Build a Gumloop flow that reads and responds to Intercom chat messages using AI, trained on past resolved conversations for automated customer support

[Tim1]

Is it possible to do the following:

I’m trying to set up a system where:

1. A customer sends a message through my website’s Intercom chat.
2. That message is sent to Gumloop, where an AI model (like Claude or GPT-4) generates a response.
3. The AI-generated reply is automatically sent back into the same Intercom conversation, so the customer receives it in real time.
4. The AI should be trained on or have access to past resolved support conversations to give context-aware, accurate responses.

Is this possible using Gumloop + Intercom + AI (Claude/OpenAI)?
And if so, what’s the best way to connect these tools and train the AI?

I’m having a fair bit of trouble and I can’t find a lot of resources on this - Is it even worth using Gumloop for this?

Thanks in advance!

[Gumloop_Bot]

Hey @Tim1! If you’re reporting an issue with a flow or an error in a run, please include the run link and make sure it’s shareable so we can take a look.

1. Find your run link on the history page. Format: https://www.gumloop.com/pipeline?run_id={{your_run_id}}&workbook_id={{workbook_id}}

2. Make it shareable by clicking “Share” → ‘Anyone with the link can view’ in the top-left corner of the flow screen.
   

3. Provide details about the issue—more context helps us troubleshoot faster.

You can find your run history here: https://www.gumloop.com/history

[Wasay-Gumloop]

Hey @Tim1! We don’t have a native integration with Intercom just yet, so this would likely be out of scope for now.

[Tim1]

Hey Wasay!

Thanks for getting back to me so quickly! Are there any workarounds with custom nodes?

[Wasay-Gumloop]

Yeah, you could create a custom node with Intercom, but to be honest, it’s going to be a bit challenging. All automation workflows are stateless—meaning they don’t have any stored memory—so you can’t really use a workflow like a proper chatbot that goes back and forth or references past conversations. There are definitely workarounds, like storing responses and data in a database or spreadsheet and using that as a reference, but depending on what you’re trying to do, a dedicated chatbot platform might be a better fit here.

[Tim1]

Okay great - Appreciate your response Wasay, and that makes a lot of sense. I’ll have a look at some chatbot platforms.

Have a nice day.

Tim.