Houdry is two products that work together. They are not the same install.
On the GPU
Install Fabric
houdry servehoudry gpu register
Ollama running, with at least one model pulled.
On the desk
Install Agent
Point it at http://HOST:8080/v1
Do not install Fabric on every laptop.
Fabric owns the GPUs and the models. Agent owns the people and the files. There is no cloud LLM API in the path.
Install Fabric on the machine with the GPU. Install Agent on the desk. Point Agent at http://HOST:8080/v1. Do not put Fabric on every laptop.
Which page should I open?
- Ops / GPU owner → Installation, then keep
houdry serveandhoudry gpu registerrunning. - Everyone else on the desk → Installation (Agent only), then Houdry Agent.
- First time, both machines in front of you → Quickstart.
- How the pieces connect → Architecture.
- Chat failed → Troubleshooting.
How they connect
On the desk
Houdry Agent
Chat, Knowledge, SOPs, files, and local tools.
Talks to Fabric at http://HOST:8080/v1
On the GPU
Houdry Fabric
houdry serve on port 8080 is the control plane.
houdry gpu register runs Ollama and returns the answer.
People never talk to the GPU process. They talk to Agent. Agent talks to Fabric. Fabric picks a model on a READY GPU.
What you can do once it is connected
- Ask questions against SOPs, inspection reports, and manuals that stay on this computer.
- Draft approval notes and registers from files that never leave the premises.
- Send drawings / scans. Fabric routes vision work to a vision-capable local model when one is loaded.
- Add another GPU workstation with
houdry gpu register. Agent does not need a reinstall.