Use n8n to answer questions from company files

They built multiple LLM assistants in n8n to query company data, call AutoCAD Forge APIs, and use parsing, embeddings, and retrieval.

Done with n8n· Advanced· Mar 2026

What they did
In a company n8n project, they wired up several LLM assistants to work across internal data. The setup handled AutoCAD Forge API calls and used parsing, embeddings, and RAG to answer questions from company information.
How it went
It was good enough for a lot of the work, but the results were uneven. They ran into painful API handling, assistants that mixed up which Supabase to use, and extra human effort for revised engineering drawings and SMP schemes.
Tips
n8n can be a quicker place to start than a Python build, but that does not mean the workflow will be easier to maintain. If you need reliable API logic or document handling, expect more cleanup and testing than the setup phase suggests.

Try it yourself with n8n

Set up an n8n workflow that can answer questions about [company data source] in [chat app]. First, ingest the files, split them into useful chunks, create embeddings, and store them in [vector database]. When I ask a question, retrieve the most relevant sources and draft an answer with clear citations. Before you do anything that would change, send, delete, or submit data, stop and ask me to approve the exact action. Stop after [three test questions] are answered correctly and the sources are shown.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.