Frequently Asked Questions: Valsight AI Assistant
When will the Valsight Assistant be rolled out for me?
From the 7.0 release onward, the Valsight Assistant can be enabled. A key prerequisite is your active consent to the required contract amendments and/or data protection terms. Usage is token-based, optional, and can be enabled per instance on request; we will inform you separately before activation.
Are my data secure?
Yes. The AI has no access to numeric values; it bases its answers exclusively on the Valsight documentation and on project structures information like models, dimensions and formulas to generate explanations and suggestions. Customer data is kept separate and confidential and our LLMs do not train on or store any customer information.
Existing access rights remain unchanged: the AI operates in read-only mode and can access projects only with the respective user permissions - it cannot see more than the signed-in person. An external review of the Valsight Assistant by an independent auditor (penetration test) has already been completed with no security vulnerabilities identified.
Can AI change my projects?
No. The Assistant runs in read-only mode - no write permissions and no changes to models, data, or scenarios. It can explain and make suggestions; implementation remains with you.
Can I trust the answers?
As decision support - not as a single source of truth. The Assistant derives its answers from your model structure and the available data, and helps make relationships visible. The business review remains with you - please validate suggestions and explanations before adopting them.
What can the AI Assistant help me with?
The AI Assistant can help you:
Understand model structure and node relationships
Explain formulas and calculations
Navigate data sources and identify relevant tables
Learn Valsight features and operations
Get quick answers to modeling questions
It works best for explanatory and guidance questions about your specific model.
What data does the AI have access to?
The AI can see:
Model structure (nodes, connections, hierarchies)
Formulas and operation syntax
Data source names, tables, and column metadata
Project configuration and annotations
The AI cannot see:
Numeric values in your data
Actual calculation results
User credentials or personal information
Projects you don't have permission to access
Can I use AI Assistant with any model?
Yes. The AI Assistant works with any Valsight model you have access to. It automatically adapts to your model's structure and data sources. The more annotations and descriptions you've added to your model, the more contextual the AI's answers will be.
Does my chat history persist?
Yes, but only per browser tab and for the session. Each tab maintains its own independent conversation. Your chat history remains available even if you refresh the page, but closing the tab will clear the entire conversation. There is no longterm chat history.
What if the chat gets stuck or doesn't respond?
If the AI doesn't respond within 30 seconds:
Check your internet connection
Refresh the page and try again
Rephrase your question to be more specific
Try a simpler question first to test connectivity
The chat interface is designed to remain responsive. If you experience persistent issues, contact your Valsight administrator.
Can I ask questions in any language?
The AI Assistant is optimized for English. While it may understand questions in other languages, response quality and accuracy are best when using English.
What happens if the AI makes an error in its response?
The AI generates responses based on your model structure and available documentation. If an answer seems incorrect:
Rate the response with thumbs down to provide feedback
Verify the information against your model directly
Rephrase your question with more specific details
Remember: the AI Assistant is a support tool - always validate important information
The AI learns from your feedback to improve over time.
Are there limits to how much I can use the AI?
[PLACEHOLDER: Rate limit information needs clarification from Valsight team]
Usage is token-based and subject to limits. If you encounter usage restrictions:
Wait a few minutes before asking another question
Contact your Valsight administrator for information about your instance's limits
Detailed rate limit information will be provided during rollout
How do I rate AI responses?
After each AI response, you'll see thumbs up and thumbs down buttons. Click the button that reflects your experience:
Thumbs up: The answer was helpful, accurate, and relevant
Thumbs down: The answer was unhelpful, incorrect, or off-topic
Ratings are optional but help improve the AI Assistant. Once submitted, ratings cannot be changed.
Can the AI perform calculations for me?
No. The AI cannot access numeric values or perform calculations. It can explain how formulas work and what calculations your model performs, but it cannot compute actual results.
Does the AI understand my business context?
The AI understands your model's technical structure but not your specific business meaning unless you've documented it. To get better answers:
Add descriptions to your nodes
Include annotations explaining business logic
Use clear, descriptive names for nodes and data sources
The more context you provide in your model, the more relevant the AI's answers will be.
Can other users see my conversations?
No. Your AI conversations are private and not shared with other users. Each user has their own independent conversations based on their access permissions. However, your conversations will be logged for audit reasons.
What if I have a question the AI can't answer?
The AI is designed to help with model structure, formulas, and Valsight features. For questions outside this scope:
Check the full Valsight documentation
Contact Valsight support for technical issues
Reach out to your internal Valsight administrators
Consult with your team for business-specific guidance
Related Documentation
Valsight AI Assistant - Overview information
How the Valsight AI Assistant Works - Understanding the AI’s capabilities