Modern business intelligence is about interacting with your data. In this webinar, Suriyaprakasam demonstrated how the generative AI features in Bold BI® transform dashboards into conversational, predictive, and prescriptive intelligence platforms.
Key Highlights Covered
- Natural Language Queries: Ask questions and get instant visual answers.
- AI Copilot: Generate dashboards and visualizations automatically from simple prompts.
- Smart Summaries: Let AI narrate key insights from dashboards and widgets for quick decision-making.
- Real-Time Insights: Always access the most up-to-date data without manual refreshes.
- Embedded AI: Seamlessly integrate Bold BI AI features into applications using the Embedding SDK.
Real-World Applications
- Sales Forecasting: Predict demand and optimize inventory.
- Customer Support: Analyze tickets to identify common issues and reduce churn.
- Healthcare: Monitor patients and predict risks with real-time alerts.
- Finance: Automate narratives, detect anomalies, and simplify cash flow analysis.
Q&A
Q: If I start working with one data source, do I need to @mention the data source repeatedly?
A: In the unified AI chat assistant, you currently need to specify the data source name each time you interact. However, the Bold BI team is actively working on integrating AI memory to streamline this process in future updates.
In contrast, when using the AI Agent in the dashboard viewer or the AI Copilot in the dashboard designer, repeated mention of the data source name is not required. These interfaces maintain context during interactions.
Q: How do I check the current version of Bold BI?
A: On the Bold BI website, click on the question mark icon (?) in the top-right corner of the page, and you will find the App version and Database version listed in the Version section
Q: Is Bold Reports on the roadmap for gen-AI features?
A: Yes, Bold Reports is being considered for future generative AI enhancements, though the timeline and scope are still under evaluation.
Q: Is this available for .NET development?
A: Yes, Bold BI supports integration with .NET applications, including embedding dashboards and using APIs. Gen AI features can be accessed via embedded dashboards. Read the Embedding SDK API reference for details.
Q: How many visuals will the AI dashboard create?
A: Currently, the AI dashboard generates six visuals automatically. Our team is actively working to expand and improve this capability.
Q: How is everything wired underneath—the AI model and data sources? How do I set those up?
A: Bold BI uses a combination of LLMs (large language models) and internal logic to interpret queries. You need to configure your data sources, enable the AI Assistant, and ensure Widget Insights and Dashboard Summary features are turned on.
The AI Agent in Bold BI is powered by models like GPT-4.0 and GPT-4.0 Mini, guided by structured prompts to ensure consistent output. Data sources are accessed through table schemas, which provide metadata such as column names, aggregations, sorting, and filters. This metadata drives the generation of visualizations. Summarization is performed using only the visualized data, not raw data or sensitive information. The setup ensures secure and context-aware AI interactions within dashboards.
Q: Can end users/clients also interrogate the data with different widgets and build custom dashboards using this AI capability? Or is it to be done by the developer who then publishes the files?
A: The AI can be interacted with in different modes in Bold BI:
- AI Agent (viewer mode): End users can interact with the data connected to the dashboard, generate summaries, and create visuals through conversation to explore metrics more effectively.
- AI Copilot (designer mode): If users have access to design mode, they can interactively generate visuals and place them in relevant positions to build dashboards. This feature is primarily intended for dashboard developers to leverage the AI Copilot for creating widgets and dashboards, but it can also be made available to end users.
- Unified AI Assistant (server page): This assistant allows users to explore both data and dashboards, similar to the experience in viewer mode. Users can interactively analyze data and generate summaries across dashboards.
Q: Can the data sources only be structured databases? Can the AI connect to files or documents?
A: Bold BI supports structured data sources like SQL, Excel, REST APIs, and cloud services. Unstructured files and documents are not directly supported for AI analysis yet. This will be included in our future releases.
Q: How much does it cost to use AI in Bold BI?
A: AI features are part of Bold BI licensing. Specific pricing may vary based on your subscription tier. Contact the Syncfusion sales team for detailed pricing. We support the bring-your-own-key (BYOK) model, allowing users to connect their own OpenAI or Azure OpenAI accounts. To understand the token usage, refer the documentation: Managing AI Features and Configuring Your Own Key | Bold BI Docs.
To configure BYOK, please refer to the documentation.
Q: What types of data sources are supported?
A: Bold BI supports over 100 data sources, including SQL Server, MySQL, PostgreSQL, Excel, Google Sheets, REST APIs, Azure, Salesforce, and many more. Bold BI AI features are designed to work seamlessly with these data sources and future releases will continue to expand compatibility.
Q: What setup is required to get this running?
A: You need to:
- Install and configure Bold BI Server.
- Enable AI Assistant, Widget Insights, and Dashboard Summary features.
- Connect your data sources.
- Ensure proper permissions and roles are set.
Please refer to the following links explaining how to configure AI settings in both UMS and the site settings:
Q: I have an AI model in Azure Foundry and data sources like CosmosDB and Azure Blob. How will it work?
A: You can connect CosmosDB and Azure Blob to Bold BI as data sources. Integration with external AI models like Azure Foundry is not natively supported but may be possible via custom APIs.
Currently, we support only Azure OpenAI and OpenAI models. In the future, we plan to extend BYOK support to additional AI models, as well.
Q: Can this AI help to create data sources?
A: No, the AI currently does not create data sources. You must configure them manually in Bold BI.
Q: Apart from data summaries, what else does the AI do in the data part?
A: In addition to data summaries, we rely on metadata (table schema) to generate visuals. For insights and summaries, we exclusively use aggregated or summarized data derived from widget visualizations. Raw data is never used directly in this process.
Q: Is it possible to integrate it with .NET Blazor applications?
A: Yes, Bold BI dashboards can be embedded in Blazor apps using an iframe or SDK. AI features will be available within the embedded dashboards.
Read the Embedding SDK API reference for detail.
Q: How are we controlling nondeterministic or hallucinated responses?
A: Bold BI reduces hallucinated and nondeterministic responses by using controlled prompts, contextual grounding, and models that are aware of visual data during summarization. However, some unpredictability may still occur, and validation is recommended.
Q: Is it possible to train the AI specifically to a dataset?
A: Not directly. The Bold BI AI uses general models and interprets queries based on the connected dataset. Fine-tuning or training is not currently supported.
Q: Do we set the AI per tenant or per server?
A: AI settings are configured per tenant in Bold BI.
Q: Can we configure system prompt-style prefixes or parameters per dashboard or data source? How does RLS impact AI responses?
A: System prompt customization is currently not supported. We plan to introduce the ability to configure system prompt-style prefixes in future updates. Role-level security (RLS) is controlled through the configuration and setup defined within the dashboard.
Q: If I want to use ChatGPT, it asks for an API key—where do I get that?
A: You can get an API key by signing up for OpenAI. Bold BI does not require this unless you're integrating external AI manually.
Q: Is there a charge per inquiry? Do we need to purchase tokens?
A: The Bold BI AI cloud service is offered with a basic rate limit of one million tokens to support the exploration and evaluation of AI capabilities. For production use, we recommend setting up your own OpenAI or Azure OpenAI account. Alternatively, you may contact us to extend the credits on our service, where we can explore paid options based on your requirements.
To understand the token usage further, refer to our documentation.
Q: Can I integrate my own AI model and train it in Bold BI?
A: Currently, we provide support for connecting to OpenAI and Azure OpenAI services. We plan to extend these capabilities in future updates.
Q: Can I use the Syncfusion AI Assistant in my Blazor app and create visuals from there?
A: You can embed Bold BI dashboards created with the AI Assistant into your Blazor app. We currently support an AI chat assistant for both the viewer and designer modes. However, the ability to embed the AI Assistant on server pages is not yet available.
Q: How do I configure AI in Bold BI?
A: Here’s the setup process:
- Set up the Bold BI Server.
- Activate the login account through the Manage License section.
- Navigate to Admin Panel → AI Settings.
- Enable the AI Assistant, Widget Insights, and Dashboard Summary features.
- Save the changes and restart services if required.
Please refer to the following section explaining how to configure AI settings in both UMS and site settings:
Why It Matters
Generative AI empowers every team member to explore data using plain language. This democratization of data insight reduces delays, works across departments, and supports smarter strategies in fast-moving industries.
If you missed the webinar, or would like to watch it again, the recording is posted to our YouTube channel
Conclusion
Using the generative AI features in Bold BI®, it’s time to move beyond static dashboards and start having real conversations with your data.
Start your journey today.
Top comments (0)