Share product ideas and upvotes with our product team
When Fin is deployed on email, the setting "Prevent Fin from processing emails that fail authentication (DMARC)" only checks the authentication result when the message reaches Intercom. It ignores the ARC seal that a forwarder adds.Many teams don't receive email at Intercom directly. Their support address is a Google Workspace or Microsoft 365 mailbox that forwards to Intercom. The forward changes the server that delivers the email, so SPF fails for every sender who doesn't sign with a DKIM key for their own domain. Most Microsoft 365 customers on default settings are in that group. Their DMARC result then fails, and Fin skips the email, even though it was genuine and passed authentication when it reached our mailbox.For us, 26 genuine customer emails were skipped in 10 days. Turning the setting off isn't a good option, because it's the only thing stopping Fin from answering forged emails.Request:1. Let a workspace mark its forwarders as trusted (for example Google Workspace or Microsoft 365, or specific domains). For mail from those forwarders, Fin should use the authentication result recorded in the forwarder's ARC seal. This is how Gmail and Microsoft 365 themselves handle forwarded mail.2. As an alternative or addition: when an email fails the check, let Fin ask the sender for the one-time code before it runs any data connector, instead of skipping the email.Related: "Add more ways to fix: Fin did not try to answer because the sender could not be authenticated".
Content import sources created through the AI Content API (sync_behavior "api") have no Settings or Target tab in the UI. "Open settings" is greyed out for them in Knowledge → Sources and in Fin AI Agent → Train → Content. For website sync sources, the same menu opens Settings → Target → Audience control, where an admin picks a default audience and adds audience rules.Support confirmed this is expected today: for API-synced sources the default audience can only be set through the API (audience_ids and apply_audience_to_existing_content on the content import source).Why it matters:Many teams split the work. One team creates content in an external tool that syncs it into Fin. Another team owns Fin audiences in Intercom and never touches that tool. Today that audience owner can only change audiences page by page, and every new page from the source arrives as "Everyone". An app that syncs content has to request audience scopes, pull audience names and IDs, and build its own audience picker just to cover this. That goes against least privilege and moves audience decisions away from the people who own them in Intercom.What we'd like: the same Target / Audience control for API-synced sources as for website sync sources, so workspace admins can set the default audience and audience rules in Intercom, while the app keeps managing only the content.
On Human in the Loop escalation rules (Ask for input) on chat, which is currently in beta, users who chat in to our Intercom see 'Let me check with the team. I'll be back shortly.' Can we customize this wording? It appears that it is hardcoded. I would like to be able to customize this wording. When the agent on our side receives Fin’s draft, I want the agent to be able to edit Fin’s draft directly instead of making suggestions for edits via chat. That would make it a lot easier to be precise. I also want Fin to be able to record that edit automatically so it understands and can learn how to tune its AI responses to our edits.
Summary: We'd like the ability to grant teammates access to create and edit Fin Procedures without also giving them visibility into or access to Workflows, automation settings and other sensitive operational areas of our Intercom workspace. Current Behavior: Today, Procedures live within the Fin AI Agent tab, which is accessed via the "Can view Fin and Automation settings" permission. This same permission also exposes Workflows and Simple Automations. There is no way to grant procedure-building access in isolation, any teammate who can work on procedures can also see workflows. To prevent teammates from editing workflows, we can withhold the "Can manage Automation settings and inbound Workflows" and "Can manage outbound Workflows" permissions, but those teammates will still have full visibility into workflow configurations, which is not ideal from an access control and operational security standpoint. Requested Change: Introduce a dedicated permission (e.g., "Can manage Fin Procedures") that:Allows teammates to create, edit, and manage Fin Procedures Does not expose Workflows, Simple Automations, other automation settings or operationally sensitive areas of the workspace Can be granted independently of any workflow-related or data / security related permissionsUse Case: We want to involve subject matter experts and content contributors in building and refining Fin Procedures, teammates who understand our processes deeply but should not have access to the broader automation and workflow configuration of our workspace. Today, there's no clean way to enable this without overexposing sensitive operational settings. Desired Outcome:A permission model that lets us confidently delegate procedure authorship to a wider group of teammates, with clear boundaries that protect workflow integrity.
Currently, important contextual information stored in User Notes and Company Notes is not automatically surfaced within a conversation when it is handed over to a Support team.Support agents may need to manually navigate to the user's or company's profile or surface the fields on top of all the data already available in the details section to review existing notes, which adds unnecessary steps and increases the risk of missing important information.This is particularly challenging when notes contain relevant customer context, known issues, special handling instructions, or other information that Support agents should be aware of before responding.The AskIntroduce a workflow action that allows us to automatically retrieve User Notes and Company Notes and insert them into a conversation as an internal note when a conversation is handed over to a human Support agent.Ideally, this functionality would allow us to: Add a workflow action to retrieve existing User Notes, Company Notes, or both. Automatically post the retrieved information as an internal note within the conversation. Trigger the action at a specific point in the workflow, particularly before or during handover to a Support team. Clearly distinguish between User Notes and Company Notes in the generated internal note. Ensure that the information is visible only to internal teammates and not to customers. Example Use CaseWhen a customer interacts with Fin and requests human support, the workflow automatically retrieves any existing User Notes and Company Notes associated with that customer.Before the conversation is assigned to a Support agent, an internal note is generated containing the relevant information.Example Internal Note:Customer Context (Automatically Generated)User Notes: Customer prefers email follow-ups. Previously reported recurring issues with a specific feature. Customer only works Monday to Wednesday. Snooze conversations outside of working days if follow up is needed. Company Notes: Customer has a known ongoing issue currently under investigation. Special handling instructions have been documented for the account. Currently on escalation due to migration issues This ensures that every Support agent receiving the conversation has immediate access to important customer context without having to manually look for it. Expected Benefits Improved Support Efficiency: Reduces the time agents spend searching for existing customer information. Better Context During Handover: Ensures important user and company information is consistently available within the conversation. Improved Customer Experience: Reduces the need for customers to repeat information already documented. Greater Consistency: Provides all Support agents with the same contextual information across conversations. Reduced Risk of Missing Information: Helps prevent important customer notes or handling instructions from being overlooked. Desired OutcomeThe ability to configure workflows to automatically surface existing User Notes and Company Notes as internal conversation notes, particularly during handover from Fin to human Support.This would help create a more seamless AI-to-human handover experience and ensure Support teams have the context they need directly within every conversation.
Today there isn’t a way to Test Responses for Co-Pilot. It would be valuable to have a testing capability that allows us to evaluate responses and identify any content changes needed to improve them.
AI Inbox Translation currently covers conversation thread messages, but ticket properties/content (including the auto-generated merge summary and description) are excluded. Since Fin already generates these summaries, it would help support teams if they were translated into the teammate’s working language the same way thread messages are, especially for cross-region teams triaging merged tickets.
We use Fin’s multilingual workflows to support users in several languages, including languages our team doesn’t speak. This works very well during the conversation, but creates a significant gap when a conversation is handed over from Fin to our team.Currently, Fin’s AI-generated conversation summary is provided in the user’s original language. This means an English-speaking teammate working in an English-language workspace could receive a handover summary entirely in Japanese. In these cases, the summary cannot fulfill its intended purpose: helping the teammate quickly understand the conversation and take over effectively.This is particularly noticeable because Fin already allows our team to view the conversation itself in our workspace language, regardless of the language used between Fin and the user. We’d like the AI summary to work in the same way.Ideally, AI summaries should be generated or made available in the workspace language, rather than only in the user’s original language. This would make multilingual Fin workflows much more effective for global support teams and ensure the value of AI summaries carries through from automated support to human handover.
Problem:Once a content recommendation is accepted or rejected, there's no history to show what was suggested, what was decided, by whom, and when. We also can't assign a recommendation to a teammate to work on.Why it matters:We need to show which changes were made to Fin's knowledge and why, and to tell whether a rejected suggestion keeps coming back. Without assignment, recommendations rely on one person checking them, which doesn't scale across a support team.What we'd like:A history view of every recommendation, with its outcome, who decided, when, and a link to the resulting change. The ability to assign a recommendation to a teammate, with a status, the way a ticket works.
Problem:A guidance rule or procedure that is paused or in draft on the live version can't be changed inside a release. It doesn't show up there as something you can edit or set to active. As far as a release is concerned, it doesn't exist until someone switches it on in the live version, outside the release.Why it matters:Releases are how we want to ship changes to Fin: bundled, tested with evals, ramped, and reversible. A new rule or procedure is often the change itself. It's the change we most need to stage and test, and the only way to ship it right now goes around the release.What we'd like:Draft and paused guidance and procedures shown in a release, where they can be edited and set to active. They would then go live when the release publishes, with the same pause and revert as everything else in it.
Problem:Every conversation started in the Fin preview lands in the live inbox as unassigned, alongside real customer conversations. Our support engineers work from that view, so each test conversation looks like a customer waiting for a reply. It's frustrating for the person testing as well. After every preview session I assign the conversations to myself just to get them out of my teammates' way. That's an extra manual step on every test, and it's the opposite of what preview is for.Why it matters:It distracts the team and discourages testing. Test conversations pull attention away from real customers, and when you know someone will see each test, or each test means cleaning up after yourself, people test less.What we'd like:Preview conversations kept out of the inbox entirely, or assigned automatically to the person running the preview, or put somewhere separate that team views don't include. The person testing should be able to go back to their own preview conversations without anyone else seeing them.
Right now, the billing date for a subscription can’t be changed from the workspace settings. It would be really helpful to have the option to adjust the billing date so it better aligns with our internal billing cycles and finance processes.Allowing admins to update the billing date directly in the subscription settings would make planning and budgeting much easier.
Hi team,We recently started evaluating the AI Inbox Translations feature, and we're very interested in using the inbound translation functionality. It has the potential to save our support team a significant amount of time that is currently spent translating incoming messages.However, we're not entirely satisfied with the quality of the outbound translations. We'd love to be able to enable inbound translations only, while keeping outbound translations disabled.The reason is that errors in inbound translations generally don't affect our business, as we can still understand the customer's intent. On the other hand, outbound translations are not shown to us before the message is sent, so we have no way of verifying whether the translation has altered the meaning or context—which unfortunately happens from time to time.Would it be possible to support this kind of setup, where only inbound translations are enabled?Thank you.
Fin AI can already read PDF and pictures. We receive +20 excel files via email per day, and the Fin AI response depends a lot on the content. Could you please enable Fin to read other file types. Thank you.
When chatting with the Intercom team through Fin, it would be helpful if Fin could submit feature requests directly to the Product Wishlist instead of sending us here to input manually (a new procedure please! 🙂)Right now, feature requests need to be created manually in the community. Allowing Fin to capture and submit requests during conversations would make it easier to track product feedback and ideas in real time.This would be especially useful when customers or teammates suggest improvements while discussing product behavior or limitations.
We have a custom dashboard to report on Fin’s performance, and one ready-made chart we found useful for that is the “Used content performance” chart. It would be great to have this same chart for used guidance. I know I can see guidance performance by going into the guidance tab, but it would make things easier to have this in the same place as our other reporting. That would then also allow us to export this data along with our other reporting.
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.