Turn your website into a voice help desk that answers questions using your own documents. A voice agent collects the question, looks up the right information, and sends back a clear answer. Great for customer support teams that want fast, accurate replies without adding more staff.

Here is how it works. A webhook listens for a question from the voice agent. Your Google Drive files are loaded, split into small parts, turned into embeddings with OpenAI, and saved in a Qdrant vector database. An AI agent uses an OpenAI chat model and a vector store tool to search Qdrant and craft a helpful response. The workflow then returns the answer to the voice agent, which speaks it back to the visitor.

To set it up, you need accounts for ElevenLabs, Google Drive, OpenAI, and Qdrant. Replace the Qdrant URL and collection name in the nodes, connect your Drive folder, and paste the n8n webhook URL into your voice agent tool with a question field named question. Expect faster replies, lower support load, and round the clock help for FAQs, product info, and store details.

• Webhook listener receives the question payload with a question field from the voice agent
• Google Drive nodes find the folder and download files to ingest content
• Document loader and token splitter prepare text for search with clean chunks
• OpenAI embeddings convert text into vectors for semantic search
• Qdrant vector store saves and searches the vectors for the best matches
• AI agent with an OpenAI chat model and memory composes a clear final answer
• HTTP requests create and refresh the Qdrant collection before loading data
• Respond to webhook sends the final answer back to the voice agent

• Reduce first response time from minutes to seconds
• Automate up to 70 percent of common support questions
• Improve answer accuracy by pulling only from approved files
• Handle many visitors at the same time without extra staff
• Connect voice, documents, and AI in one flow for faster help

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 ElevenLabs, Google Drive, OpenAI and Qdrant. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, open any Google Drive node, choose Credential to connect with, click Create new credential, then follow the on screen steps to grant access to your Google account.
4. In any OpenAI node, create a new OpenAI credential, paste your API key from the OpenAI dashboard API keys page, and save.
5. In the Qdrant Vector Store nodes, create a new Qdrant credential. Enter your Qdrant host URL and API key if your instance requires it, then save.
6. Open the HTTP Request nodes named Create collection and Refresh collection. Replace QDRANTURL with your Qdrant base URL and COLLECTION with your target collection name. If your Qdrant setup needs an API key header, add it in the node credentials.
7. On the Google Drive Get folder node, pick the Drive and select the folder that holds your source documents. Confirm the Download Files node pulls the files you want indexed.
8. Review the Default Data Loader and Token Splitter nodes. Adjust chunk size and overlap if needed to match your document length and detail level.
9. Open the Qdrant Vector Store1 node and confirm it points to the same collection. Ensure it receives embeddings from the Embeddings OpenAI1 node and documents from the Default Data Loader node.
10. Open the Webhook Listen node and copy its production URL. In your ElevenLabs agent tool, add a tool webhook with method POST, include a string property named question, and paste the n8n URL.
11. Open the AI Agent node and make sure the input text is set to {{$json.body.question}}. Confirm the Vector Store Tool is connected and points to the Qdrant Vector Store.
12. Click Execute on the workflow and run the manual trigger path to create and refresh the Qdrant collection, then ingest your Drive contents. Test your voice agent using its test feature and verify you receive spoken answers.
13. Troubleshooting: If you get no answer, confirm the webhook received a question field. If search fails, check Qdrant URL and collection name. If embeddings fail, verify your OpenAI key and rate limits. If files do not load, check Google Drive permissions and selected folder.