Vidurai
The Conscience of the Machine
"विस्मृति भी विद्या है"
Forgetting too is knowledge
Teaching AI the ancient art of memory and strategic forgetting,
inspired by Vidura's wisdom from the Mahabharata
Why Vidurai?
While everyone races to give AI perfect memory,
we asked a different question:
What if forgetting is the key to true intelligence?
🎯 The Problem We Solve
Store everything, remember nothing useful
Current AI memory systems treat all information equally, leading to noise drowning out signal.
Costs explode with token accumulation
Every API call gets more expensive as context windows fill with irrelevant data.
No distinction between important and trivial
Your preference for vegetarian food is stored with the same weight as "hmm, let me think."
Memory becomes noise over time
Without intelligent decay, the signal-to-noise ratio only gets worse.
✨ The Vidurai Solution
Three-Kosha Memory Architecture
Inspired by Vedantic consciousness layers
Annamaya Kosha
The Physical Sheath
High-speed, volatile working memory. A sliding window for immediate context. Built for speed, not permanence.
Manomaya Kosha
The Mental Sheath
Active episodic memory with intelligent decay. Memories fade based on relevance and usage, not just time.
Vijnanamaya Kosha
The Wisdom Sheath
Deep archival memory. Distilled wisdom, core preferences, foundational knowledge. Updated, never forgotten.
Vismriti Engine - The Art of Strategic Forgetting
Production-verified performance metrics (v1.5.1)
Verified Token Reduction
Real savings, not estimates
Cost Savings @ 10K Users
Production-grade optimization
Recall Reliability
With configurable decay
🧠 Vismriti RL Agent - The Learning Brain
The first self-learning memory optimizer that gets smarter with every decision
❌ Traditional Systems
- Hardcoded rules
- Same for everyone
- Manual tuning required
✅ Vidurai RL Agent
- Learns via Q-learning
- Adapts to your patterns
- Intelligence from experience
Epsilon-Greedy Exploration
30% explore → 95% exploit over 1000 episodes
Multi-Objective Optimization
Balances cost vs quality
Continuous Learning
Improves with every decision
Intelligence emerges from experience, not rules.
Viveka Layer - The Autonomous Conscience
What separates Vidurai from a simple database:
- Autonomous Importance Scoring: Scores memories based on emotional significance, user goals, and surprising content
- Dharma Alignment: Programmable ethical guardrails to prevent storing biased or harmful information
- Learns What Matters: Adapts its understanding of importance based on user interactions
🚀 Quick Start
pip install viduraifrom vidurai import Vidurai
# For critical applications (recommended for v1.5.1+)
memory = Vidurai(enable_decay=False)
# Or with configurable decay for general use
# memory = Vidurai(decay_rate=0.98) # Slower decay (default: 0.95)
# Simply tell Vidurai what happened
memory.remember(
session_id="user123",
content="My name is Alice. I'm a vegetarian planning a trip to Japan."
)
memory.remember(
session_id="user123",
content="Hmm, let me think about the dates."
)
# Later, recall only what's relevant
context = memory.recall(
session_id="user123",
query="What are some good food options there?"
)
# Returns: "User is Alice. User is a vegetarian.
# User is planning a trip to Japan."
# The trivial "hmm" is automatically forgotten.📖 Philosophy
Vidurai is named after the wise minister from the Mahabharata, known for:
Speaking truth to power
Ethical decision-making
Strategic wisdom
We embed these principles in code, creating AI that doesn't just remember—it remembers wisely.
🤝 Join the Sangha
Vidurai is built by the community, for the community.
Be part of building the conscience layer for AI.