Turn scattered API docs into clean, reusable schemas. This build researches each service, extracts real API operations, and outputs a structured schema file. It suits product and engineering teams that need fast vendor due diligence, partner onboarding, or an internal API catalog.

The run starts with a list in Google Sheets. It calls Apify to search the web for official docs and scrapes pages to collect content. Text is split into chunks, embedded with Google Gemini, and stored in Qdrant for fast search. The flow then identifies a service’s products, finds matching docs in the vector store, and uses Gemini to extract endpoints and methods. Results write back to Google Sheets. A code step builds a custom JSON schema and uploads it to Google Drive. Event routing, batching, and wait steps coordinate the three stages end to end.

Plan for Google Sheets and Drive access, an Apify token, a Google Gemini API key, and a Qdrant URL and key. Expect research time to drop from hours to minutes per service, with more consistent results. Useful for partner evaluations, integration planning, and building a private API knowledge base. Start with a few services in the sheet, run a test, and confirm each stage writes results before scaling up.

• Web search and scrape with Apify to find and collect official docs from each service.
• Duplicate filtering and result checks to keep only relevant pages before processing.
• Chunking and metadata tagging so long pages are split and labeled for better context.
• Google Gemini embeddings to turn text into vectors and store them in Qdrant for fast search.
• Targeted retrieval from Qdrant to pull only docs tied to a service’s products and features.
• LLM powered extraction with Google Gemini to list endpoints, methods, and parameters.
• Code step to group operations into a clean JSON schema ready for reuse.
• Google Sheets logging of status and results, and Google Drive upload of final schema files.
• Event routing, batching, and wait steps to coordinate research, extraction, and generation stages.

• Reduce manual research from 3 hours to 15 minutes per service
• Automate 80 percent of repetitive web search and scraping tasks
• Improve data consistency by 70 percent with structured extraction
• Handle 10 times more services with batching and vector search
• Connect Google Sheets, Apify, Google Gemini, Qdrant and Google Drive without custom code

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Google Sheets, Google Drive, Google Gemini, Apify and Qdrant. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, create Google Sheets OAuth2 credentials. Grant access to the spreadsheet used for research, extraction, and results logging.
4. Create Google Drive OAuth2 credentials. Grant access to the folder where schema files will be uploaded. Copy the folder ID and set it in the Upload to Drive node.
5. Get an Apify API token from your Apify account. In the Web Search For API Schema node, set HTTP Header Auth and add Authorization with value Bearer YOUR_APIFY_TOKEN. In the Scrape Webpage Contents node, use HTTP Query Auth and add token as a query parameter.
6. Create a Google Gemini API key in Google AI Studio. In n8n, open the Gemini nodes, choose Create new credential, and paste the API key. Use the same credential for both chat and embeddings nodes.
7. Set up Qdrant credentials with the base URL and API key. In the Qdrant vector store nodes, select the credential and set a collection name for document embeddings.
8. Open the Google Sheets nodes like Get All Research, Append Row, and Get API Operations. Point them to your spreadsheet and map the correct columns for service name, status, and extracted operations.
9. Review the EventRouter switch rules and make sure the event values used in your sheet match the research, extract, and generate paths.
10. Run a small test with the manual trigger. Check Google Sheets for pending items moving through Research, Extract, and Generate. Confirm Apify returns results and the scraper produces page content.
11. Verify Qdrant collections exist and contain vectors. Use the Search in Relevant Docs nodes to confirm that queries return documents. If empty, adjust search terms in the query templates.
12. If scraping fails, reduce batch size in the scraper options or increase memory. If Drive upload fails, confirm the folder ID and Drive credential. For empty extractions, refine templates or increase chunk size in the text splitter.