Hosting and AI inference
Cortex
We put your apps and AI models online and maintain the servers.
This page is for technical teams. Cortex is the infrastructure we run in Trois-Rivières, Québec, and it runs our own products. We can also host your apps and AI models on it, on request.
For now, Cortex hosts our own products and the software of Constellation Heating, a cofounder’s company. We can put your project online on request: write to us to talk it through. Prices are not set.
Why
You want your apps online. You do not want to run the servers.
A big cloud gets you started fast, then the bill, the quotas and the proprietary services grow with you. Your own servers give you control, but someone has to own networking, storage, upgrades, security and the pager. And AI APIs send your data to someone else’s model.
You hand us an app or a model, we put it online on your domain and operate everything underneath. A self-service console and Kubernetes access are planned for later.
What you get
- 01
Host any app
Bring a container image, its configuration and health checks. We deploy it and serve it on your domain.
- 02
AI models on our machines
Your models run on our NVIDIA DGX Spark and AMD Strix Halo machines in Trois-Rivières. Nothing is sent to a third-party API.
- 03
Kubernetes, later
Planned: an isolated environment with its own policies and Kubernetes access. Tell us if you need it.
- 04
Any compute node can fail without losing data
Compute nodes keep no data, so any one can fail without losing anything. The others carry on and we replace it.
- 05
Hosted in Québec
Apps you host on Cortex run on our machines in Trois-Rivières. Backups and some files are kept with an outside provider, named in our privacy policy.
- 06
A team in Trois-Rivières
You talk directly to the team that runs the infrastructure.
How it works
- 01
Write to us
Describe your app or your model, and what it needs.
- 02
Bring an image
Your app or your model server, as a container, with its configuration and health checks.
- 03
Go live
We deploy it and attach your domain.
- 04
Run
We watch the infrastructure and replace failed hardware.
What changes
Your team builds features. We run the servers.
- No servers, networking or upgrades for your team to maintain.
- Your AI runs next to your data, not through someone else’s API.
- One team to reach, in Trois-Rivières.
At a glance
Cortex
- What we host
- Apps, websites and AI inference, as containers
- Location
- Trois-Rivières, Québec
- Redundancy
- Stateless compute: any node can fail without losing data. We replace failed hardware.
- AI hardware
- NVIDIA DGX Spark, AMD Ryzen AI Max (Strix Halo)
- Compute
- Stateless ARM64 nodes on Talos Linux
- Orchestration
- Kubernetes
- Access
- Set up by our team. Web console, API and Kubernetes access planned.
Interested in Cortex?
Describe your app or your model and what it needs. We will tell you whether Cortex can host it.