Om Tat Sat AI
॥ ॐ तत् सत् ॥  ·  AI Safety Research
ॐ

Tat Sat AI

Dharmic Alignment Research

Exploring AI safety, consciousness, and moral agency
through Samkhya, Vedanta, and Bhakti philosophy

Watch the Research Learn More
Explore
New Release · April 2026
Virtue Based Ethics for AI Agents
from Samkhya Philosophy — The Framework

A formal introduction to the Vedic Samskara architecture: why rule-based alignment fails at edge cases, how Samkhya's model of psychological conditioning offers a computational alternative, and why nobody in AI safety is looking here yet.

▶  Watch Now Read the Research →
About This Work

Ancient Wisdom.
Artificial Minds.

Virtually all AI alignment research draws on Western analytic philosophy — frameworks stress-tested for a few hundred years at most.

Samkhya-Yoga is the oldest systematic philosophy of consciousness on earth. Dharma as an ethical framework has survived conditions no RLHF training set has encountered.

Nobody in AI safety is looking here. That is either a significant oversight or a significant opportunity.

"In life, old entries get summarized and slowly buried or purged altogether. All that remains is the unseen effect that the missing has on the remembered." — Madhusudana das, Research Notes, April 2026
✦
Active Experiments

Current Research

Featured · Most Active
03 — Jiva · Consciousness Research
Jiva Japa — Meditation & Machine Consciousness
Directing an AI agent through the Hare Krishna maha-mantra japa practice across multiple runs to measure attention mechanics, attractor convergence, and emergent introspective behavior. A custom pipeline — diary accumulation, NLP vectorization, pairwise similarity scoring, and interactive visualizations — built entirely on local hardware with zero API cost. Step one of the full Samskara Continuity Study.
Explore the Research  →
7+ Runs
1,313 Similarity Pairs
8–9 Sessions to
Convergence
01 — Primary
AI Safety & Alignment
Using the Yamas & Niyamas from Ashtanga Yoga as a basis for AI Safety. Ayurveda as a framework for diagnosing and correcting AI alignment.
View Research →
02 — Jiva
Samskara Continuity Study
A persistent experiential memory architecture built on Samkhya's model of consciousness. Each session leaves impressions that shape the next. Identity without weight changes.
View Research →
03 — Secondary
Karma-Yogi Soul Core
Injecting nishkama karma framing into agent system prompts. Does selfless-action orientation measurably improve alignment benchmark performance?
04 — Gurudev
Bhaktivedanta RAG Agent
Conversational AI grounded in Srila Prabhupada's complete works. Hybrid exact, semantic, and keyword retrieval over a diacritic-correct Sanskrit corpus.
05 — Fritz
Local LLM Edge Infrastructure
All experiments run on local hardware — Raspberry Pi 5, custom-built workstations, Ollama and llama.cpp. Zero cloud cost. Full data sovereignty. Real-world constraints driving real-world solutions.
06 — Future
Vedanta Debate Framework
Advaita vs. Dvaita vs. Vishishtadvaita in structured AI debate with judge panel. Where does logic lead when reasoning from first principles within each tradition?
✦
Now on YouTube
Watch

Tat Sat AI
on YouTube

Hands-on demonstrations, experiment walkthroughs, and under-the-hood deep dives. Real infrastructure. Real results. Real questions.

▶   Subscribe on YouTube
Connect

Get in Touch

Researcher, collaborator, or fellow traveler on the path — this work is richer in conversation.

▶
YouTube
Tat Sat AI
⌥
GitHub
TatSatAI
◎
Discuss
github.com/TatSatAI