Harsha demonstrates building an automated reporting agent that connects to Google Drive to access data, uses ChatGPT to create a metrics calculation skill, and runs weekly to generate consistent team metrics reports without manual intervention. The agent operates independently with a service-account setup, executes scheduled analyses every Friday, and provides transparent activity logs for monitoring its performance and outputs.
In this video, Harsha demonstrates how to build a simple reporting agent designed to automate the creation of a weekly metrics readout for a team. The process begins by connecting the agent to a data source, specifically Google Drive, where relevant files and spreadsheets are stored. This connection allows the agent to directly access and work with the data, eliminating the need for manual data handling each week.
Harsha sets the connection to be agent-owned, which functions like a service account. This setup enables the agent to perform scheduled or background tasks independently, without relying on any individual’s personal configuration. This approach ensures that the reporting process remains consistent and uninterrupted regardless of personnel changes.
To enhance the agent’s capabilities, Harsha uses ChatGPT to improve the workflow. Instead of manually coding every instruction, he describes the desired job, and ChatGPT suggests creating a metrics calculation skill. This skill encapsulates the team’s processes, definitions, and best practices, guiding the agent on which metrics to focus on, how to interpret them, and how to structure the weekly report. This makes the workflow more reliable and repeatable.
The agent is then scheduled to run automatically every Friday with a simple trigger message like “run analysis.” This automation means the team no longer needs to remember to initiate the reporting process each week, saving time and reducing the risk of missed reports. The agent consistently executes the reporting workflow on this set cadence.
Finally, Harsha shows how to review the agent’s activity history to monitor its work. The team can inspect each run, see the steps the agent took, the tools it used, and the output it generated. In a typical run, the agent accesses spreadsheet data, calculates metrics, creates charts, and compiles the analysis into a comprehensive readout ready to be shared with the team. This transparency provides confidence in the agent’s performance and the accuracy of the reports.