gmail-export

AI-powered UX case study extraction from email threads

Pipeline for analyzing Gmail exports and generating UX design case studies using LLM classification

PythonGmail APIClaude

A toolkit for turning Gmail threads into UX case study content. Export, classify, analyze, generate—all powered by LLM analysis.

Problem

UX case studies require narrative construction from project history. That history often lives in email threads—client conversations, stakeholder feedback, design decisions. Manually extracting and structuring this content is tedious.

Architecture

Python pipeline with AI-driven analysis:

┌─────────────────┐
│   Gmail API     │
│  (thread export)│
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│    Classifier   │
│  (LLM-powered)  │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Case Study     │
│   Analyzer      │
└────────┬────────┘
         │
         ▼
┌─────────────────────────────────┐
│      Modular Writeups           │
│  ┌──────┐ ┌──────┐ ┌──────┐   │
│  │01-ctx│ │02-prb│ │03-sol│   │
│  └──────┘ └──────┘ └──────┘   │
└─────────────────────────────────┘

Three-Stage Pipeline:

  1. gmail_export.py — Fetch threads via Gmail API
  2. thread_classifier.py — LLM categorizes by project/topic
  3. case_study_analyzer.py — Generate structured narratives

Key Pattern: Modular Content Generation

Output is structured into reusable pieces:

writeups/
├── 01-context.md      # Project background
├── 02-problem.md      # Challenge definition
├── 03-approach.md     # Methodology
├── 04-solution.md     # Design decisions
└── 05-results.md      # Outcomes

Each piece can be assembled into different case study formats—portfolio, presentation, detailed write-up.

Custom AI Commands

The .claude/commands/ directory defines domain-specific operations:

  • Thread classification prompts
  • Narrative extraction templates
  • Case study assembly workflows

Turns generic LLM into specialized UX documentation tool.

Current Status

Why This Exists

Every UX project generates documentation debt. Emails contain decisions, context, and rationale that never make it into final deliverables. This pipeline recovers that knowledge and structures it for portfolio use—AI doing the tedious extraction work humans avoid.