Moltis

What is Moltis?

Moltis is a secure, AI assistant that runs locally on your own hardware. It is a single, Rust-native binary designed for privacy, keeping your keys and data on your machine. Users deploy it to get a full-featured assistant with multi-channel support, voice capabilities, and long-term memory that is always on and local.

Application scenarios

* Private AI assistance: Run a conversational AI agent that processes all data on your personal hardware or server.
* Multi-platform automation: Manage interactions and tasks through integrated platforms like Telegram, WhatsApp, Discord, and Microsoft Teams.
* Secure task automation: Perform web browsing and automation in isolated sandboxes to protect your main system.
* Local model experimentation: Download, set up, and run various local large language models (LLMs) directly.
* Persistent AI agent: Utilize an assistant with long-term memory that remembers context across conversations.
* Voice-enabled interaction: Talk to your AI assistant using built-in voice capabilities.

Main features

* Local & Secure Execution: Your AI runs on your own hardware, with data never leaving your machine and sandboxed execution by default.
* Single Binary Distribution: The tool is one self-contained binary with no runtime dependencies for simple deployment.
* Multi-Channel Support: Access and control your assistant through a Web UI, Telegram, WhatsApp, Discord, Teams, or a direct API.
* Voice Interface: Communicate with your assistant using voice input and output.
* Long-term Memory: The agent uses a hybrid vector and full-text search system to remember context and information.
* Sandboxed Browsing: Run browser automation sessions in isolated Docker containers for safer interaction with the web.
* Local LLM Management: Run your own models locally with automatic download and setup included.
* Skills & Extensibility: Use plugins, hooks, and MCP (Model Context Protocol) tool servers; the agent can even create its own skills at runtime.
* Flexible Isolation: Choose between full filesystem access or per-session isolation via Docker, Podman, Apple Container, or WASM.
* Streaming-first Responses: Get instant, token-by-token streaming replies for a smooth conversational experience.

Target users

This tool is built for individuals and teams who prioritize data privacy and control, such as tech enthusiasts, developers, and privacy-conscious professionals. It suits anyone wanting to self-host a powerful, always-on AI assistant on their own hardware, from a Raspberry Pi to a dedicated server.

How to use Moltis?

Install the single binary on your machine. The website provides several one-command installation methods, including using a shell script, Homebrew, Docker, or installing from source via Cargo. After installation, you typically access the Web UI at http://localhost:13131 to complete setup. You can then configure your AI models and connect to supported channels like Telegram or Discord.

Effect review

Moltis presents a compelling proposition for the self-hosted AI space by bundling a extensive feature set—from multi-platform chat integration to voice and memory—into a single, security-focused binary. The emphasis on sandboxing, local execution, and protection against threats like supply-chain attacks addresses core concerns for users deploying AI on private infrastructure. For the user who wants a full-featured, always-available digital assistant without sending data to third-party servers, Moltis's integrated design and local-first architecture offer a robust and private alternative to cloud-based services.