autoMate – An AI "Discovery Engine" that finds automation gaps in your business docs

Built with n8n
, , , ,
1

The Inspiration

Most automation tools wait for a prompt, assuming the user already knows what to build. As an AI/ML Expert and Engineering Student at ENSICAEN, I realized that many businesses are sitting on “dead” documentation (PDFs, SOPs, CSV logs) without realizing they are actually blueprints for automation.

What it does

autoMate transforms n8n into an Autonomous Consultant. Instead of manual brainstorming, you feed the system raw business context:

  • Discovery Phase: A RAG pipeline (built with LangGraph and Gemini 3) analyzes your documents to find hidden “If-This-Then-That” logic.

  • Architecting Phase: It uses the Model Context Protocol (MCP) and Context7 to fetch live node documentation, ensuring the generated JSON is accurate and hallucination-free.

  • Execution: It generates a production-ready n8n JSON file that you can import instantly.

Technical Deep Dive

  • Engine: Gemini 3 for high-level reasoning and workflow synthesis.

  • Orchestration: LangGraph manages the multi-agent state, from document grading to final code generation.

  • Connectivity: Real-time documentation fetching via MCP ensures the generated nodes use correct parameters.

Watch the Demo

I Built an AI Agent That DISCOVERS Automations You Didn’t Know You Needed

I’d love to hear the community’s thoughts on using AI as a discovery tool rather than just a builder. Would you like to see this as a native n8n node or template?

4

the discovery angle is what makes this stand out — most tools assume you already know what to automate. feeding it business docs and having it spot if-then patterns is a clever use of rag. curious how it handles ambiguity though — like when an sop says ‘review and approve if appropriate’, does the pipeline flag that as automatable or skip it? also, how reliable is the generated workflow json in practice — do you usually need to tweak it after import or does it work out of the box?