Report Copilot: Core Workflow
Ready to add county
Beta Note: Report Copilot is currently in beta and will be gradually shipped to clients in 1H 2026.
Report Copilot is offered in two modes, each suited to a different use case:
- Report Builder Mode: For users who want to create reports with AI and bypass the use of SQL Entirely.
- Open the Report Copilot panel, describe your report in plain English, answer clarifying questions such as grain or time window, review the generated SQL and discovery summary if you choose, then run and save. Iterate with follow-up prompts until the output is correct. This works best when prompts are specific about metrics, filters, dimensions, and output shape.
- Data Exploration Mode: For users who want to derive insights with no SQL familiarity. Ask questions in natural language and see data and answers quickly, without writing or editing SQL. This mode is ideal for ad hoc exploration and quick answers.
In Builder Mode, you can copy generated SQL into the editor to run it, save the report, or publish it. In Data Exploration Mode, the focus is on questions and results rather than saving report definitions.
To enable Builder mode, select Enable AI to the right of the main toolbar. The prompting pane opens on the right. Now you have many options for the AI to assist you in making a report.
Building a New Report - Report Builder Mode
The process is largely self-guided. Start by telling the agent what type of report you want, including the fields, filters, aggregations (such as count or sum), and sorting. The Report Copilot will ask for the report name, purpose, and any additional details needed to clarify the request. After you answer these prompts, the agent presents a plan for the report and asks for confirmation. Once confirmed, it generates the report, testing and refining the query before returning the final results in the output pane. You can continue giving instructions to refine the report until it meets your needs. When finished, save the report using the SQL Editor tools. You may also attach documentation or specifications to your prompt to help the agent understand your request, though very large files may exceed system limits.
Start Copilot
The report copilot will prompt for the report name, report purpose, and any other clarifying addition it needs. Please answer these prompts and then it will describe its plan for the report and ask for your confirmation. Once you confirm it will build the report, and as it is generating the report
Answering first dialog
Now the system has a few more things to clarify. Clarification of what the claim amount really means, which closed claims should be included, how to define the date, and report purpose, and even interprets and corrects a typing error.
Answering second dialog
Continued clarifications.
Final dialog
Once you confirm that the proposed report plan looks correct, the agent uses its knowledge of SQL queries along with its understanding of your database and the BriteCore system to generate the report. Below are the results from your first AI-generated query.
Claim report with results
Modifying an Existing Stock Report
In this use case, BriteCore has a similar report and you desire to make slight modifications to it. It is easiest to open the report in the editor and run it once manually, so you can see the outputs in its standard configuration. Then tell the Report Writer in english what you want to do, such as:
“Please remove claim_id and add the description field”
Opening stock report
Generating Data Insights in Data Exploration Mode
Building a report in Data Exploration mode is very similar, except the tool uses all of its knowledge to infer the answers to all of the aforementioned questions. Rather than asking, it chooses fields it believes relevant to the query, picks the filter that chooses rows by what it believes the most likely interpretation of your English is, and gives you a result based on that. You can ask the agent to modify its assumptions and guide it to the report or insights you need.