Ask a Google Drive PDF questions in chat

Ask a Google Drive PDF questions in chat

An n8n workflow that lets you chat with a PDF. One run reads a PDF from Google Drive and stores it in Pinecone. After that, you ask
questions in n8n's chat window and an OpenAI model answers from the pieces of the PDF that match.

What is inside

• workflow.json: the workflow as n8n exports it, 16 nodes, three of them sticky notes. It holds no credential. Two notes are
  coloured frames that group the nodes, and one holds short try-it steps.
• README.md: these notes.

What we changed from the original copy

• The Google Drive, OpenAI and Pinecone nodes were linked to credentials in their author's n8n. The links are gone, so select your
  own.
• Google Drive ignored the link in Set Google Drive file URL and always downloaded one file picked by hand. It now reads the
  link from that node, and the link is a placeholder, <GOOGLE_DRIVE_FILE_ID>, that you replace with your own.
• OpenAI Chat Model had no model set, so it used n8n's default, gpt-5-mini. It is now set to gpt-5.6-terra.
• Both embeddings nodes had no model set, so they used text-embedding-ada-002. They are now set to text-embedding-3-small. It
  makes vectors of the same size, 1536, so the index setup does not change.
• The try-it note now has a step for the Drive link.
• The Chat Trigger's webhook id is gone, and n8n makes a new one when you import the file.
• Nothing else.

Before you start

• An n8n instance, on n8n Cloud or on your own server.
• An OpenAI API key.
• A Google account that can open your PDF, linked to n8n through its Google sign-in.
• A Pinecone account, an API key, and an index with 1536 dimensions.

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Add an OpenAI credential. Select it in OpenAI Chat Model, Embeddings OpenAI and Embeddings OpenAI2.
3. Add a Google Drive OAuth2 credential and select it in Google Drive.
4. In Set Google Drive file URL, replace the whole value with the share link of your PDF. In Drive, open the PDF, choose Share and
   copy the link.
5. Make an index with 1536 dimensions in Pinecone. Add a Pinecone credential in n8n, then select the credential and the index in
   Insert into Pinecone vector store and in Read Pinecone Vector Store. The two nodes must use the same index.
6. Press Execute workflow to load the PDF into the index. Check in Pinecone that the index now holds records.
7. Open the chat from the button at the bottom of the canvas and ask a question about the PDF.

How a run goes

Loading, started by hand:

1. When clicking 'Test Workflow' button starts the run and Set Google Drive file URL holds the link in a field named file_url.
2. Google Drive downloads the file named by that link.
3. Insert into Pinecone vector store takes the file through Default Data Loader as binary data. `Recursive Character Text
   Splitter cuts the text into pieces of 3,000 characters that overlap by 200. Embeddings OpenAI` turns each piece into a vector.
4. The insert node has Clear Namespace on, so every load first empties the namespace and then writes the new pieces.

Asking, started from the chat:

1. When clicking 'Chat' button below passes your message to Question and Answer Chain.
2. Vector Store Retriever asks Read Pinecone Vector Store for the closest pieces. Embeddings OpenAI2 turns your question into
   a vector first. The retriever returns four pieces, which is n8n's default.
3. Question and Answer Chain gives those pieces and your question to OpenAI Chat Model and returns its answer in the chat.

Change the look

• Which PDF it loads: the link in Set Google Drive file URL.
• How big a piece is: the chunk size and the overlap in Recursive Character Text Splitter. Load again after a change.
• How many pieces the model sees: the Limit field of Vector Store Retriever.
• The model: the model field of OpenAI Chat Model.
• The instructions the model gets: the options of Question and Answer Chain.

Good to know

• Clear Namespace: it is on, and no namespace is set. Each load empties the default namespace of the index before it writes. Use an
  index that holds nothing else, or set a namespace in both Pinecone nodes.
• Models: OpenAI's deprecations page, read on 12 November 2026, lists gpt-5-mini for shutdown on 11 December 2026 with
  gpt-5.6-terra as the replacement. The chain has no tools, so Mindket's copy uses gpt-5.6-terra. The page does not list
  text-embedding-ada-002. The new embeddings model gives vectors that do not mix with the old one's, so load the PDF again if
  your index already holds vectors from the old model.
• One PDF at a time: a load replaces the earlier one. To keep several PDFs, use a namespace for each.
• Answers: the model answers from the pieces it is given, so a question about something outside them may get a weak answer.
  Check anything that matters against the PDF.
• Your data: the text of the PDF goes to OpenAI and to Pinecone, and each question goes to OpenAI.
• Not run: nothing here was run against Google, OpenAI or Pinecone. These notes come from n8n's source and the vendors' pages, read
  on 12 November 2026.