Nova
SQB's unified corporate AI platform for knowledge, documents, speech and internal tools

Overview
Nova is a corporate AI platform for SQB employees. It unites a chat interface, search over the bank's approved sources, work with text and scanned PDFs, audio transcription, and a single connection layer for internal tools. Department tools connect to Nova over the MCP protocol through one contract: the platform checks the user's rights before every call and asks for confirmation before any action. Instead of separate assistants per department, Nova gives the bank one shared AI foundation.
What Nova does
Nova.Chat
One chat interface for work questions and continuing the current conversation's context.
Nova.RAG
Search over SQB's approved sources, respecting access rights and citing the source used.
Nova.Vision
Works with text and scanned PDFs: extracts text and answers questions about the document.
Nova.Speech
Transcribes uploaded audio or a mic recording with timestamps and answers questions about it.
Nova.Core
The intelligent core that routes a request to the right mode and merges the processing results.
Nova.API & Nova.Tools
A single point to connect bank systems, calculations and department tools to the AI platform.
What an employee can ask
Nova collapses the path an employee used to walk across many systems and folders into one chat - an answer with a source link:
- “I joined in April - when can I take leave and for how many days?”
- “Who is the director of the operations department?”
- “Which department could this letter be addressed to?”
- “Review this document and check it for legal compliance”
- “Did the operator resolve the client's issue and how did the client react?”
Security
For a banking AI platform, security is part of the product. Nova is built around trusted access, data isolation and processing inside SQB's infrastructure:
- Access only after authentication via the bank's IdP/SSO.
- Search fragments are filtered before the context reaches the model.
- Chats, documents, audio and OCR/STT results are isolated between users.
- Protection against prompt injection in user queries and indexed documents.
- Actions, denials and errors are logged per the bank's audit requirements.
Want to know more about this project?
Reach out to the SQB AI team - we'll share the details and show the platform in action.
Contact us