What is actually built
Written from the repository rather than from a pitch. Every count below is computed when this page loads; where a figure would have to be invented, the section says so instead.
The product, counted
Computed from frontend/screen_table.js and frontend/data_sources.js when this page loaded - the same registries the terminal navigates by.
What makes it different
- Every figure on every screen traces to a named source with an as-of date. Where a source is unavailable the screen says which one and why, instead of substituting a zero - that rule is enforced by the test suite, not by policy.
- It runs entirely on the reader's own machine if they want it to. Nothing about a reader reaches a third party: no trackers, no analytics vendor and no embedded third-party frames. The one record this app can keep about you - a log of which screens you opened - is off until you turn it on, stays on this install, and is deleted when you turn it off.
- The analysis is algorithmic and inspectable rather than generated. A reading that says a margin moved names the two figures it compared.
The no-third-party claim is enforced by tests/test_no_third_party_embeds.py, which fails the build if an iframe to an external host is added.
How it would make money
Four tiers are defined in app/services/entitlements.py. The pricing page is disabled while billing is off, which is the default in this edition, so no money can change hands. The terminal needs a free account; the paid tiers would add usage headroom rather than withholding analysis.
What is unproven
There are no users, no revenue and no retention data, so none is shown. A traction chart drawn from nothing would be the exact failure this product exists to avoid, and an investor who found one here would be right to stop reading.
The build is a working terminal with an honest empty-state discipline. That is what is being shown; everything commercial is ahead of it.
This section is deliberately empty of figures.