CaseDraft

Manually translating a set of UI screens into structured test cases is slow, repetitive, and easy to shortcut under deadline pressure. CaseDraft is an internal documentation tool that takes Figma exports or file paths and uses AI to auto-generate test cases and use cases, with output formatted for direct copy-paste into Word — the goal being an AI first pass reliable enough that a human only needed to review and refine, not write from scratch.

What I Built

• Figma image ingestion — accepts screenshots or direct exports
• File-path input mode — point the tool at local design files for analysis
• Selenium-based UI extraction for scraping structural context when needed
• AI-generated test cases and use cases derived from the extracted UI context
• Word-ready formatted output — structured text that pastes cleanly into documentation without reformatting

Tech Stack

• Language: Python
• Extraction: Selenium
• AI: OpenAI API (or similar LLM)
• Output: Structured text, Markdown/RTF-compatible

Key Decisions

• Used Selenium for structural UI extraction rather than relying purely on image analysis, since some Figma exports needed DOM-level context that a screenshot alone couldn't provide
• Formatted AI-generated output for direct Word paste rather than a custom viewer, since the goal was fitting into the QA team's existing documentation workflow, not replacing it
• Kept a human review step central to the design rather than positioning the tool as fully automated, treating AI output as a strong first draft rather than a final deliverable

Outcome / Impact

Cut hours of manual documentation work down to minutes for QA and product teams, turning designs directly into a usable first draft of test coverage.