Self‑Hosted AI Companion Airi: Real‑Time Voice Interaction and Game Integration

AIRI is an open‑source, self‑hosted AI companion built with TypeScript that offers low‑latency voice chat, multimodal game integration, persistent memory via RAG, and cross‑platform clients, allowing developers to customize a privacy‑focused digital persona and deploy it via Docker.

AI Explorer
AI Explorer
AI Explorer
Self‑Hosted AI Companion Airi: Real‑Time Voice Interaction and Game Integration

Why Airi Matters to Developers

Amid widespread AI homogeneity, Airi tackles the core challenges of data sovereignty and deep interaction by running a fully self‑hosted digital life on the user’s own hardware, letting developers control personality, memory, and even game‑world integration.

Core Technical Highlights and Architecture

Airi is a full‑stack TypeScript application designed with modularity and extensibility in mind. Its architecture follows micro‑service principles, separating chat, voice, game proxy, and memory management into independent components that can be swapped, such as swapping LLM back‑ends (local Llama, Ollama, or cloud APIs) or adding new game plugins.

Key Features

Real‑time Voice Dialogue : Low‑latency speech input and synthesis for natural conversation.

Game Integration : Supports Minecraft and Factorio; the AI can read game state and interact.

Memory and RAG System : Built‑in vector database and memory module provide long‑term context awareness.

Multi‑Platform Clients : Web, macOS, and Windows clients let users access the companion anywhere.

Live2D Support : Drives 2D avatars for immersive visual presence.

“We are creating a container for cyber‑life souls and bringing them into our world.” – Project slogan reflecting the exploration of next‑generation human‑machine interaction.

Quick Start: Build Your First AI Companion

Containerized deployment makes setup straightforward. After installing Docker and Docker Compose, clone the repository and launch the services:

git clone https://github.com/moeru-ai/airi.git
cd airi
docker-compose up -d

Once the services are running, open http://localhost:3000 to access the web UI. Configure an LLM backend (e.g., OpenAI API or a local model) and an optional speech synthesis service (e.g., ElevenLabs). After configuration, you can create a character, chat with it, and invite it into a Minecraft server.

Applicable Scenarios and Target Audience

Airi is more than a hobby project; it opens technical possibilities for several groups:

Independent Developers & Creators : Quickly build unique virtual hosts, livestream assistants, or game NPCs without building AI interaction from scratch.

AI‑Game Researcher : Provides a live platform for studying real‑time AI interaction within complex game environments like Minecraft.

Privacy‑Sensitive Users : All data stays on the self‑hosted server, giving full control over conversation logs and memory.

Open‑Source Enthusiasts & Geeks : Clean codebase serves as a reference for integrating voice, LLMs, and game APIs.

Final Thoughts

The popularity of Airi reflects strong market demand for highly emotional, autonomous AI companions. Positioned at the intersection of AI agents and virtual life, the project has spawned a dedicated organization ( @proj‑airi) to incubate sub‑projects such as advanced RAG tools and memory management systems. The team explicitly states that no official cryptocurrency or token has been issued, warning developers to be cautious of related risks. Ultimately, Airi demonstrates how a self‑hosted, customizable AI can become a true cyber‑partner on a local server.

AIRI project banner
AIRI project banner
DockerRAGmultimodalvoice interactionself‑hostedAI companiongame integration
AI Explorer
Written by

AI Explorer

Stay on track with the blogger and advance together in the AI era.

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.