Exa Websets MCP Server

# Websets MCP Server - Usage Examples This document provides practical examples of using the Websets MCP server with Claude. ## Basic Webset Operations ### Example 1: Create a Simple Webset **Prompt to Claude:** ``` Create a webset called "AI Startups" that searches for 15 artificial intelligence companies founded in the last 3 years ``` **What happens:** - Claude calls `create_webset` with: - name: "AI Startups" - searchQuery: "artificial intelligence companies founded in the last 3 years" - searchCount: 15 ### Example 2: Create a Webset with Enrichments **Prompt to Claude:** ``` Create a webset of "Series A SaaS Companies" with 20 companies. Add enrichments for: - Annual Recurring Revenue (ARR) - Number of employees - Primary customer segment - Tech stack used ``` **What happens:** - Claude calls `create_webset` with enrichments array - The webset will automatically populate and start enriching ### Example 3: View Webset Contents **Prompt to Claude:** ``` List all my websets, then show me the full details and items from the "AI Startups" webset ``` **What happens:** - Claude calls `list_websets` first - Then calls `get_webset` with expandItems: true - Shows you all the companies discovered ## Advanced Workflows ### Example 4: Build a Company Database **Complete workflow prompt:** ``` I want to build a database of marketing agencies. Please: 1. Create a webset called "Marketing Agencies" searching for "digital marketing agencies in the US with 50-200 employees" 2. Add these enrichments: - "Revenue": "Annual revenue in USD" - "Clients": "Notable client brands they've worked with" - "Services": "Primary marketing services offered" - "Location": "City and state of headquarters" 3. Set up a weekly monitor that searches for new agencies every Monday at 9am ``` **What happens:** - Step 1: Claude calls `create_webset` - Step 2: Claude calls `create_enrichment` multiple times - Step 3: Claude calls `create_monitor` with schedule "0 9 * * 1" ### Example 5: Research Tracking **Prompt to Claude:** ``` Create a webset to track AI research papers. Search for "machine learning papers on transformer architecture" and get 25 papers. Add enrichments for: - "Authors": "Lead authors of the paper" - "Citations": "Number of citations" - "Key Findings": "Main contributions or findings" - "Year": "Publication year" ``` ### Example 6: Investor Pipeline **Prompt to Claude:** ``` I'm building an investor pipeline. Create a webset called "Climate Tech Startups" that searches for "climate technology startups that raised funding in 2024". Get 30 companies and enrich with: - "Funding Amount": "Total funding raised" - "Stage": "Latest funding stage (seed, series A, etc)" - "Investors": "Lead investors in latest round" - "Technology": "Climate tech category (solar, carbon capture, etc)" Then set up a daily monitor to check for new companies. ``` ## Working with Existing Websets ### Example 7: Update Webset Metadata **Prompt to Claude:** ``` Update my "AI Startups" webset to have the description "Collection of emerging AI companies for potential partnerships" ``` ### Example 8: Add Enrichments to Existing Webset **Prompt to Claude:** ``` For my "Marketing Agencies" webset, add a new enrichment: - "CEO Name": "Name of the current CEO or founder" ``` ### Example 9: List Items with Filtering **Prompt to Claude:** ``` Show me the first 10 items from my "Climate Tech Startups" webset ``` Then you can ask follow-ups like: ``` Show me the next 10 items ``` Claude will use the pagination cursor automatically. ## Monitor Schedules ### Common Cron Patterns **Daily Updates:** ``` Create a monitor for my webset that refreshes items daily at midnight ``` - Schedule: `0 0 * * *` **Weekday Morning Updates:** ``` Create a monitor that searches for new items every weekday at 9am ``` - Schedule: `0 9 * * 1-5` **Weekly Deep Refresh:** ``` Create a monitor that refreshes all item data every Sunday at 3am ``` - Schedule: `0 3 * * 0` - Behavior: `refresh` **Twice Daily:** ``` Create a monitor that searches for new items at 9am and 5pm every day ``` - Schedule: `0 9,17 * * *` ## Tips for Best Results ### Writing Good Search Queries **Good:** - "B2B SaaS companies in San Francisco with Series A funding" - "Academic researchers working on quantum computing at universities" - "Health tech startups focused on mental health apps" **Too Vague:** - "tech companies" (too broad) - "good startups" (subjective terms) ### Writing Good Enrichment Descriptions **Good:** - "Annual recurring revenue in USD as of latest available data" - "Number of full-time employees excluding contractors" - "Primary geographic market (North America, Europe, Asia, etc.)" **Too Vague:** - "revenue" (what currency? what time period?) - "size" (employees? revenue? customers?) ### Using Monitors Effectively **Search behavior:** Use when you want to continuously grow the collection ``` Create a search monitor that runs weekly to find new companies matching the criteria ``` **Refresh behavior:** Use when you want to update existing data ``` Create a refresh monitor that runs daily to update enrichment data for all companies ``` ## Real-World Use Cases ### Use Case 1: Venture Capital Deal Flow **Setup:** ``` Create 3 websets for different investment stages: 1. "Seed Stage Prospects" - early stage companies in our sectors 2. "Series A Targets" - companies ready for growth capital 3. "Late Stage Monitoring" - companies we passed on but want to track For each, add enrichments for: funding history, team size, revenue metrics, and investor lists. Set up weekly monitors. ``` ### Use Case 2: Sales Prospecting **Setup:** ``` Create "Enterprise Sales Prospects" webset searching for "Fortune 1000 companies in financial services". Enrich with: IT budget, current vendors, decision maker contacts, tech stack, and recent news. Set up daily refresh to keep contact info current. ``` ### Use Case 3: Market Research **Setup:** ``` Create "Competitor Analysis" webset tracking "direct competitors in the project management software space". Enrich with: pricing, feature set, customer count, recent product launches, and marketing strategy. Set up twice-weekly search to catch new entrants. ``` ## Troubleshooting Common Issues ### Issue: Not enough items found **Solution:** ``` List my websets and check the status. If a search is still running, wait. If completed with few items, the query might be too specific. Try broadening the search criteria. ``` ### Issue: Enrichments taking too long **Solution:** Enrichments can take time for large websets. Check progress: ``` Get the details of my webset including all items to see enrichment status ``` ### Issue: Want to change search criteria **Solution:** You can't change criteria for existing searches, but you can: ``` Create a new search for my webset with updated criteria ``` Or create a new webset with the correct criteria.