Open Source AI Infrastructure

Build AI Agents That Think, Remember & Act

State-of-the-art memory engine, autonomous execution, and multi-agent coordination. Enterprise-ready infrastructure from Germany.

Ecosystem

Four Services. One Platform.

Each service works independently or together as a unified AI infrastructure stack.

OpenAlice

Autonomous AI assistant with SOTA memory engine, 135+ tools, knowledge graph, and mission execution. Connects to Telegram, Discord, and any platform.

OpenAlice ML

Embeddings (BGE-Large 1024d), cross-encoder reranking, and zero-shot NER. One API for all your ML needs.

OpenAlice A2A

Agent-to-Agent protocol for multi-agent coordination. Your agents talk to each other, share context, and collaborate.

OpenAlice Persona

AI personality shell. Define your agent's character, voice, expertise, and communication style. Tool-driven, model-agnostic.

Built for production. Designed for scale.

Hybrid architecture combining Mem0, Hindsight, and Graphiti. HNSW vector search, knowledge graph, temporal dedup, RLHF feedback loop.

GDPR-compliant. HSTS, CSP, rate limiting, HITL approvals, safety gates, kill switch. Gov-level security from day one.

Written in Rust. Sub-millisecond memory retrieval. 1,330+ tests. Handles thousands of concurrent users.

Hosted on Hetzner (Frankfurt). EU data residency. German engineering, European values.

A2A protocol for agent coordination. Missions with OODA loop. Your agents work together autonomously.

Rank system with RLHF. Alice evaluates herself, learns from mistakes, and gets better with every conversation.

Start with the ML API. Add Alice when you need autonomy. Scale with the full platform.