Open source · Container-isolated · Private

Your AI forgets you
every single time.

Lethe is an AI assistant with persistent memory. It runs on your machine with any cloud API or a local LLM, talks to you via Telegram, and remembers everything. Runs in an isolated container with access only to directories you choose.

Get Started View Source ↗
// what this feels like
Memory changes everything.
Persistent, associative, always-on memory — across every session. Skills it learns are never forgotten. Here's what that looks like:
3 weeks since last mention
you: how's the thing going
lethe: The Springer resubmission? You uploaded the revised manuscript on the 2nd but I never saw a confirmation email come through. Want me to check the submission portal?
learned once, never forgets
you: check my university emails
lethe: [recalls Graph API skill note, runs unimail list]
You've got 3 new ones since yesterday. Nathalie sent the defense schedule, and there's a quarantine notice from IT.
noticed on its own
lethe: You haven't mentioned the visa renewal since March 12. The window closes in 9 days. Want me to draft the email to the consulate, or are you already on it?
// architecture
A cognitive system, modeled on the brain.
Specialized actors that think, remember, and act independently — coordinated through structured message channels.
cortex
Executive layer. Your direct interface. Handles tasks, delegates work, manages context.
hippocampus
Associative recall. Searches notes, archival memory, and conversations. Surfaces what's relevant before you ask.
dmn
Default Mode Network. Scans goals in the background, connects dots, catches what you've missed.
01:24:18 dmn background round complete. discovered new repo, updated project tracking
01:24:19 hippocampus recall triggered. found 2 notes + 3 conversation matches
01:24:20 cortex processing message. context loaded, calling bash → read_file → responding
01:24:21 notes auto-extract. saved skill: "Deploy API via Graph endpoint"
01:24:22 brainstem heartbeat. all systems nominal
// principles
Why it works this way.

🏠 Your machine, your data

Runs in an isolated container on macOS and Linux. It can only access directories you explicitly allow. Memory, conversations, files — everything stays on your machine. Works with any cloud API or a local model.

🧠 Memory that compounds

Not just chat history. Lethe builds structured knowledge — skills it learns, conventions you set, procedures it discovers. The longer you use it, the less you repeat yourself.

🔧 Fully open, fully yours

MIT licensed. Every prompt, every memory decision, every tool call is in the source. Customize the persona, swap the model, rewrite the rules. No black boxes.

// get started
Two minutes to memory.
1

Install

One command. Works on macOS and Linux.

curl -fsSL https://lethe.gg/install | bash
2

Say hello

Open Telegram and message your bot. Lethe remembers from here on out.

// you'll need

1

Build llama.cpp

git clone https://github.com/ggml-org/llama.cpp && cd llama.cpp
cmake -B build -DGGML_CUDA=ON && cmake --build build -j$(nproc)
2

Start the model server

Download a Gemma 4 31B GGUF and run:

llama-server --model gemma-4-31B-it-Q8_0.gguf \
  --split-mode tensor --jinja --reasoning-budget 4096 \
  --ctx-size 98304 --parallel 2 --flash-attn on -fit off
3

Install Lethe & configure

curl -fsSL https://lethe.gg/install | bash

# then set in .env:
LLM_PROVIDER=openai
LLM_API_BASE=http://localhost:8090/v1
OPENAI_API_KEY=local

// you'll need