Share via


Generate coded actions with natural language in an agent

Enabled for Public preview General availability
Admins, makers, marketers, or analysts, automatically Jun 16, 2025 Jul 18, 2025

Business value

The business value is end-to-end business processes execution without manual sequential workflows to perform complex mathematical calculations. You can automate complex operations like write actions to Dataverse, send emails, and create CRM cases. This capability lowers costs by minimizing large language model (LLM) calls, making agent systems about 30% more affordable. The solution delivers higher-quality output with the ability to create graphs.

Feature details

This feature makes prompts actionable with external sources by using logs for validations and better exception handling. This feature also helps to overcome limitations of a native large language model (LLM). The native prompt builder experience in Microsoft Copilot Studio is built on Fluent UI 9 (non-iFramed experience). It removes the wizard for prompt actions by integrating code interpreter elements to work with the new prompt builder experience.

Here are some tasks you can perform with this feature: You can turn on the toggle to generate the code from Adv. Setting. You can turn off the toggle to stop generating the code. You can select the checkbox to write data to Dataverse tables in Adv. Setting. You can open the Preview code to autogenerate Python code and run it. You can edit the Python code and re-run the code. You can copy and paste the code in a code interpreter you find on the internet. You can undo code changes by using the Undo button. When you turn on the setting to write data to Dataverse tables, you can use the Run button to write actions to Dataverse tables.

Geographic areas

Visit the Explore Feature Geography report for Microsoft Azure areas where this feature is planned or available.

Language availability

Visit the Explore Feature Language report for information on this feature's availability.

Additional resources

Use code interpreter to generate and execute Python code (docs)