How KAALOS works

KAALOS is an operational layer for business timing alignment. It combines panchang calculations, founder/team context, and calendar availability to recommend supportive windows for important business work. This page describes the data flow and the guarantees we make.

1. The recommendation pipeline

1. You create a task with title, description, duration, deadline, stakes, and flexibility.
2. OpenAI classifies the task into a business-energy category (strategy, finance, sales, legal, hiring, marketing, product, engineering, operations, conflict, relationship, creative, spiritual_remedial, general). When the AI's confidence is low we respect your selected category.
3. KAALOS generates candidate time windows from your working hours and any imported busy blocks.
4. DivineAPI returns the panchang, choghadiya, and inauspicious-timing data for the relevant date and location. We never invent calculations; if a call fails we tell you and lower confidence.
5. A deterministic scoring engine ranks each candidate window. The score is the canonical evidence; AI does not pick the time.
6. OpenAI explains the top windows using only the scoring engine's evidence bundle.
7. You accept a window. KAALOS creates an internal calendar event and, if you opted in, pushes it to your Google Calendar.

2. What we never do

• We never fabricate panchang, dasha, or chart values. If the calculation is unavailable we say so and the recommendation is downgraded.
• We never let the AI override the scoring engine. AI explains; it does not decide.
• We never use astrology to label a person as weak, unlucky, cursed, or unsuitable. Birth-time precision and consent reduce confidence — they do not produce stigmatising verdicts.
• We never make hiring, firing, compensation, or financial decisions automatically.

3. Data we touch through Google APIs

When you connect Google Calendar, KAALOS uses two minimum scopes: calendar.events (to create an event after you accept a recommendation) and calendar.freebusy (to know which windows are already booked). Free/busy returns opaque busy intervals — no event titles, descriptions, attendees, or locations. We do not use Google data for advertising, analytics, or model training. For the full data-handling notice see the Privacy policy.

4. Why the scoring engine is deterministic

A versioned scoring engine (currently rule_version v1) lets us explain every recommendation in plain language. Every scored window includes a breakdown of panchang, task suitability, profile support, business context, calendar availability, data completeness, and penalties. When weights change we bump the rule version so old recommendations remain interpretable.

5. Confidence is separate from score

A high score is not the same as high confidence. Confidence reflects how complete the underlying data is — calculation completeness, birth-time precision, calendar availability certainty, and AI schema validation. KAALOS surfaces both numbers on every recommendation so you can decide accordingly.

6. Privacy by default

• Birth profiles default to private. Founder profiles default to admin_only.
• Team birth data is opt-in by the individual; founders cannot silently store it.
• Managers see suitability tags, never raw chart judgments.
• Sensitive reads and writes are audit-logged.

7. Disclaimer

KAALOS is a cultural, spiritual, and productivity decision-support tool. It helps you align work with traditional timing systems and business context. It does not guarantee outcomes or replace professional legal, financial, medical, investment, or business advice.

Read the full Privacy policy and Terms of service.