There is a badge in this app that says Training needed. It is the most commercially stupid thing we have built, and we are not removing it.
Every other product in this category answers every question. It has to — a blank space looks like a bug, and a plausible paragraph costs nothing to generate. So the incentive runs one way for everybody: always have an answer, and never distinguish between the ones you have tested and the ones you have invented.
We split those two apart and put the split on the screen.
What the badge actually means
The engine's coverage is built one Moon sign at a time. For a sign to count as finished, its interpretive layer has to exist and have been checked against real charts and the lives behind them. That is slow, and it does not parallelise — it is the bottleneck of the entire project, and it will be for a while yet.
So a sign sits in one of three states, and the app tells you which:
| State | What it means |
|---|---|
| Validated | Built and tested against real lives. The app reads this sign in full. |
| Partly built | The framework is there; validation is still running. Where a case is not covered, the app says so. |
| Framework only | The classical layer is complete and correct. The engine's own interpretive layer is not built yet, and it will not pretend otherwise. |
The chart itself, the periods, the transits and the sky are computed the same way for every sign — that is astronomy, and astronomy is finished. What varies is the interpretive depth, and that is the part we refuse to fake.
Why not just write something?
Because we would never find out we were wrong.
The value of this engine is not that it produces sentences. It is that each sentence traces to a rule that could have failed and did not. The moment we let an unvalidated sign emit confident prose, we lose the only signal that tells us which parts of the system actually work — and we lose it quietly, which is the worst way to lose it.
There is a second reason, less noble and just as real: a plausible wrong reading does damage. People make decisions on this. Someone who reads that their marriage house is afflicted, from a sign whose rules nobody has checked, carries that for years. We would rather show a gap.
What we do instead of guessing
- Say which frame a claim comes from. With no birth time there is no ascendant, so anything counted from it would be a noon guess wearing the clothes of a fact. The app switches those readings off and tells you exactly what still holds — your Moon sign, your nakshatra, your entire dasha timeline — and what does not.
- Show the rule. Every finding names the rule behind it, so a disagreement has somewhere to land.
- Take corrections. If the app says something about your life and your life says otherwise, you can file that against the specific finding, from inside the app. Those go into the corpus as counter-evidence. A rule that cannot be contradicted is not a rule.
The commercial version of this argument
We are aware of what this costs. A competitor with an unvalidated engine ships twelve complete signs on day one and looks like a finished product next to ours.
They are also selling something nobody can check, to an audience that has learned not to trust them — which is why the reviews of the five biggest astrology apps are mostly about money and cancellation, not about accuracy. Nobody argues with a horoscope. They just stop paying for it.
An admission is expensive once. A plausible sentence is expensive forever.