Your data stays in
Prompts, files, and answers never leave your network or cloud account unless you choose to route them out.
Platform
Like the vendordashboard. Yours.
Each deployment is the same four layers, set up inside your environment. To your people it feels like using the big AI vendors. Underneath, every piece runs on infrastructure you own.
01
The stack
01
Dashboard
Where employees and customers chat with the models, organise projects, and upload files — signed in with the accounts they already have.
02
Gateway
One OpenAI-compatible endpoint that handles keys, rate limits, routing, logging, and usage reports for every app and user.
03
Models
Open-weight models for chat, code, embeddings, vision, and speech, served for high throughput and swapped as better ones are released.
04
Infrastructure
GPU servers on your premises, in your own cloud account, or both — monitored, patched, and sized to your demand.
02
Switching an app over
Apps built on the OpenAI SDK move to your own platform by changing two values. Everything else stays the same.
from openai import OpenAI
client = OpenAI(
base_url="https://ai.yourcompany.com/v1",
api_key="YOUR_GATEWAY_KEY",
)
reply = client.chat.completions.create(
model="your-chat-model",
messages=[{"role": "user", "content": "Summarise this contract"}],
)03
Why it's built this way
Predictable cost
You pay for capacity you own, not per token. Heavy use doesn't send the bill up.
No lock-in
Standard APIs and open-weight models mean you can change models — or providers — without rewriting your apps.
Want the details for your environment? See Services.