FAQ: Differentiation, Data Safety, Partnership
Standard answers to the questions users, partners, and researchers ask most. Honest register throughout: only verifiable facts, no invented numbers, no exaggeration, probabilistic wording.
Product & Positioning
How is this different from traditional fortune-telling?
We do not make fortune verdicts; we do structured measurement. Traditional destiny knowledge is the knowledge base, reverse-engineered via Symbolic Reasoning into executable, reproducible, traceable rule systems, plus original Vector Relation Analysis modeling. Output is always directional reference; fatalistic assertion words are banned by system red lines. Think "individual-regularity research + data technology company", not mysticism.
How is it different from bazi apps and AI fortune-telling mini-programs?
Three points: depth (an engine base, 7 clusters / 213 vector items, 7 categories / 38 interpretation capabilities vs one-off charts with templates); continuity (year/month/day calendars, commitment-action state machine, reference–action–review loop — users stay); engineering & integrity (~1900 backend tests, 18 ground-truth regressions, anti-fatalist red lines, source citations).
Will strong LLMs replace this?
Quite the opposite — LLMs are the opportunity. They commoditize general knowledge but lack data about the specific individual: they can say how "people like this" generally are, not what "this person" is experiencing at this stage. The engine supplies exactly that individual structured data. And all rule computation is LLM-independent — on model failure the system explicitly falls back to deterministic results.
Data & Verification
Where does user data come from? Is it lawful?
Three principles: users create profiles voluntarily (birth info + gender); data is isolated per tenant and profile; identity-irrelevant information is de-identified before external model calls. Enterprise-side analysis strictly requires lawful, authorized data. Profile deletion is three-stage and thorough, leaving no residue.
How accurate are the results? How do you verify?
Three verification layers: rule-layer determinism (same input + same rule version → same result); ground-truth regression (18 cases + 5 user baseline snapshots, high-precision tolerance); reality-feedback loop (in-product verification and feedback calibration). We never promise 100% accuracy — we promise verifiability, traceability, and continuous improvement.
What does "500+ calculation points" mean? Is it built?
The 500+ point system is under construction. Source files completed a first cleaning pass: 167 candidate points (98 personal, 69 business), 58 relation-model records, 13 vector constructors. Only items passing strict acceptance are counted — the figure reflects system scale under construction, not completed count.
Partnership & Business
How can we partner? What models exist?
Four models: software (channel resale), capability (API & technical integration), training (curriculum), ecosystem (compute, channel, professional-service partners). See Partnership Models, or submit an application.
Is there regulatory risk in destiny content?
We built three risk controls proactively: anti-fatalist red lines (no certainty assertions), professional-advice boundaries (health/legal/finance always disclaimed), and honest register (in-progress features labeled). The product sells self-cognition and decision reference — not luck-changing.
The company is young — is the product real?
The AI Destiny Analysis System is a working cloud product; register and try it now. We also disclose honestly: several advanced modules are in staged construction with per-item lists in the product. Principle: finished is called available; unfinished is clearly in-progress.