Share product ideas and upvotes with our product team
Current problem: As of now, when Fin resolves a conversation it’s not possible to assign it to a corresponding team (team based on language, product, etc.) This affect reporting and also it’s harder to navigate when ALL Fin conversations are “unassigned”. Solution: Using workflows, enable that after Fin resolves a conversation it’s possible to automatically assign it to the corresponding team.
It would be highly beneficial if we could train Fin using the batch test page. Currently, we can rate its answers, but this does not contribute to its training. Allowing each rating to train Fin enhances its performance before user availability.
When reviewing Fin accuracy I am surprised with Fin’s “choice” or articles.For example a user is asking how to export some data, which is a common topic for which we have an article. Fin answers sourcing the “How to delete your account” article and says “Sorry I don’t have data .. “Rather than creating a snippet which we then have to maintain, it’d be much better an easier if the feature “improve answer” had an option to unlink and re-link the source for that specific answer.
I’d like to request a setting that allows teams to toggle the way Fin references help articles — specifically, the option to disable the small reference number format and instead display the full article title as a clickable hyperlink or list it clearly at the bottom of the conversation.We’ve noticed that Fin now includes article references as small superscript numbers at the end of a sentence, requiring customers to hover over or click to see the linked article. While this may streamline the chat visually, it creates a usability issue for our customer base.Why this matters:Most of our customers are Gen X and Boomers, and our team is concerned that they may not intuitively understand that the small number is a clickable link. This could result in customers missing important resources and increase their reliance on live support.Suggested solution:Introduce a setting to choose how article references appear in Fin: Show full article titles as clickable links within the response OR list source articles at the bottom of the conversation OR allow teams to toggle between reference number format and traditional linking based on their audience We’d love to see this added so we can better tailor the experience to our customers’ needs.Thanks for considering!
We would like to create guidance the can reference specific folders that we have set up for snippets. For instance, “if a client is asking about forms, first ask them which type of form they are using. If they say they are using a X form, only respond based on snippets from the X folder and if they respond with Y form, only respond with snippets from Y folder.”Same idea as well with Knowledge content, if we could use Guidance to have it only source articles from a specific section (church management, Giving, apps, etc) that could really help.
Intercom have advised that there is no way to see the AI response and edit it before sending to a customer.We respond to complex questions, so we want to improve AI’s reponses, we can’t be the only company who wants a choice imbetween the one to not use AI and let it send messages to customer without review.
It would be great to have a feature in the "Analyze" portion of Fin AI Agent, to mark that I've "Reviewed" a conversation. Or maybe even leave notes for myself and other managers to see outside of the context of the conversation, only in the context of the “Fin Performance” or “Optimize.” Thinking like UI green check marks I can select when I’ve reviewed something, or flag a conversation to revisit and work on later. We've been reviewing all of our Fin "sent to teammate" to see how we can get more of those resolutions from Fin. Sometimes the answer is perfect, and the customer just doesn't read... And I'd like to somehow notate that I've "Reviewed" a conversation so I don't review it again in the future
It would be great if the suggestions could be filtered or manually grouped so that subject experts only have to review the content that relates to them. The current system is only effective for if one person is able to review them all.
Translate CSAT feedback into default workspace language. Right now, we’re translating the conversation but then when a user leaves feedback in their native language, I still need to use a third-party tool to translate the feedback to understand it.
I'd like to submit a feature request on FIN AI Agent Suggestions. It would be great if it could be possible to sort the list. A use case would be the ability to sort suggestions based on number of conversations used to generate the suggestion so we can begin or focus on optimization efforts that would have the most impact.
Would love to see Intercom add a Quality Management tool (similar to EvaluAgent, MaestroQA, Scorebuddy, Balto, etc.) using the power of Fin. All the key features of a tool like this such as Quality Standards/Attribute creation, Manual Scorecards, Reporting, Calibration, etc. but with the ability to train fin on the standards and forms to do auto scoring to the standards. This could be incredibly powerful when combined with the CX score...tells the story from the voice of the customer (CX) and from the voice of the business (QA). @Arlene Trompczynski @Gerald Prado
We use an external AI software to tag and escalate conversations… this means we’re linking Intercom to that external software.We’d like to see conversations auto tag based on machine learning from Fin and keep it all within the Intercom ecosystemAlong these lines, we’d like to import our NPS data into Intercom so then the tagging can happen on the feedback as well. For HIPAA compliance, we track feedback from NPS pretty closely, especially in the (extremely rare) case a customer reports an injury - because we need to document that accordingly for FDA purposes. Having Fin be able to 1) tag an interaction based on machine learning from our interactions and then 2) have a workflow where if a certain tag is used, an action happens like automatically filing a back-office ticket
When the “Participant is Teammate” option is selected for a conversation Topic, the text and content of Fin’s answers should be included when looking for keywords and assigning topics.We use Fin on a majority of our cases, and typically include teammate responses as part of topic identification. This means that at the moment, not including Fin as a teammate means topic identification does not work properly for a majority of cases.Users will often make spelling mistakes or use a very large variety of non-standard words and phrases, which do not get picked up by the Topic assignment ML algorithms. However, Fin can use context to understand the customer’s message, then identify and restate the topic in a way that is standard with our help center content, which would then be caught by topic tagging. This is currently a large gap that is pushing us to look for topic identification/AI categorization tools outside of Intercom, instead of being able to use the readily available tools we’re already paying for.
It would be nice if we could configure the Summarize Conversation option in the AI Compose feature.For example, what headings to include in the summary etc.Apologies if this is already a feature. Thank you!
We’re a SaaS platform based in Sweden with a large Nordic user base. To support our users better, we’ve enabled multilingual support in Fin. However, I’ve noticed that Fin often confuses Norwegian and Danish. In most cases, when a user writes in Norwegian, the response is generated in Danish.While speakers of Scandinavian languages (Swedish, Norwegian, and Danish) can generally understand each other, this mix-up leads to poor user experience. It would be a significant improvement if Fin could better distinguish between Norwegian and Danish. For now, I will have to disable one of the languages, likely the one with fewer users, to prevent confusion.
Problem to Solve:The AI Generated CSAT reports are AMAZING, and there’s so much information, but it is a little cumbersome to get an understanding of the overall reason the AI generated score was given without digging into each of the individual conversations. Request: Would be great to be able to add a AI Generated Summary to the report that gives a combined overview of why it provided the score that it did. 🎥 Video ExplanationAmazon Example:
In the beta testing area of Fin, I’d like to be able to point it to specific article when it provides the wrong answer or content. Sometimes it pulls similar content but is the wrong answer and the only option is to change the article or add a snippet. I want to be able to point it to the correct article rather than right a snippet or change the article it chose to use.
Currently, it appears Fin only applies a maximum of two Guidance items per response, despite multiple pieces being relevant. This limitation reduces the accuracy and completeness of Fin’s replies, especially when Guidance is written according to best practices—split into smaller, focused pieces. The behaviour contradicts the idea of modular Guidance, as only a subset is ever used.Problem:We've followed Intercom's recommendation to create granular and objective-specific Guidance. However, we're consistently observing that Fin only references two items per reply, even when three or more are clearly relevant. This behaviour reduces answer quality, especially for complex queries. It feels like a soft cap is being enforced despite there being no documented limit.Suggestion:Lift or increase the current restriction on the number of Guidance items Fin can reference per response. Alternatively, provide a setting to configure how many pieces of Guidance can be applied or surfaced dynamically. Transparency around the prioritisation logic would also help in optimising Guidance strategy.
Currently, it appears Fin only applies a maximum of two Guidance items per response, despite multiple pieces being relevant. This limitation reduces the accuracy and completeness of Fin’s replies, especially when Guidance is written according to best practices—split into smaller, focused pieces. The behaviour contradicts the idea of modular Guidance, as only a subset is ever used.Problem:We've followed Intercom's recommendation to create granular and objective-specific Guidance. However, we're consistently observing that Fin only references two items per reply, even when three or more are clearly relevant. This behaviour reduces answer quality, especially for complex queries. It feels like a soft cap is being enforced despite there being no documented limit.Suggestion:Lift or increase the current restriction on the number of Guidance items Fin can reference per response. Alternatively, provide a setting to configure how many pieces of Guidance can be applied or surfaced dynamically. Transparency around the prioritisation logic would also help in optimising Guidance strategy.
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.