Turn a Google Drive document into a chat ready knowledge base. Team members and customers can ask questions and get answers that come directly from the file. This fits product manuals, policy papers, onboarding packs, and whitepapers.

When you click Test Workflow, the flow sets a file URL, downloads the file from Google Drive, splits the text into chunks, builds embeddings with OpenAI, and writes vectors into a Pinecone index. For chat, a message triggers an agent that embeds the question, pulls the best matching chunks from Pinecone, and uses an AI chat model to write a clear answer. The model is guided by the retrieved text, which helps keep answers on topic and reduces guesswork. People get what they need fast without reading long PDFs.

You will need a Pinecone index with 1536 dimensions, plus working credentials for Google Drive, OpenAI, and Pinecone. Expect faster answers and fewer repeat questions on complex documents. This setup works well for help centers, internal wikis, and sales enablement content. It also supports more files over time by repeating the load step for new sources.

• Manual trigger loads and indexes a Google Drive file for quick testing
• Set node stores the file URL so you can switch documents easily
• Google Drive node downloads the file directly from a shared link
• Recursive text splitter creates 3000 character chunks with 200 overlap for better recall
• OpenAI embeddings convert text into vectors that can be searched
• Pinecone insert mode writes vectors and clears the namespace before each load
• Chat trigger accepts incoming questions and starts the answer flow
• Pinecone retrieve as tool exposes the index to the agent with a named tool
• OpenAI chat model composes the final answer from retrieved facts
• Question and Answer agent coordinates the model and vector search tool

• Reduce manual document search from 30 minutes to 2 minutes per question
• Cut support response time for document questions by about 60 percent
• Connect Google Drive, OpenAI, and Pinecone in one flow
• Scale from a single file to many without changing the chat step
• Keep answers tied to the source text to lower errors and rework

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 will need accounts with Google Drive, OpenAI and Pinecone. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, create a Google Drive credential. Double click the Google Drive node, choose Credential to connect with, click Create new credential, then follow the on screen steps to sign in and grant file access.
4. Create an OpenAI API Key in your OpenAI account. In n8n, double click the Embeddings OpenAI and OpenAI Chat Model nodes, click Create new credential, paste the API key, and save.
5. Create a Pinecone API Key and environment in your Pinecone dashboard. In n8n, open both Pinecone nodes, click Create new credential, and enter the API key and environment values.
6. In your Pinecone project, create an index with 1536 dimensions. Use a metric that fits your embedding choice. Make sure both Pinecone nodes point to the same index.
7. Open the Set Google Drive file URL node and replace the sample link with the shared link to your document. Ensure the file is accessible to the connected Google account or set sharing to Anyone with the link.
8. Check the Recursive Character Text Splitter settings. Keep chunk size near 3000 and overlap near 200 for long PDFs, or adjust for shorter notes.
9. Click Test Workflow to run the loading path. Confirm the Google Drive node downloads the file and the Pinecone insert node finishes without errors. If you see a dimension mismatch, confirm your embedding model uses 1536 dimensions and your index matches.
10. Open the chat panel or use the chat trigger endpoint to send a question. The agent should embed the query, retrieve chunks from Pinecone, and return an answer grounded in the source text.
11. If the chat returns empty answers, verify the Pinecone retrieve as tool node is connected to the agent as a tool and that the same index name is used in both Pinecone nodes.
12. If Google Drive download fails, check file permissions and that the link is a direct file link. For repeated loads, note that clear namespace will reset prior vectors on each run.