Starting from Bold BI v10.1, on-premise users can configure their own AI models with credentials for AI features.
BOLD AI Service
Once users configure the Bold AI Service, we manage AI functionalities on our end seamlessly. it requires no credentials and operates on a cloud-based service.
Versions up to 13: Works with your existing Bold BI license.
Version 14+: Runs per-tenant with 1.5 million free tokens monthly, zero billing.
We provide 1.5 million tokens every month per tenant, which is free of charge and will not be billed.
How are the Tokens Applied for Bold BI Service Tenants?
Each tenant has its own dedicated 1.5 million token pool per month.
Tokens cannot be shared or transferred between tenants.
All AI usage regardless of the number of users (e.g., 10 or 100 embedded users) is deducted from the same monthly tenant token pool.
The full 1.5 million tokens are automatically reset on the first day of every month.
Understanding Token Usage
AI token usage varies based on the datasource size, column count, and complexity of the query. It is not constant and can differ notably across dashboards and operations. For example, in a Sales Dashboard Analysis scenario
Dashboard: Sales analysis Dashboard
Datasource: Sales
Number of Columns: 43
Purpose
Average Token Usage (Min - Max)
AI Summarization
0.6K – 0.75K tokens
Widget property updates
4K – 5K tokens
Widget creation
8.90K – 9.1K tokens
NOTE: Token usage estimates may vary based on data size, structure, and query complexity. If usage is high or frequent, additional tokens may require prior payment.
Open AI
Users can integrate their own OpenAI model for AI features by entering an API key.
Once configured, we can identify the selected model from our end.
Azure AI
For Azure AI configuration, users need to provide their Azure AI model name, resource name, deployment name, and API key.
NOTE: We recommend using gpt-4o deployment for Azure AI to achieve better responses.
The following section explains how to configure AI settings in both UMS and Site Settings.