Microsoft Customer Stories

Overview

Search and retrieve customer stories from Microsoft’s official Customer Stories site via its internal API.

Workflow

1. Analyze the user’s request to determine appropriate filters
2. Run scripts/search_stories.py with selected filters to find matching stories
3. Review results and optionally run scripts/fetch_story.py on specific stories for full details
4. Summarize findings for the user

Prerequisites

Install requests in the Python environment: pip install requests

Step 1: Map User Request to Filters

Translate the user’s natural language request into API filter parameters. Consult references/filters.md for the complete list of available filter values.

Mapping guidelines:

Common mappings:

• “中小企業” / “SMB” → --org-size 50-999-employees or --org-size 1-49-employees
• “大企業” / “Enterprise” → --org-size 10000-employees
• “RAG” / “検索拡張生成” → --query RAG + --products azure/azure-openai
• “日本” → --region asia/japan

Step 2: Search Stories

Run the search script:

python scripts/search_stories.py --products azure/azure-openai --region asia/japan --query "RAG" --top 10

Arguments:

• --query / -q: Free text search
• --products / -p: Product filter (e.g., azure/azure-openai, azure/azure-ai-search)
• --region / -r: Region filter (e.g., asia/japan, europe/germany)
• --industry / -i: Industry filter (e.g., healthcare, manufacturing)
• --business-need / -b: Business need (e.g., artificial-intelligence)
• --org-size / -o: Organization size (e.g., 50-999-employees)
• --service / -s: Service filter (e.g., fasttrack)
• --includes: Stories that include (e.g., videos,partners)
• --top / -t: Number of results (default: 12)
• --skip: Pagination offset

Output is JSON with totalCount, hasMorePages, and stories array.

Step 3: Fetch Story Details

For interesting stories, fetch the full content:

python scripts/fetch_story.py 25666-softbank-corp-azure-ai-foundry

Accepts a story slug or full URL. Returns JSON with title, description, content.

Output Format

Always respond in the same language the user used. If the user asks in Japanese, reply in Japanese. If in English, reply in English.

Present results to the user in this format:

## Search Results: {totalCount} stories

### 1. {title}
- **Industry**: {industry}
- **URL**: {url}
- **Summary**: {summary from content}

### 2. {title}
...

When the user asks for details on a specific story, provide a more detailed summary of the story content including key challenges, solutions, and outcomes.