How to create automatic export actions in Copilot in Power BI

This tutorial shows how to configure Copilot in Power BI to create automatic export actions (PDF and CSV) from prompts. This feature is useful to automate quick reports and share structured files without manually navigating the service. By automating these exports you can save time — for example, generate 20 daily reports in seconds instead of 30–60 human minutes per report — and reduce human errors when selecting filters and pages.

Prerequisites

• Account with access to Power BI Service and Copilot enabled in the organization.
• Report published in the workspace with edit/view permissions.
• Basic knowledge of Power BI Desktop and navigation in Power BI Service.
• Permissions to create and run Copilot actions (admin/owner or configured by the admin).

Additionally, ensure you have access to the delivery destination (OneDrive/SharePoint) and that you know quotas: for example, CSV files smaller than 10 MB and PDF up to 50 MB are more reliable for automatic transfer. If you work with very large datasets, consider filtering the data to reduce volume before export.

Step 1: Prepare the report and define export pages

Organize the report to have specific pages that you will export. Practical examples: create an “Executive Summary” page with KPIs and charts (1 page, ideal for PDF) and another “Detailed Data” page with a paginated table to export to CSV. Use bookmarks and roles to define predefined versions (for example, filters for the current quarter).

Tip: limit the visuals on the export page to avoid the PDF exceeding an A4/A3 layout. If the page has many tables, create a compact view with only the necessary columns. Practical test: reduce the column list from 30 to 6 to see improvements in time and file size.

Step 2: Create a Copilot prompt with export instructions

Open the report in Power BI Service and enable the Copilot pane. The prompt should be clear and prescriptive — Copilot follows literal instructions. Indicate page, format, layout, columns and file naming pattern. Use date placeholders to automate daily versions.

Exporta a página 'Resumo executivo' para PDF com layout A4 e inclui header com título. Exporta a página 'Dados detalhados' para CSV apenas com as colunas: Cliente, Produto, Quantidade, Valor. Nome dos ficheiros: Relatorio_Resumo_{yyyyMMdd}.pdf e Dados_Detalhados_{yyyyMMdd}.csv

Practical example: if today is 2026-10-10, the files will be named like Relatorio_Resumo_20261010.pdf and Dados_Detalhados_20261010.csv, which makes organization and automatic ingestion in later pipelines easier.

Step 3: Create a Copilot Action for export

If your organization allows Copilot Actions, create an action that links the prompt to automatic execution. In the Copilot pane choose "Create action" and paste the prompt; then define parameters such as file name, format and columns. A well-defined action can have manual and scheduled triggers to cover ad-hoc and automated scenarios.

{   "name": "ExportarRelatoriosDiarios",   "triggers": ["manual","agendado"],   "steps": [     {"type":"export","page":"Resumo executivo","format":"pdf","fileName":"Relatorio_Resumo_{date}"},     {"type":"export","page":"Dados detalhados","format":"csv","fileName":"Dados_Detalhados_{date}","columns":["Cliente","Produto","Quantidade","Valor"]}   ] }

Note: adapt the JSON above to the syntax your tenant supports. The most common pattern is to have one step per file and explicitly define columns for CSV to avoid exporting unnecessary fields.

Step 4: Test the action manually

Run the Copilot Action manually from the report to verify that the files are generated with the intended content and format. Test with a data subset (for example, the first month of the year) to validate content in under 2 minutes.

Observe Copilot log messages for errors — typical logs include timestamps, execution ID and error messages. A common error example is «coluna inexistente: ValorUnitário» when we requested «Valor». Fix the prompt or the data model as needed.

Step 5: Schedule exports and configure delivery

For periodic exports, configure the scheduled trigger in the Copilot Action (daily, weekly). Choose the time: e.g., daily at 07:00 for reports closing the previous day. Define delivery: direct download, email sending (if supported) or saving to OneDrive/SharePoint. Check destination permissions and storage quotas.

{   "schedule": {"frequency":"daily","time":"07:00"},   "delivery": {"type":"sharepoint","site":"/sites/Finance","folder":"/Relatorios/Diarios"} }

Practical example: configuring saving to "/Relatorios/Diarios" on SharePoint allows the team of 5 users to automatically access the files and for an ETL process to pick those files every morning.

Step 6: Handle common errors

Frequent errors include: lack of permissions in the workspace/destination, columns referenced in the prompt that do not exist, and export limits (for example, exporting more than 200 pages or files larger than allowed). For each error, fix the prompt, adjust permissions or split the export into multiple steps.

How to resolve: 1) Permissions — confirm the identity executing the action has write permission on SharePoint; 2) Columns — run a simple query in Power BI Desktop to list columns; 3) Size — apply date filters to reduce the file or share in parts (e.g., by week).

Verify the result

Confirm that the PDF and CSV files were generated with the correct names and that the data/visuals match expectations. For CSV, open in a text editor/Excel and validate delimiters and number of columns; for PDF, check layout, headers and bookmarks. Review the Copilot Action logs for messages and execution timestamps and keep a record of the last 30 runs for auditing.

Conclusion

With this approach you can automate recurring exports using Copilot in Power BI, save time and reduce manual errors. Recommended next steps: integrate email notifications, parameterize date/periods in prompts and validate destination security with the IT team. Practical tip: start with a manual action and test with a small dataset (100–1,000 rows) before scheduling in production.