yod 0.2.1

Web-based personal memory assistant backend (FastAPI + Neo4j + Qdrant + OpenAI)

---

Memory Infrastructure for AI Agents

Evidence-preserving memory with full citation provenance

Yod extracts structured knowledge from conversations, stores it in a knowledge graph, and retrieves relevant context with full citations. Unlike memory systems that compress into summaries, Yod preserves the exact quotes that created each memory — enabling explainable AI.

Why Yod?

Feature	Yod	Mem0	Notes
MERGE Operation	✅ ADD/UPDATE/DELETE/KEEP/MERGE	⚠️ ADD/UPDATE/DELETE/NOOP	Yod can combine partial truths
Citation Provenance	✅ Quote + Source links	⚠️ Timestamps only	Yod traces to exact evidence
PII Redaction	✅ Built-in	❌ Not mentioned	Yod redacts before LLM/storage
Quote Storage	✅ Exact quotes stored	❌ No quote tracking	Yod preserves source evidence
Session Scoping	✅	✅	Both have user/agent/session
Temporal Queries	✅	✅ (Mem0g)	Both support point-in-time
Enterprise Security	⚠️ AES-256	✅ SOC2/HIPAA	Mem0 ahead on certifications
Multi-LLM Support	✅	✅	Both support multiple providers
Self-Hosted Full Stack	✅ docker-compose	⚠️ Cloud-first	Yod runs locally in 5 min
Open Source Completeness	✅ Frontend + Auth + Deploy	⚠️ Core only	Yod is fully open

Quick Start

Cloud API

pip install yod

from yod import YodClient

client = YodClient(api_key="sk-yod-...")  # Get a key at https://yod.agames.live

client.ingest_chat("My name is Alex. I work at Google.")
response = client.chat("Where do I work?")
print(response.answer)     # "You work at Google."
print(response.citations)  # [Citation(quote="I work at Google")]

Self-Hosted

git clone https://github.com/agames-live/yod.git && cd yod
cp .env.example .env       # Add OPENAI_API_KEY
docker-compose up -d

Open http://localhost:3000 and sign in with Google/GitHub.

Features

• Evidence-Preserving — Every memory links to the exact quote that created it
• Smart Conflict Resolution — LLM-based decisions: ADD, UPDATE, DELETE, KEEP, MERGE
• Entity Resolution — "my dog Max" links to the same entity across conversations
• Temporal Versioning — Query memories as they existed at any point in time
• Session Scoping — Isolate memories per conversation or agent
• Multi-LLM — OpenAI, Anthropic Claude, or local Ollama models
• Full Stack — Frontend, auth, and infrastructure configs included

Architecture

┌─────────────────────────────────────────────────────────────┐
│                        Yod API                              │
├──────────────────────────────┬──────────────────────────────┤
│     Ingest Pipeline          │       Query Pipeline         │
│  ┌────────────────────────┐  │  ┌─────────────────────────┐ │
│  │ Text → Semantic Chunks │  │  │ Semantic Search (Qdrant)│ │
│  │ LLM → Claims + Entities│  │  │ Graph Traversal (Neo4j) │ │
│  │ Conflict Resolution    │  │  │ Citation Assembly       │ │
│  │ Evidence Storage       │  │  │ LLM Response Generation │ │
│  └────────────────────────┘  │  └─────────────────────────┘ │
├──────────────────────────────┴──────────────────────────────┤
│  Neo4j (Graph)    │   Qdrant (Vectors)   │   Redis (Cache)  │
└─────────────────────────────────────────────────────────────┘

Documentation

Resource	Description
API Reference	Authentication, endpoints, rate limits
Python SDK	SDK guide with async support
LLM Integrations	OpenAI, Anthropic, LangChain examples
Configuration	LLM providers, sessions, memory settings
Testing Guide	API testing with curl and SDK

Roadmap

Feature	Status
Multi-LLM providers (OpenAI, Anthropic, Ollama)	✅ Done
Session/Agent scoping	✅ Done
LLM memory decisions (MERGE)	✅ Done
Redis distributed caching	✅ Done
TypeScript SDK	In Progress
SOC2/HIPAA certification	Planned
Framework integrations (CrewAI, LangGraph)	Planned

Development

# Backend
poetry install && uvicorn app.main:app --reload

# Frontend
cd frontend && npm install && npm run dev

# SDK
cd sdk && pip install -e ".[dev]" && pytest

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

Priority areas: TypeScript SDK, framework integrations (CrewAI, LangGraph, Vercel AI SDK)

Chat Interface

License

Apache 2.0 — See LICENSE