Analytics Agent
The Analytics Agent can generate charts, tables, dashboards, and downloadable files, and text-based responses. It is developed and maintained by the BAI team.
You can use Analytical Agent to:
- Answer questions about business data.
- Create charts and visualizations.
- Generate dashboards.
- Perform advanced analytical tasks, such as trend analysis, anomaly detection, root-cause investigation, and forecasting.
- Export results to downloadable files.
- Save results for future use.
- Provide analytics services to other AI assistants, such as the Infor GenAI Assistant.
You can ask the Analytics Agent questions, such as:
- What were total sales last quarter?
- Show revenue by region.
- Identify the drivers behind the decline in customer retention.
- Forecast sales for the next six months.
- Create a dashboard that shows product performance by category.
How the Analytics Agent works
Unlike traditional AI assistants that return text responses or invoke a limited set of predefined functions, the Analytics Agent can perform multiple analytical steps within a single request.
When processing a request, the Analytics Agent can:
- Determine which data is required.
- Retrieve data from the appropriate Birst space.
- Perform calculations, comparisons, and analysis.
- Generate the required output format.
- Return the completed result to the user.
Depending on the request, the response can include:
- Text-based insights
- Charts and visualizations
- Formatted tables
- Dashboards
- Downloadable files
Because the Analytics Agent combines several operations into a single workflow, the agent can answer complex, multi-step questions without requiring users to guide each intermediate step.
Context Awareness
The Analytics Agent uses information from the context where the agent starts.
For example, when the agent is associated with a dashboard or report, the agent can recognize:
- Existing filters
- Current report selections
- Data that is already displayed
Context awareness reduces the need to repeat information that is already available in the current analytics session.
Analytics Agent compared to a traditional AI Assistant
A traditional AI assistant provides information about data. The Analytical Agent creates outputs that you can use directly, such as charts, files, or dashboards.
This table shows the comparison between a traditional AI Assistant and Analytical Agent:
| Capability | Traditional AI Assistant | Analytics Agent |
|---|---|---|
| Multi-step questions | Requires multiple exchanges to complete several analytical steps. | Performs multiple analytical steps in a single request. |
| Output | Text only | Text, charts, tables, dashboards, downloadable files. |
| Advanced analysis | Limited to what a pre-built function returns | Performs trend analysis, root-cause analysis, anomaly detection, forecasting, and other analyses. |
| Context awareness | Limited or unavailable | Uses dashboard, report, and filter context. |
| Reusability | N/A | Results can be saved and re-run later with updated data |
Access to the Analytics Agent depends on the customer's configuration.
A customer can enable or disable capabilities, such as:
- File exports
- Dashboard creation
- Saved results
- Access from other AI assistants
If the Analytics Agent is unavailable, or an expected capability is unavailable, contact the account team or product team to verify configuration settings.
Limitations:
- The Analytics Agent can access data that the requesting user has permission to view. The Analytics Agent does not provide additional data access.
- When used inside another assistant rather than directly in Birst, it can only reply in text, no charts, dashboards, or files in that mode.
- The Analytics Agent processes one question at a time within a conversation.
- Broad or open-ended requests may not complete successfully. Smaller and more specific requests typically produce better results.
- Conversations and generated files are temporary and expire after a period of inactivity.
- Availability of capabilities, such as exports, saved or reusable results, and dashboard creation depends on the deployment configuration.