. tax status. The officer-compensation trigger gets resolved, the state clock gets computed from your formation date, and the whole thing gets scored for whether a preparer could work from it.
A rules engine decides
A versioned rules engine, not a language model. Each conclusion carries its rule id, its pinned version, the facts it used, the sources it cites with retrieval dates, and whether a professional still has to look. Where a model is used at all, it asks the question in plain words and explains the answer. It never sets one.
The difference shows up in one place: open any finding on the homepage and read the rule behind it. A chatbot cannot show you that, because there is nothing behind its answer to show.
Four checks, then a gate
- Who owns it.
- Every dollar that moved, classified, with funding kept apart from revenue.
- What the company paid you, through every channel, and whether that triggers wages.
- The state's own clock: when the annual report was due, and the dissolution window after it.
Then a fifth: whether the record, taken together, is complete enough for a licensed preparer to start from. That gate can say not ready or complete enough. It has no fine.
We are not a firm, and we do not file
Twenty refusals, published in full, at what we will not do. Most are things you can buy elsewhere today.
Free at the door
The check is free and always will be. The tiers above it are listed on pricing, without prices, because we have not set them.