Your task
All challenges
Challenge 10
Ask Your Google Drive
Build a RAG assistantA RAG assistant searches a selected knowledge source for relevant passages before an AI model writes its answer. that searches a Google Drive knowledge folder before answering and names the files that support each answer.
Your bonus task
For the question "Does the event provide an airport shuttle?", reply exactly "I could not find this in the event documents." and add nothing else.
Example use cases
- An event team keeps venue, volunteer, and sponsor facts in Drive and wants one grounded chat assistantA grounded assistant limits its factual claims to evidence found in the supplied documents instead of relying on general model knowledge. for all three.
- New volunteers need reliable setup answers with the correct handbook filename.
- Organizers want unsupported questions to produce an honest fallback instead of a plausible guess.
Before you start
- 01
Sign up for n8n Cloud or open an up-to-date n8n workspace, then create a new workflow.
- 02
Create or sign in to a Google account, then add a Google Drive connection by following the n8n Google credential guide.
- 03
Create an OpenAI account, create an API key, and store it in n8n using the OpenAI credential guide. API usage may incur a small charge.
- 04
Download the venue guide, the volunteer handbook, and the sponsor logistics file. Upload only these three `.txt` files as direct children of a new Drive folder, then copy its folder ID.
- 05
Use Simple Vector Store only for this workshop demo: it keeps its index in n8n memory, can expose it to other users of the same instance, and loses it after a restart or low-memory cleanup. Re-run ingestion before testing chat after either event, and do not use sensitive documents.
Nodes you'll use
What the workflow must do
- 01
One Manual Trigger run finds and downloads all three direct-child `.txt` files and indexes their chunksIndexing chunks means splitting documents into smaller passages and storing searchable meaning-based representations of them. under `challenge_5_event_docs`, with `file_name`, `file_id`, and `source_url` metadataMetadata is identifying information stored beside each passage, such as its source filename, Drive ID, and link. preserved on every chunk.
- 02
After ingestion, "Where and when should volunteers check in?" returns the North Entrance at 08:00, and "When and where may sponsors deliver materials?" returns Loading Bay B from 07:00 to 08:00; each answer ends with a `Sources:` line containing the correct fixture filename.
- 03
The AI Agent calls the `event_documents` retrieval toolThe retrieval tool searches the stored document passages for the ones most closely related to the user's question. before every answer and, for the bonus test question, returns the exact not-found sentence without a `Sources:` line.
Tip
Need a tip?
Reveal up to five tips, one at a time.
Still stuck? Ask on the n8n community forum
Solution
See the workflow challenge answer
Reveal the completed workflow only when you are ready to compare it with your own.
Submit as solved
Ready to submit?
Submit when your team has a working workflow to demonstrate.