Turn a Google Drive file into a searchable knowledge chat. Team members can ask questions and get clear answers that come from the file, which helps with training, support, and daily work. It suits teams that store guides, policies, or product docs in Drive and need quick answers.

Here is how it works. Click Test Workflow to load the data. The workflow reads a Google Drive file, breaks text into 3000 character chunks with 200 overlap, builds embeddings with OpenAI, and saves them into a Pinecone index. When you click Chat and type a question, the question is embedded, Pinecone returns the most relevant chunks, and the OpenAI chat model writes an answer using that context. The setup uses Google Drive for storage, Pinecone for fast search, and OpenAI for language understanding, which delivers faster answers at scale.

To set it up, you need a Google Drive account, an OpenAI API key, and a Pinecone index with 1536 dimensions selected in both Pinecone nodes. Paste your file link into the Set node, then run Test Workflow to index the content. Next, open Chat and ask questions. Expect shorter search times, fewer context switches, and better accuracy for internal knowledge, onboarding, SOPs, and product manuals.

• Manual Test Workflow trigger loads and indexes your file for search
• Google Drive node downloads the file using a shared link
• Recursive text splitter uses 3000 character chunks with 200 overlap for better retrieval
• Default data loader prepares binary file content for processing
• OpenAI embeddings create vectors that match a 1536 dimension index
• Insert into Pinecone stores vectors and can clear the namespace before each load
• Chat trigger collects questions and routes them to the Q&A chain
• Retriever pulls the most relevant chunks from Pinecone for each question
• OpenAI chat model writes the final answer using the retrieved context
• Set node holds the Google Drive file URL so you can switch sources fast

• Reduce document search time from 30 minutes to 1 minute
• Streamline knowledge lookups by up to 80%
• Improve answer accuracy by using source context, cutting off topic replies by 70%
• Handle 10 times more questions at the same time with vector search
• Connect Google Drive, OpenAI, and Pinecone in one chat flow

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 Drive, OpenAI and Pinecone. See the Tools Required section above for links to create accounts with these services.
3. In the n8n editor, double click the Google Drive node. In the Credential to connect with dropdown, click Create new credential. Follow the on screen steps to connect your Google Drive account with OAuth2. Make sure the Drive file is shared with that account or set to Anyone with the link can view.
4. Open the OpenAI Embeddings and OpenAI Chat Model nodes. For each, choose Create new credential, then paste your OpenAI API key from your OpenAI account API page. Save the credentials.
5. Double click both Pinecone nodes. Create a new credential using your Pinecone API key and environment from the Pinecone console. Save and select the same index in both nodes.
6. In your Pinecone dashboard, create an index with 1536 dimensions. Use a suitable name and region. Confirm this exact index is selected in the Read and Insert Pinecone nodes.
7. Open the Set Google Drive file URL node and paste the share link of the file you want to index. Ensure the link is accessible to the Google Drive credential you used.
8. Check the Insert into Pinecone node options. Leave Clear Namespace enabled if you want to replace older data when reloading content.
9. Click Test Workflow at the bottom of the canvas. Watch the Google Drive, splitter, embeddings, and Insert nodes for successful runs. In Pinecone, confirm vectors were added.
10. Click Chat at the bottom of the canvas and ask a question that is clearly answered in the document. Review the response for relevance and clarity.
11. Troubleshooting: If you see a dimension mismatch error, confirm the embeddings model outputs 1536 dimensions and your Pinecone index matches. If the Google Drive node fails, check file permissions. If answers are empty, verify the correct index is selected in both Pinecone nodes and that the document actually loaded.