I’ve been back and fort with the support team on this for weeks, and made sure to set up Finn correctly.
I have this very simple instruction asking Finn to not reply to my customers if it doesn’t know the answer to their question. No matter how I phrase it or how many ways I repeat it, Finn will “behave” for a while and then go back to replying, mostly embarrassing nonsense responses.
It’s ridiculous! I keep having to apologize to my customers for the dumb responses it will give to super basic questions.
See below my current instructions - I can’t think of another way to repeat this… I keep adding them.
Is Finn just not ready for real world usage? Or is it just my account that’s cursed?
I can’t keep coming up with different ways to phrase this simple instructions (and the support team can’t either, aparently).
Page 1 / 1
Hi @Adam Tal ,
Thank you for posting and sharing this feedback. I can absolutely see how frustrating that experience with Finn's responses must be, and I want to assure you we take this very seriously.
This is valuable for us and the community. As Finn is a new feature, we are constantly refining how it interprets instructions. One general tip we've found is that consolidating similar rules into a single, authoritative instruction can often improve consistency.
To resolve your specific issue, I've created a direct conversation for you in Intercom where I'll be looking into this personally. Please check your messages there. We'll get this sorted out for you.
Best, Mat
Hey @mateusz.leszkiewicz thanks.
As it stands now, it seems like a bug, and the developers were supposedly notified, but I’ve been assured this is on low priority and unlikely to be solved soon.
The way I see it, looking at Intercoms advert for a million dollar refund if a company isn’t happy with Finn, I’d consider this bug quite a priority 😂.
For now, it seems like my account is “plagued” and the team as no solution for me other than a refund. I’ve suggested migrating the history to a new account, but didn’t get a response to that.
Dave is wonderful and really took the time to understand the issue and offer suggestions, unfortunately, these were kind of taking a step back. I did them anyway and it seems to make matters worse.
I’ve now been redirected to another team member as Dave is unavailable, who suggested maybe I should just give up and do this as a workflow/chatbot without Finn.
I explained that my current use-case is simple enough that it can be achieved with canned responses and a filter in Gmail. The fact that Finn can’t do it, is… uhm… surprising, I guess?
Needless to say, I’m not paying for intercom and Finn for this use-case, the idea was to start with something very simple to test Finn out and then expand to more complex stuff - but so far it’s failing the simple things.
I’m quite baffled to be honest.
Wondering if it’s possible that I’m the only Intercom customer with this issue? (in which case, why not start a new account that doesn’t have this weird rare bug?)
I’ve been offered a 3 month refund, but I’m already all set up and have Intercom integrated, I would really prefer a solution.
Hey @Adam Tal sound frustrating not being able to get it to work for you.
I have something similar working for me in Handover and escalation where it will handover to humans based on Guidance so thought I’d give a try see if I can make it work for you situation.
Using the below under Handover and escalation guidance I can make the AI response to let the customer know it can’t help, then it will handover to a human. This seems to work as it still also the AI to provide a response before handing over. Perhaps this will work for you.
“If you don't have or know the answer to a customer's question, let them know you can't help them with this and route to a human agent who will be able to support them.“
Guidance Applied
You can see below where the Guidance was applied in the conversation alongside some other Guidance in this case. However as it was in test it was not possible to pass to a human/agent.
You might need to review all your order Guidance encase there is conflict in these stopping this working. If you hit issues on it not working suggest disable them all and re-enable one by one starting with the above testing after each one is enabled. You’ll eventually find the one that conflicts, if any and be able to adjust it.
AI applied the guidance in conversations events