Index / LettaBot · updated Oct 24, 2026

LettaBot

letta-ai/lettabot · abandoned · rank 39 of 50 by stars

A multi-channel personal AI assistant with unified persistent memory across Telegram, Slack, Discord, WhatsApp, and Signal, built on the Letta platform. Now officially deprecated and archived in favor of Letta Code's Channels and Remote Environments.

#deprecated#multi-channel#persistent-memory#chatbot#letta
Compare vs OpenClaw
GitHub ↗

Facts

Repository

Contributors23
Open issues84
Last commitMay 3, 2026
Release cadence~3 days
Latest releasev0.2.0-alpha.12

Runtime

LanguageTypeScript
Memory80 MB
Boot time200 ms
Deploymentself-hosted · cloud
Setup difficultyMedium
Plugin ecosystemLimited

Posture

LicenseApache-2.0
Local-firstNo
Cloud dependencyRequired
Multi-userYes
Privacy postureMixed

Community

Sentiment45% positive
Reddit mentions0
Web results10

Security breakdown

Composite 55 / 100 · how these are scored

Sandboxing3 / 10

higher is safer

API security6 / 10

higher is safer

Network isolation4 / 10

higher is safer

Telemetry safety5 / 10

higher is safer

Shell access risk8 / 10

higher is riskier

Model access

Read from the repository, not written by a model · 60 files examined

2 providers · custom endpoint

Direct
MistralOpenAI
Compatible
OpenAI-compatible

Pinned models

claude-sonnet-4-20250514 released 2025-05-14 — 18 months old (dated in the model id)

Pin last edited 9 months ago

Evidence

Decision

Why choose LettaBot over OpenClaw?

Why choose this

  • Unified persistent memory across five chat platforms via Letta's memory architecture
  • Built-in scheduling, heartbeats, and voice transcription out of the box
  • One-click Railway deployment template

Tradeoffs

  • Officially deprecated and archived — no future development
  • Requires Letta cloud API key or self-hosted Letta server, unlike OpenClaw's local-first design
  • Smaller community (326 stars) with zero Reddit presence

Best fit

  • Existing deployments that still work and need short-term continuity
  • Developers studying a hackable multi-channel agent reference implementation
  • Letta users evaluating migration to Channels + Remote Environments

Avoid if

  • You are starting a new project (officially deprecated)
  • You need ongoing maintenance, security patches, or new features
  • You want a local-first agent without a Letta cloud/server dependency
Good Confidence78%

Evidence is strong on deprecation status (explicit README warning and archive notice) but thin on community sentiment, with zero Reddit matches and only directory-listing web mentions. Security and performance metrics are estimates from the stack, not measured data.

AI layer reviewed Oct 5, 2026 · how this is written

Star activity

326 stars today

Overview

LettaBot was a personal AI assistant built on the Letta Code SDK that unified conversations across Telegram, Slack, Discord, WhatsApp, and Signal into a single agent with persistent, cross-session memory. It also supported Bluesky Jetstream feed ingestion, voice message transcription via Whisper, heartbeat check-ins, agent-created reminders and recurring tasks, and local tool execution (file reads, code search, shell commands) on the host machine.

Architecturally, it acted as a multi-channel gateway in front of the Letta platform, requiring either a Letta cloud API key or a self-hosted Letta Docker server. Configuration was YAML-driven, and the codebase was explicitly designed to be 'hackable' by AI coding agents. Deployment was streamlined via a Railway template.

The project is now officially deprecated and archived. Letta directs all users to Letta Code's Channels (for messaging), Remote Environments (for always-on agents via outbound WebSocket), and letta cron (for scheduling), with a stated ~20-minute migration path. Compared to OpenClaw, LettaBot offered stronger managed-memory infrastructure but weaker local-first posture, a smaller ecosystem, and — decisively — no ongoing maintenance. New users should not adopt it.

Nominate a clone

Add a new Claw

Paste a GitHub repository and tell us why it belongs on the tracker.

Opens a prefilled issue on GitHub — every nomination is public. Comfortable with a PR? Adding the repo to projects.json is faster.