Share product ideas and upvotes with our product team
We need an internal version of Fin that CS and Sales can use to quickly ask product questions without creating tickets or needing to access demo accounts where fin is installed.Today we use Fin in a demo workspace, which adds unnecessary users and overhead. An internal Fin powered through the existing Slack integration would let teams get fast, accurate product answers without triggering support workflows.This would reduce friction, eliminate extra setup, and make Intercom a true internal knowledge assistant.
Many of our articles/guidance/snippets are audience-filtered, and we want to ensure that answers work as expected for those audiences specifically, but when using the Fin preview testing environment, there doesn’t appear to be a way to pretend we are within a specific audience to ensure that the answers are working as expected. This is essential to check our measures are working effectively and without it we’re essentially going in blind and hoping for the best!A good solution would be a simple dropdown within the testing environment that says “I am in X audience” so that only articles/guidance/snippets applicable to that audience are used during testing.
It would be beneficial if, based on escalation rules, Fin could create a back-office ticket to other teams.
I would like to have a tab in the Testing section for Fin Voice that is similar to the one we have now for Fin Chat/Email. Including the Evaluate Answer feature would speed up the time it takes to manage changes to guidance by showing us exactly what content is being used to craft replies. I want to be able to batch test questions from customers that come through over phone calls as there are differences in what customers ask over the phone, and differences in how they respond to Fin from the moment the call starts. It would also help track how the changes made in voice guidance affect how Fin responds over key topics as Fin responds differently over voice.
As we’re testing out the AI Suggestions feature for Fin to improve articles and create new Snippets, I find myself wanting to be able to review a historical record of all content that we have edited using the AI Suggestions feature.I could be missing it but I don’t think I am. Having a hub that shows me a list would be very helpful for my own reference but also for sharing with my stakeholders when reports are due.
A way for AI to summarize cases from a particular user for a period of time. For example, they gave a negative NPS score but left no comments so I would like to have a snapshot of what her support conversations have been about without having to click into each one. This could be a huge step towards customer retention efforts and evaluating the health score of a user or organization.
Turn the chat history into a knowledge base for an AI agent that can search up previous chats. That way it saves the time of searching previous chats if a recurring issue pops up. If it could summarize the problem in that previous chat as well as the solution give + link the chat source itself that would be ideal.
You can set a response time (in seconds) for FIN AI.It responds very quickly; in our culture, customers don't send a single question. Therefore, FIN responds immediately to a sentence, leaving the FIN's response incomplete or inconsistent.
Instead of having the 'was that helpful?' auto-question, create yes or no buttons to click. It might be easier for customers to provide feedback.In general, end-users need to type "Yes," which can take more time during the typing process. Intercom Find Digital Agent Typing process
Hi Team, Is it possible that tables can also be integrated or capture when uploading documents? Thank you!
Right now, Fin can analyze and understand images uploaded through the chat and provide answers through there. That should be expanded to where if a user uploads a file such as a PDF, DOC, CSV, XSLX, etc then Fin can analyze the content of that doc and compare it to the knowledge it’s been given to provide an answer to the users question Example:A user uploads a CSV file asking why the file failed to upload to our system. Fin analyze the content of that CSV file and sees that the headers do not match up with the formatting of the headers we’ve given it. Fin can provide an answer stating that the file failed to upload due to a header formatting and give an appropriate solution based on internal documentation
At the moment, when an outbound message created by API (via third party), the user’s reply can’t be addressed by Fin because 'Conversation starter is: API' is missing. The product team has confirmed there is not any trigger available addressing this specific scenario.
I think it would be a great improvement to have the ability for the AI to be able to respond to conversations before routing them to a team member, either providing more context as to why the routing is taking place. This would be great for some guidance changes that would allow for responses before closure etc.
Part of the rationale around the CX score is to replace the need of CSAT scoring. I don’t send CSAT requests out as often as I used to, but I still like to offer the option. However, when chatting with Intercom’s iteration of Fin, I’m asked to leave a CSAT score every time, which seems like overkill on its own, but it also makes the bot frustrating to use, which makes me use it less and less.
We are having a hard time having Fin AI Agent being able to read and answer questions based on multi-column tables. It seems like this would be a pretty easy thing for an AI to manage understanding.
We would like a singular point within the knowledge base on intercom, where all of our information/data/knowledge/processes are kept in one place. This would include; public articles, snippets, guidance, workflows and tasks, query types/tags all broken down for the information, by customer issues and enquiries.We would like to, for example, have this listed by the query type, where any information lives related to the topic as well as how we would handle it and how Fin/our workflows would handle it. We want this to be in one place so that we can spot duplicates and gaps and have information updated on a master document without having to duplicate the work or update multiple sources.
It would be great if the event of moving a conversation to WhatsApp could exist for workflows, so that we could set up the WhatsApp chat to be greeted with information from the conversation they moved, to keep things more organized and easier for them to find in the WhatsApp chat. Many of our customers migrate a large number of different cases to WhatsApp, making it difficult for them to know what they were referring to in previous chats that were moved to WhatsApp.
We have multiple sources of knowledge for Fin, including confluence, websites, and articles hosted on Intercom. Occasionally, the sync will fail for confluence or a website. I’d like a notification to be sent to the admins of the account when this happens so that we can review and re-sync as needed.Fin relies entirely on these sources to be able to help our customers and agents, so it seems like a big oversight in Intercom that if the content fails to sync nobody is notified.
We are currently monitoring all customer conversations where Fin is involved to ensure an excellent customer experience. While a report already exists that pulls all conversations in which Fin has participated, there is no mechanism to leave the evaluation of the quality of Fin’s responses with real customers.I propose introducing a simple rating system within Intercom that allows us to assess Fin’s performance with three categories: Good, Acceptable, or Poor. This would work in a similar way to the evaluation logic currently used in the Test module.Implementing this functionality would enable us to: Track the accuracy and helpfulness of Fin’s contributions. Monitor Fin’s “health” over time. Compare Fin’s performance across different topics and contexts. This improvement would provide valuable insights into Fin’s effectiveness and ensure we continuously deliver a high-quality customer experience.
Currently if you offer multilingual support in AI, you will need to create duplicate snippets in every supported language. This creates extra work.Having the same multilingual functionality for snippets as articles would be much more convenient. Articles do support multiple language versions within a single article using translated content objects, which makes management much easier.Having snippets work similarly to articles with built-in language version management would definitely streamline the multilingual content workflow.
Many times we receive duplicate conversations from users who are looking for assistance quickly, it would be awesome if Fin could recognize they are opening multiple conversations and maybe mention to the user that they have another chat open and to further correspond there or automatically close out the duplicate and leave an AI summary in one.
Already have an account? Login
No account yet? Create an account
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.