Infrastructure
AI on infrastructure you control.
RAI brings the model, document retrieval, and conversation experience into a locally deployed environment. Organizations gain a useful AI service while retaining control over where their workloads run.
The RAI commitment
- Data remains within the country.
- AI runs on locally deployed infrastructure.
- Organizations retain control of their information.
Self-hosted by design
RAI is deployed on infrastructure inside the country instead of depending on a public AI endpoint abroad. This creates a direct, understandable boundary around the data used for every request.
Layer 1 of 4
Interface
The RAI web app is served from the node. Sign-in uses a secure server session, so no access token is kept in the browser.
- Web app
- Secure session
Layer 2 of 4
Retrieval
Your PDFs are indexed on the node, and the passages relevant to each question are found there too.
- PDF index
- Up to 25 MB per file
Layer 3 of 4
Model
The language model runs on the node itself. Prompts and passages go to it without leaving the country.
- Self-hosted model
- Streaming answers
Layer 4 of 4
Storage
Conversations and documents are stored on the node, in your account, until you delete them.
- Conversations
- Documents
One local path from question to answer
A user sends a prompt or a document, RAI retrieves the relevant context and runs the model locally, and the answer returns without leaving the country.
One request, step by step
Prompt received
What is required above the tender limit?
Operated on your terms
A self-hosted deployment lets each organization align access, monitoring, updates, and capacity with its own operational needs. RAI provides the AI layer without taking data ownership away from the organization using it.
rai node · control room1 of 4 · scroll to continue
Decide who can sign in and use the node.
Finance team
Members · can sign in
Legal team
Members · can sign in
External reviewer
Guest · no access
Next · About RAI