sqlite-vec · vec0
Local RAG Engine
SQLite + sqlite-vec vector search. Every interaction is semantically indexed for O(log n) retrieval. Cross-mode: use a cloud AI for chat while a local model handles embeddings.
Running locally · zero telemetry · GNU AGPL v3
Your mind, unbound.
Your data, sovereign.
Memo is a private, local-first AI second brain. It runs models locally, remembers everything in a vector store, and never phones home. Chat through external APIs when you need more power, or stay 100% offline — your call.
one-click installer · llama.cpp bundled · no terminal required
or install with one command — macOS & Linux
29
Go Packages
8
Agent Tools
15+
Slash Commands
v3.1.2-beta
Latest Version
the premiseMost AI assistants send your thinking to someone else's servers.
Most AI assistants send your thinking to someone else's servers.
Memo doesn't.
01
Private by construction
Zero telemetry, no training on your chats, no cloud dependency. Optional encrypted backup only if you turn it on.
02
Real memory, not a longer prompt
Every interaction is embedded and indexed in a local vector store. Relevant context is retrieved automatically on each turn.
03
Best of both worlds
Run chat through a powerful external API while a tiny local model handles embeddings — or stay 100% offline. Your call, per model.
04
Native, not a web wrapper
A real Flutter desktop app on Windows, Linux, and macOS, with a mobile companion. It feels like software you own.
what's inside
Nine engines, one quiet surface.
A real RAG memory and a tool-calling agent, paired with an interface a first-time user can navigate.
12+ tools · sandboxed
Agent Engine
12+ built-in tools with a sandboxed execution pipeline. File read/write, shell commands, web search, symlink protection, per-tool timeouts, and a 6-policy permission system.
15 cmds · task loop
Terminal CLI & Task Loop
A Claude Code-style terminal chat with 15+ slash commands, persistent sessions, and an autonomous CEO/Worker task loop. Permission prompts surface inline — no need to leave the terminal.
multi-model · parallel
Orchestra Mode
A chief model plans the task, 8 specialist roles execute in parallel, and the chief synthesizes the result. Assign different models per role — Claude for reasoning, Gemini for speed.
7 providers · fallback
7 External Providers
OpenAI, Gemini, Claude, Grok, Groq, OpenRouter, and Ollama — with automatic fallback chain. Switch mid-conversation with /model. API keys encrypted with AES-256-GCM.
GPU offload · bundled
Bundled llama.cpp
Fully managed server lifecycle: auto-download, GPU offloading (NVIDIA/AMD/Metal), VRAM detection. No Docker, no terminal — everything from one native desktop app.
whatsmeow · QR pair
WhatsApp Bridge
QR pairing with whatsmeow. Read, search, and reply to messages. AI drafts responses, summarizes threads. Profile photos, group names, auto-reconnect. Everything stays local.
AES-256-GCM · E2E
Cloud Sync & Backup
E2E-encrypted Google Drive sync (AES-256-GCM, PBKDF2). Auto-sync every N messages. Full .memo archive export. Double-confirm wipe. Your data is encrypted before it leaves your machine.
LAN · ngrok · Tailscale
Mobile & Remote
Thin Flutter companion for Android/iOS over LAN or built-in ngrok tunnel. Built-in Tailscale tunnel for private networking. Access your brain from anywhere with token auth.
under the hood
Your words become memory. Memory becomes context.
Every message flows through a six-stage pipeline — embedding, search, context construction, and response — all on your machine, in milliseconds.
💬01
You ask
Typed, spoken, or from WhatsApp.
🧮02
Embed
Converted into a 768-dimension vector by a local embedding model.
🔍03
Search
Cosine similarity against thousands of memories. Top‑K in milliseconds.
🧩04
Inject
Relevant context silently placed in the system prompt.
✨05
Generate
LLM streams back an informed response via SSE.
♻️06
Persist
Message and response asynchronously embedded. Memory grows.
built for
One brain. Many workflows.
Memo adapts to how you work — whether you are shipping code, writing a thesis, or just want an AI that respects your privacy.
Software Developer
Drop in your codebase. Ask the agent to refactor modules, write tests, or explain a complex function. Orchestra mode splits work across specialist models — one writes React, another writes Go. All offline.
Agent modeOrchestraCode gen
Researcher & Writer
Upload papers, notes, and references. RAG memory connects ideas across weeks of work. Ask "what was that citation about reinforcement learning?" and get the exact paragraph — no keyword search needed.
RAG memoryMultimodalSearch
Privacy-Conscious User
Your conversations, files, and memories never leave your machine. No account, no cloud dependency. Use external APIs when you need more power — keys are AES-256 encrypted on disk. You hold all the cards.
Zero telemetryE2E syncOffline
no terminal · no cloningBring your second brain
Bring your second brain
home.
One click installs Memo with llama.cpp bundled. Open the app, pick a model from Discover, and start chatting.