The customer is answered
It reads the way your team writes, because that is where it came from. No queue, no wait, and no reason to go asking for a person.
You run Intercom with Fin answering your customers. Automatifie answers from what your team has already told customers, so fewer people push for a person and fewer conversations land on your agents. When the match is not certain, a person takes over first. Fin writes its own replies, and Intercom gives you no supported way to substitute that step. Teams who want what we do move across.
We will get back to you about setting it up on your support history. By submitting you agree to our Privacy Policy and Terms of Use.
Answers that sound certain and are wrong, and no way to catch one before the customer reads it.
“Sometimes Fin provides the wrong information — for example, it will walk someone through troubleshooting mobile app issues even when they're actually on a desktop.”
Quite a lot, and it is worth being fair about it. Guidance steers how Fin responds to particular questions or customers, and Custom Answers let you fix the exact wording for specific questions. But Intercom documents a maximum of 100 active pieces of guidance per workspace, says Custom Answers are hardcoded and not supported alongside Guidance, and states that guidance cannot route a handover to a specific team, tag a conversation, set priority, or take other actions.
Every one of those controls steers Fin. None of them replaces the step where Fin composes the reply, and that step is the entire thing we do differently. There is no hook that says "do not generate this one, use the answer my agent already sent". So we do not pretend to plug in.
Source: Intercom, Fin guidance limits
"How much is it?" on its own means nothing. Read against the chat so far it means "how much is the yearly plan". That is the question we go looking for.
Not a help article. The reply one of your own agents sent to that question, out of your support history. There is nothing for you to write first.
Their order, their plan, their delivery date, read live out of your system and dropped into the reply by code. The model never writes those values, so it cannot invent one.
Is there a close enough answer, and is it about the same thing this customer asked about. A similarity score can only tell you the first one.
It reads the way your team writes, because that is where it came from. No queue, no wait, and no reason to go asking for a person.
Before the customer ever sees a guess. Your agent gets the draft, the history and the reason it stopped.
76-90%what bot vendors advertise41%what really gets solvedHow many customer questions an AI support bot actually settles end to end, next to the number on the brochure.
Tell us what your setup looks like and we will walk you through how this fits, or tell you honestly if it does not.
We will get back to you about setting it up on your support history.