Private AI Memory Platform — self-hosted, fast, cost-efficient
Apache 2.0Free ForeverSelf-Hosted

The Private
AI Memory Platform.

Your AI. Your hardware. Your rules.

Run locally in one command:

git clone && docker compose up

AI should be private. Not as a feature — as the foundation.

Every time you ask an AI a question, your data travels to a distant data center, gets processed on someone else's hardware, and a record of your query lives on infrastructure you'll never audit. For individuals, that means your personal thoughts, health records, financial details, and private conversations are one breach away from exposure. For enterprises, it means trade secrets, customer data, and proprietary research flowing through third-party systems with opaque data-retention policies.

We're building Mem-Dog to end that trade-off. A complete AI memory platform — 42 specialized agents, semantic search, a temporal knowledge graph, 300+ integrations — that runs entirely on hardware you control. Your laptop, a Mac Mini on your desk, or a server rack in your office. The AI never phones home because there's no home to phone.

Speed through smarter architecture. Instead of routing every request to a massive cloud model, Mem-Dog uses a 6-layer classification pipeline that matches each piece of data to the smallest model that can handle it well. A simple text note gets a fast 4-billion-parameter model. A complex document gets a 27-billion-parameter one. Multimodal content gets a vision model. This tiered routing means most queries finish in milliseconds on modest hardware — because you're not waiting for a 200B model to parse a grocery list.

Cost efficiency through local inference. Cloud AI bills grow linearly with usage — every token costs money. With local models running on Ollama, the marginal cost of a query is the electricity to run it. For individuals, that means unlimited personal AI for the price of a home server. For enterprises, it means predictable infrastructure costs that don't scale with headcount or query volume. No per-seat licensing. No surprise invoices.

Intelligence through better data processing. Privacy doesn't have to mean dumb. Mem-Dog's 42 agents work in concert — classifying, extracting entities, generating embeddings, building knowledge graphs, and creating AI-powered viewpoints — all locally. The result is a system that gets smarter the more you use it, building a private knowledge base that understands relationships between your data, tracks how facts change over time, and surfaces insights through multi-signal search that combines vector similarity, keyword matching, and graph traversal.

Private by Design
Data never leaves your network
Fast Locally
Tiered models, millisecond routing
Cost Efficient
No per-token fees, ever
Genuinely Smart
42 agents, knowledge graph, RAG

How it works

Data flows from any source through a real-time AI pipeline into a private, queryable data layer — all automatic, all configurable.

1/7

Data Sources

300+ apps

WhatsAppSlackEmailTelegramDiscordNango OAuthFilesWebhooks (whk_*)
2/7

Ingest Pipeline

Normalize & route

Per-User WebhooksUniversal EnvelopeNATS StreamingNango Credentials
3/7

AI Processing

42 agents

6-Layer ClassificationSmart Model RoutingViewpoint GenerationEmbedding Creation
4/7

Memory Layer

10 memory types

Versioned Storagepgvector EmbeddingsPer-User ScopingEncrypted Credentials
5/7

Query & Search

Semantic RAG

Natural LanguageCosine SimilarityInline CitationsDigiMe Agent
6/7

DigiMe (OpenClaw)

Chat agent

15+ ChannelsWhatsApp · Slack · TelegramConversational RAGIngest via Chat
7/7

MCP Server

Agent access

SSE TransportClaude DesktopCursor Integration8 Tools

System Architecture

300+ Apps
Slack, WhatsApp, Email...
Web UI (Next.js)
Upload, Chat, Playground
Webhook Gateway
Normalize → Route
API (FastAPI)
Ingest, Query, Manage
NATS Pipeline
42 AI Agents
Supabase
Postgres + pgvector
Neo4j
Graphiti KG
Search + RAG
Vector · FTS · Graph · Hybrid
DigiMe (OpenClaw)
WhatsApp · Slack · 15+ channels
MCP Server
Claude Desktop · Cursor

Fully Configurable AI

Bring your own models, tune every parameter

Model Garden
10+ providers
Smart Routing
Per data-type
Agent Configs
Custom prompts
Processing Flags
Toggle per type

Built for every use case

From personal knowledge management to enterprise compliance — one platform, infinite possibilities.

Personal Knowledge Base

Capture from WhatsApp, email, Slack. Semantic search across everything. Never lose a conversation or idea.

Team Memory

Shared org memory across channels. Auto-classify meetings, decisions, action items. Queryable by anyone.

Customer Intelligence

Ingest support tickets, CRM, chat logs. AI extracts sentiment and trends. 300+ app connections via Nango.

Research & Analysis

Ingest PDFs, papers, web pages, datasets. AI viewpoints and summaries. Semantic connections across sources.

Compliance & Audit

Every mutation versioned. OpenTelemetry tracing. Per-item ACLs. Immutable audit trail.

IoT & Sensor Data

GPS, biometric, weather, industrial sensors. Specialized agents for time-series and geospatial.

Legal & Contract Intelligence

Ingest contracts, NDAs, legal briefs. AI extracts clauses, obligations, deadlines. Temporal graph tracks amendments over time.

Healthcare & Clinical Notes

Process medical records, imaging reports, lab results. DICOM-aware agents. Knowledge graph links patients, conditions, treatments.

Education & Training

Ingest lectures, textbooks, course materials. AI generates study guides and flashcards. Students query their learning history.

Sales Enablement

Connect Salesforce, HubSpot, email. AI summarizes deal history, extracts action items. Search across all customer touchpoints.

Media Monitoring

RSS feeds, social media, news APIs. Real-time sentiment analysis. Track brand mentions across channels with temporal trends.

Real Estate & Property

Ingest listings, inspections, contracts. AI extracts property details, comparables. Knowledge graph links properties, agents, transactions.

Meeting Intelligence

Zoom, Teams, Google Meet recordings. AI transcribes, extracts decisions, action items. Searchable meeting memory across your org.

Visual Asset Management

Ingest photos, videos, design files. Multimodal AI describes and tags visuals. Search images by description or concept.

Knowledge Management

Connect Notion, Confluence, Google Docs. AI indexes and cross-references documentation. Ask questions across all your wikis.

Travel & Logistics

Ingest itineraries, shipment data, GPS tracks. Geospatial agents plot routes. Temporal graph tracks delivery timelines and delays.

Flexible. Powerful. Complete.

Everything you need to build a private AI memory platform for your data.

Versioned StorageAI Enrichment300+ AppsDigiMe AgentReal-Time PipelineSemantic SearchSecure by Default
Powered by OpenClaw Runtime

Meet DigiMe

Your AI memory assistant that lives inside your messaging apps. Built on the OpenClaw agent runtime — ask questions, search your knowledge base, and ingest data through natural conversation. No context switching, no extra tools.

Talk to your data from anywhere

Lives in WhatsApp, Telegram, Signal, Slack, Discord, Matrix, and 15+ more channels
Natural language queries against the full mem-dog RAG system with cited answers
Semantic search across all ingested data — vector, keyword, hybrid, and graph modes
Ingest new data directly from conversations — forward a message to save it
Retrieve and summarize memories, timelines, and facts on demand
Runs on your infrastructure — your conversations never leave your network
DigiMe · WhatsApp
What came out of last week's standup?
Here's a summary from the 3 standups last week: • **Mon** — API rate-limiting merged, deploy scheduled for Wed • **Wed** — Deploy completed, 2 hotfixes pushed same day • **Fri** — Sprint retro: velocity up 15%, carried over 1 ticket
[1][2][3]
Which ticket was carried over?
**MEM-247** — "Add batch ingest endpoint." Blocked on schema review; moved to next sprint with high priority.

Built on OpenClaw

Open-source AI agent runtime

15+ Channels
WhatsApp, Slack, Telegram...
RAG-Powered
Cited answers from your data
Real-Time
Instant responses via webhook
Self-Hosted
Runs alongside mem-dog

Built for developers

SDKs in 5 languages, agent framework adapters, and a full REST API. Get started in minutes.

Quick Start
$ git clone https://github.com/BuildGeekAI/mem-dog
$ cd mem-dog && docker compose up
# UI: localhost:3000 | API: localhost:8080
# Neo4j: localhost:7474 | 3 Ollama tiers

Developer Highlights

SDKs: Python, TypeScript, Go, Rust, Ruby — full API coverage
Adapters: LangChain, CrewAI, OpenAI Agents SDK
Graph Memory: Temporal knowledge graph with Neo4j + Graphiti
Memory Compression: LLM-powered summarization for long-term recall
MCP Server: 8 tools for Claude Desktop, Cursor, and more

Deploy anywhere

Production-grade infrastructure with real-time processing, tenant isolation, and complete observability. Run locally or on Google Cloud.

0
Apps
15 categories
0
AI Agents
6-layer routing pipeline
0
Model Tiers
Small to multimodal
0
Memory Types
Timeline to semantic
0
API Endpoints
Full REST coverage

How Mem-Dog compares

Each tool does one thing well. Mem-Dog combines all of them — integrations, AI processing, memory, search, and a knowledge graph — into a single self-hosted system.

Feature
Mem-Dog
Dify.ai
Mem0
Zep
LangMem
Core Architecture
Runs fully on-premise / at home
Self-hosted AI models (Ollama)
Near-zero marginal cost per query
Data never leaves your network
Single docker compose up
Data & Integrations
300+ app integrations (Nango-powered)
Per-user webhook endpoints (whk_<ulid>)
Multi-channel ingest (WhatsApp, Slack, etc.)
File/image/video/audio ingest
Versioned storage with full history
Automatic OAuth token refresh
AI & Intelligence
42 specialized AI agents
5 model tiers with smart routing
Knowledge graph (Neo4j/Graphiti)
Temporal fact tracking (valid_at/invalid_at)
RAG chat with inline citations
Model Garden (BYO providers)
Memory & Search
10 memory types
5 search modes (vector, FTS, hybrid, graph, full)
Memory compression (LLM summarization)
Per-user scoping & ACLs
Memory expiry with configurable TTL
Developer Experience
Python SDK with adapters (LangChain, CrewAI)
Multi-language SDKs (TS, Go, Rust, Ruby)
Built-in conversational agent (DigiMe)
Interactive playground in UI
OpenTelemetry distributed tracing

Dify.ai

Low-code LLM app builder

Strength

Beautiful drag-and-drop workflow builder for creating AI applications. Good for prototyping LLM chains quickly. Self-hostable with Docker.

Where Mem-Dog goes further

No persistent memory layer, no multi-channel ingest, no webhook pipeline, no knowledge graph, no per-user webhook endpoints, no temporal reasoning.

Mem0

Memory layer for AI agents

Strength

Clean API for adding long-term memory to LLM applications. Good conversation and user memory primitives. Growing open-source community.

Where Mem-Dog goes further

Cloud-only for full features, limited to 4 memory categories, no data ingest pipeline, no 300+ app integrations, no AI enrichment agents, no knowledge graph, no versioned storage.

Zep

Long-term memory for AI assistants

Strength

Strong temporal knowledge graph with Graphiti fact extraction. Good at tracking how facts change over time. Reranking and triple search.

Where Mem-Dog goes further

Cloud-managed only, no self-hosting, no data ingest pipeline, no app integrations, no AI enrichment, no webhook endpoints, no built-in UI or playground.

LangMem

Memory management for LangChain agents

Strength

Native LangGraph integration. Good primitives for thread-level and cross-thread memory. Backed by LangChain ecosystem.

Where Mem-Dog goes further

LangChain-only, no multi-channel ingest, no app integrations, no knowledge graph, no versioned storage, no search modes beyond vector, no AI enrichment pipeline.

Nango connects your apps. Dify builds AI workflows. Mem0 adds memory. Zep tracks facts. Mem-Dog does all of it — on hardware you own, at a fraction of the cost.

Frequently Asked Questions

What hardware do I need to run mem-dog?
A laptop or any machine with Docker. For production, a Mac Mini (M2/M4) with 16 GB RAM runs the full stack including local AI models. For cloud, any GKE/EKS/AKS cluster works.
Is mem-dog free and open source?
Yes. Mem-dog is Apache 2.0 licensed — free forever, no usage limits, no telemetry. Self-host on your own hardware or cloud. No vendor lock-in.
How is this different from ChatGPT memory or Notion AI?
ChatGPT memory is locked inside OpenAI. Notion AI only works with Notion data. Mem-dog ingests from 300+ apps, runs 42 AI agents on your hardware, and gives you a temporal knowledge graph with 5 search modes. You own everything.
What data formats are supported?
Text, Markdown, HTML, JSON, CSV, PDF, images, audio, video, and binary blobs. The 42-agent pipeline auto-detects the format and routes to the right agent for classification, analysis, and embedding.
Can I use my own AI models?
Yes. Model Garden supports 10+ providers with per-user routing and fallback chains: Ollama (local), Ollama Cloud, Google Gemini, OpenAI, Anthropic, and more. You can run fully offline with local Ollama models.
How do I get started?
Clone the repo and run docker compose up. That starts the full stack: API, UI, PostgreSQL + pgvector, Neo4j, Redis, 3 Ollama instances, and the webhook pipeline. UI at localhost:3000, API at localhost:8080.
Does mem-dog send my data anywhere?
No. There is no telemetry, no analytics, no phone-home. All processing happens on your machine. The AI models run locally via Ollama. External providers (Gemini, OpenAI) are opt-in only.

Ready to try mem-dog?

Get running in under 2 minutes. No account needed.

git clone https://github.com/BuildGeekAI/mem-dog && cd mem-dog && docker compose up

Enterprise inquiries: pagarwal@buildgeek.ai