ALF OS
VerifiedSelf-hosted Go-based AI assistant for Telegram with semantic memory and flexible LLM backends.
What is ALF OS?
ALF OS is an open-source personal AI agent built in Go that operates inside Docker on user-owned hardware. It links a chosen LLM to Telegram messaging and layers on persistent memory, voice handling, and configurable access levels managed through a browser interface.
The architecture separates concerns into capability execution, memory storage, AI decision-making, sandboxed security, and runtime orchestration. This design enables mixing backends per response tier while enforcing isolation between processes and restricting permissions for safety.
It targets users who want a private, subscription-friendly or local AI companion that avoids JavaScript-heavy frameworks and keeps data on their own machines rather than third-party servers.
Capabilities
What you can build with ALF OS
Daily chat companion
Maintain ongoing conversations in Telegram where the agent recalls past details across sessions using vector search.
Tiered task handling
Route simple queries to lighter models and complex requests to stronger ones based on an automatic classifier.
Voice-enabled notes
Transcribe spoken messages and store them with semantic indexing for later retrieval through the dashboard.
Install ALF OS
curl -fsSL install.alfos.ai | shcurl -fsSL install.alfos.ai | sh- 1Run the one-line installer script to fetch the Docker setup.
- 2Execute alf init and follow the prompts for directory, Telegram token, and dashboard settings.
- 3Start the container and open the Control Center in a browser to select a backend and tiers.
- 4Authenticate the chosen LLM provider if using a CLI or API option.
- 5Send a test message to the Telegram bot to verify the connection.
Works with
ALF OS: pros & cons
Pros
- +Single static Go binary reduces dependencies and potential vulnerabilities compared to Node.js stacks.
- +Semantic memory uses local ONNX embeddings and SQLite for recall without external services.
- +Flexible backend support lets users mix subscription CLIs, APIs, or local Ollama models per tier.
- +Strong process isolation and non-root execution provide defense-in-depth security.
Cons
- –Requires Docker and at least 2 GB RAM, limiting use on very lightweight devices.
- –Initial setup involves multiple configuration steps including Telegram bot creation.
- –Voice transcription depends on a separate sidecar container that must be managed.
Frequently asked questions
Only if using a cloud backend; local Ollama models work completely offline after initial setup.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other agent frameworks options worth comparing.
