Private AI on your own hardware.
iPhone to Mac cluster to private LLM. No cloud APIs. Your data never leaves your network. Deploy AI agents across Apple Silicon nodes connected via Thunderbolt 5 RDMA.
Powered by
Built from scratch. Every layer.
Custom daemon. Custom dashboard. Custom tools. No wrappers, no frameworks, no cloud LLM APIs for inference. Electron, web, and iOS ship today — TUI and widgets are in the same repo.
asmi
The Control Plane
Custom Rust daemon on every node. Cluster telemetry and remote management over HTTP (port 9090) — no SSH required for day-to-day ops.
Learn moreRDMA Mesh
TB5 Tensor Parallelism
Thunderbolt 5 connects Mac nodes for distributed inference. JACCL splits models across nodes for multi-node Apple Silicon serving.
Learn moreModel Pipeline
Hugging Face to Inference
Browse HF Hub → download → MLX convert → quantize → deploy. Indexed local inventory with pluggable storage and auto-discovery.
Learn moreNatural Language Ops
Talk to Your Cluster
Cluster tools with rich preview renderers. Deploy models, check health, inspect processes — all through chat. Auto-heal pipeline included.
Learn moreFully Private
Zero Cloud APIs
Tailscale mesh VPN. Inference and conversations stay on your hardware — no cloud LLM APIs. No vendor lock-in.
Learn moreHermes Agent
37 skills. Zero cloud.
Every r1o install ships with Hermes -- a local AI agent that comes pre-loaded with 37 skills across 17 categories. Documents, GitHub, research, cluster ops, and more. All running on your hardware.
GitHub
6 skills
Apple
3 skills
Documents
3 skills
Creative
2 skills
Research
2 skills
Software Dev
3 skills
Cluster
2 skills
MLOps
1 skill
Fully offline
All 37 skills work without internet. No API keys needed.
Self-expanding
Hermes learns new skills as you use it. Custom skills save to ~/.hermes/skills/.
Skills Hub
Install more from the Skills Hub marketplace anytime. Community-built, locally run.
Zero cloud dependency
Everything runs on your Mac. Your data never leaves your network.