Configure local endpoints (LM Studio, Ollama) or cloud providers (OpenRouter, OpenAI, Groq, DeepSeek, Google Gemini).
Configure separate astrology calculation providers for Admin and for All Other Users. Admin and users can simultaneously query separate calculation engines.
Configure calculation engines (PyJHora Local, Jagannatha Hora MCP, VedAstro, FreeAstroAPI). All enabled sources can be assigned above.
Directives provided to the AI on every conversation turn. Guides planetary calculation tool calls and authentic classical interpretation.
Non-autoregressive System 1 decision engine. Delivers sub-40ms calibrated intent classification, MCP tool pruning, emotion/persona calibration, and zero-cost memory extraction gating.
http://127.0.0.1:8100 (Managed by Engine/run_laya_service.py & start.bat)
Test any user query or astrological prompt to see real-time Laya System 1 predictions (Domain, Emotion Tone, Profile Requirement, Life Fact Gating, UI Intent).
• Sub-Repository: Laya is cloned in Engine/laya/ from NandhaKishorM/laya.
• 1-Click Updates: Double-click update-engine.bat to pull the latest upstream updates for PyJHora, pyjhora-mcp, and Laya in one step without breaking anything.
• Resilient Fallback: If the Laya service is restarting or disabled, the application continues to answer queries using the standard full tool workflow.