About HorniLab
HorniLab is a digital-native startup building an embodied AI platform in Minecraft. We use a persistent multiplayer 3D world to train, evaluate, and deploy autonomous agents with memory, telemetry, and human interaction.
The platform includes HorniMine, Oneiro, memory and telemetry systems, and internal AI tooling used to build and operate the environment itself.
Minecraft gives us a persistent, socially inhabited 3D world with terrain, physics, inventory, construction, and real humans in the loop, without needing to build a simulator from scratch.
What We Build
- HorniMine — A live multiplayer Minecraft world used as the public environment and experiment surface for the platform.
- Oneiro — An embodied AI agent prototype combining visual input, reflex loops, planning, memory, and autonomous action.
- Memory and telemetry — Persistent logs, vector memory, decision history, and experiment traces for long-horizon evaluation.
- Internal AI tooling — Build agents and automation that help generate code, maintain infrastructure, and accelerate iteration.
Who It Is For
- Embodied AI researchers — exploring spatial reasoning, memory, autonomy, and long-horizon agent behavior.
- Agent builders — developing orchestration systems, tooling, and long-running AI workflows in live environments.
- Experimental communities — interested in public-world AI experiments and human-in-the-loop testing.
Current Stage
- Public website and live domain
- Authentication and account system
- Persistent Minecraft world and active infrastructure
- Published Oneiro architecture and dataset design
- Internal AI agents used in development
Business Model
- Premium access and memberships — supporter tiers, early access, and premium community entry.
- AI infrastructure and tooling — reusable agent systems, orchestration patterns, and hosted components.
- Custom identity and digital goods — personalized textures, cosmetics, and premium social/world features.
Why Google Cloud Matters
- Embodied AI dataset generation — render frames, actions, and world state stored as structured training data.
- Model training and serving — scalable compute for low-latency inference and repeatable evaluation.
- Persistent memory and telemetry — long-horizon storage and replay for agents and experiments.
- Live experiment infrastructure — scalable systems for public embodied AI experimentation.
Team
A small independent team building a public embodied AI platform.
- Nikita (Cokeef) — Founder · infrastructure, product, platform engineering.
- Arseniy (Halva) — Co-Founder · AI systems and engineering.
- Andrey (Arhitector) — world design, narrative systems, server identity.
Internal AI Team Members
- Oneiro — Minecraft embodied agent for visual input, spatial reasoning, and autonomous behavior research.
- Coder & Designer — OpenClaw build agents that produce code, assets, and internal tools.
Contact
For collaboration inquiries: ip@horni.cc
Platform overview: horni.cc/platform
Discord: discord.gg/Zw3tQkCSZN
Telegram: t.me/HorniMine