Sprint Ticket Master

Before sprint planning, someone has to turn a rough task list into structured tickets: a clear title, a description, an estimate, an owner and a category. That list usually lives in a spreadsheet.

Sprint Ticket Master automates the step. Upload an Excel file with a Task Description column, review what was extracted, and Gemini writes a full ticket for each task, shown as cards or a table and editable on a Jira-style detail page.

SheetJS parses the workbook in the browser, so nothing leaves the machine until you choose to analyse. Files without a Task Description column are rejected with a clear message before any request is made.

The tasks go to a single Next.js API route, which keeps the Gemini key on the server, calls gemini-2.0-flash, extracts the JSON array from the reply and assigns ticket IDs. Without a key configured, the route returns built-in demo tickets and the interface says so in a banner.

The prompt teaches the output by example: one complete ticket in JSON with all eight fields (title, three-point description, story points, assignee team, reporting lead, parent ticket, category, status), followed by the numbered task list.

It also grounds the model in a fixed company context, with four teams, four leads, four categories and a parent-ticket ID format, so assignments land on real options instead of invented ones. Temperature 0.2 keeps estimates consistent between runs.

Tickets open in a card view for a quick read or a table for scanning the whole sprint. Each one links to a detail page where the title, story points and status can be edited and saved for the session. The screenshots below show the app's built-in demo tickets.

The prototype proves the flow end to end. The next version would harden it: