Hey - I believe you’d need a separate Intercom organisation for this, and even then it sounds a bit like the wrong approach. Unless your employees are spending a large amount of their time in your product (and therefore are always one click away from asking Fin a question) I would recommend building from wherever they already spent the most time.

My experience is that would typically be your chat tool like Slack, Teams or Discord. I’ve had a short search (not an endorsement, just an example) and something like Runbear would connect one of those three platforms to Confluence, and allow your team to search for information Confluence - and would likely be a lot easier than having to work out how to prevent customers from accessing internal information! 

You know your own company and business needs better, but that approach would be my suggestion!

Conor

​@Sujai Arakali Good news, I believe you’ll be happy to test out Copilot here for your internal team -- Intercom’s Copilot feature does exactly this, it helps your team look up and craft answers based on all your resources both internal and external. 

Team members already have access to a certain number of Copilot uses free each month and then you can grant seats to team members who need unlimited usage for an additional monthly fee.  Very good for onboarding new team members especially so they can get answers to their questions without needing to ask other team members about the answers as they are getting up to speed etc. 

Then customers have access to Fin AI which only uses the external sources you’ve added to the system and not the internal resources in this case.  Check your Knowledge Hub on how to connect internal resources there’s a whole category for this information.  You can connect tools like Notion as well for internal documentation to feed Copilot.  

Hope this helps!

​@Sujai Arakali, to follow up on Nathan’s answer, Copilot is definitely the way to go here. Use the Knowledge Hub to import your internal documentation and enable it of copilot only (not AI agent, which is your public facing Fin).

I just tried Copilot and it seems like the right tool, however it needs to be very much improved. I tried a few questions and it could never find content from conversations. (Our knowledge hub is limited, but conversations data is large)

So I went through the tedious process of exporting all conversations to Amazon S3, downloading them locally, splitting them in chunks of about 1MB (wihtout breaking conversations in the middle), uploading them to NotebookLM, and the difference is not even close (and it uses 3 years of data, not 4 months).

I can ask any questions such as:

What are the top 5 feature requests from the last 6 months?
Which users requested Feature X?

Or even much deeper reports like:
If you were the CEO of a SaaS company with churn rate of X, monthly new subscribers count of Y, and considering all the conversations, which would be the 5 priorities for the company?

I plan on using it to autogenerate Knowledge Hub articles from the Questions and Answers in conversations

I would add to the other answers with our use case information. If you’re talking internal employees that aren’t your Customer Support Team, it can be costly to assign them a full inbox seat for them to utilize copilot. We actually found it best to host all of our internal content and information in Confluence; since we’re in the Atlassian suite of products. And Confluence has an AI ASK feature that works extremely well for getting answers to internal employees. And confluence syncs with Copilot as a data source. So my Support Team members with inbox seats can get the full use there, but I don’t have every employee in Intercom with an inbox seat, and have a lot of our important product info and everything stored internally.