Intercom’s chatbot Fin stands out for its impressive accuracy, speed, and ability to handle complex customer queries with natural, human-like responses. It’s tightly integrated into the Intercom platform, so you can use it right alongside help center articles, custom bots, and live support agents without friction.

Fin’s real-time knowledge base parsing is especially impressive—it can answer questions directly from your docs without needing manual setup or flowchart logic. It also does a great job knowing when to escalate to human agents, which helps reduce customer frustration and maintain high CSAT.

There isn't much to dislike about Fin, and it keeps getting better. The main limitation is Fin’s dependence on well-structured documentation. If your help articles are out of date or poorly written, Fin will reflect that. So it' s not really a Fin issue, but very important when setting internal expectations.

Fin is helping us scale support without scaling headcount. It handles a high percentage of Tier 1 questions—like asset management, and basic troubleshooting—allowing our human agents to focus on complex issues. The result: better resource allocation, faster support, and higher customer satisfaction.

Fin by Intercom offers a high degree of customization, which is a significant advantage. I particularly appreciate the control it provides over the AI's tone, allowing us to deliver tailored responses that align with our brand voice. Furthermore, its ability to structure responses effectively, drawing information from the help center resources it has access to, is impressive. Our support agents have found it so useful that they even leverage our Fin-powered chatbot to find answers for their own customer inquiries, which speaks volumes about its utility and accuracy.

One area that could certainly be improved is the experience with custom actions. We've encountered consistent issues with a custom action that utilizes a POST request; it repeatedly fails without providing clear or actionable error messages. This lack of specific error feedback makes troubleshooting extremely difficult, leaving us unsure of what exactly needs to be fixed. Additionally, the structure of the custom action's body is not very code-friendly, which adds to the complexity and frustration when trying to implement or debug these actions.

Before implementing Fin, our support team spent a significant amount of time answering the same questions — like order statuses, product setup, or integration guidance. With Fin, we're now able to automate those interactions while still delivering clear, actionable answers. For example, we’ve configured Fin to detect edge cases like delayed or lost orders and guide customers through next steps, without needing to involve an agent. Just this week, we are able to hit 60% deflection in our two main channels which is the Chat and ERM (Error Reporting Module) We're also leveraging Fin in combination with custom logic in Make.com and N8N, which allows us to surface dynamic, real-time data like stock availability or damaged item detection. This flexibility has made Fin a key part of our support strategy — not just a bot, but an automation layer we can build upon. The biggest benefit has been faster resolution times and more bandwidth for our agents to focus on complex issues.

What I like best about Fin by Intercom is its incredible efficiency and natural language processing capabilities. It excels at understanding and instantly resolving a significant portion of customer queries, dramatically reducing the burden on our human support team. Ease of Use: From an agent's perspective (when Fin hands off or for Fin Co-pilot), the interface is intuitive and clean. For administrators, the no-code workspace to train and manage Fin's knowledge and behavior is highly user-friendly. The ability to fine-tune responses and set up guidance with natural language makes it accessible even for non-technical users. Ease of Implementation: This is a major strong point. Fin can be set up in under an hour, especially if you already have a structured knowledge base. Frequency of Use: Fin is designed for continuous, high-volume use. It works 24/7 across multiple channels (chat, email, SMS, social) and languages, ensuring consistent, instant support. Ease of Integration: Fin offers excellent integration capabilities. It seamlessly connects with various popular helpdesks and can pull information from different knowledge sources and Customer Support (for the end-user): This is where Fin truly shines. It provides instant, 24/7 support to our customers, eliminating wait times and delivering accurate, personalized answers around the clock.

While Fin is a powerful tool, there are a few areas where it could be improved: Handling of Complex or Nuanced Issues: Despite its advanced capabilities, Fin can still struggle with highly intricate, multi-layered queries that require a deeper level of human understanding or judgment. Number of Features: While the core AI agent functionality is robust, some users might find certain advanced customization options or specific niche features to be somewhat limited compared to highly specialized AI platforms. For instance, while it has good reporting, extremely granular custom reporting might require additional effort Assumed Resolutions" and Billing Clarity: There's a point of contention regarding how Fin counts "resolutions," especially when an AI response is followed by human intervention, but the conversation is still marked as resolved by Fin.

Fin by Intercom allows us to scale our customer support without proportionally scaling our human team. It provides a higher quality of immediate service to our customers while simultaneously empowering our agents to do more impactful and fulfilling work, ultimately leading to a more efficient operation and happier customers

We love the concise answers Fin provides in chats and via email. We noticed our customers prefer fast and concise replies by Fin rather than human responses as they often want just to get an answer to their question quickly rather than wait for a human to reply.

I also enjoy the multilanguage support. Fin replies to customers in multiple languages, which means clients who speak other languages than our team can also get prompt assistance. Additionally, I love a vast list of features to customize Fin responses: custom answers, guidance, suggestions. All these things were something we didn't even ask for, but Intercom added them and they work so well!

I love Fin to 99%, but there is one irritating thing: when you forward an email to Intercom, you cannot control how Fin responds and in what language. Fin considers me, as a sender because I am forwarding it, and it's irritating because if the email is in Spanish, but my language is English, Fin will reply in English.

Providing fast and accurate responses to customers without human interaction, communicating with customers in languages which our customer support team doesn't support.

Fin is impressively accurate at handling routine queries, which significantly reduces the burden on our support team. Its natural, conversational tone means customers often don't realise they're speaking to an AI, which helps maintain a professional and consistent experience.

I particularly value how easy it is to train and refine Fin. Updates to articles and content are quickly reflected in its answers, and the visibility into what it's unsure about helps us proactively improve our support material. It gives us genuine confidence that customers are getting the right information, even outside working hours.

While Fin is generally reliable, there are times when it confidently gives incorrect or outdated answers, especially if the source content isn't perfectly structured. It would be helpful if it could recognise uncertainty more consistently and route queries to a human sooner when needed.

Fin is solving the problem of repetitive, time-consuming queries that previously tied up our support team. By handling these automatically, it reduces our response times and allows our agents to focus on more complex or high-priority issues.

It also ensures customers get consistent, accurate answers 24/7, even outside office hours, which has noticeably improved our customer experience. For us internally, Fin acts as an extension of the team—scalable, reliable, and easy to update. It’s helping us manage a growing volume of enquiries without needing to constantly scale headcount.

We've found Intercom’s AI assistant, Fin, incredibly helpful at Butternut Box, especially in delivering instant, accurate answers to customer queries, which frees up our team to focus on more complex support. Fin’s ability to understand context and provide consistent responses has really streamlined our customer experience.

One thing we'd love to see in the future is Dutch added as a supported translation option, it would be a big win for our Dutch-speaking customers and international team members helping us out.

Fin by Intercom has been a great addition to our customer support at our company. One of the biggest problems it solves is reducing the volume of very basic, repetitive questions our team used to handle daily, like order tracking, delivery times, or how to update account details. Fin now takes care of those instantly and accurately, which means our team can spend more time on complex or sensitive queries where the human touch really matters. It’s been a real time-saver and has improved both team efficiency and the customer experience.

Intercom's implementation of AI is the best I've had the pleasure of working with, and it's only getting better.

Things I like about it -

It's easy to set up - the better your knowledge base is, the better Fin becomes.

It's configurable based on your needs - want to tell it to use a different format of English based on a user's location? No problem.

Want to make sure bugs or log-in issues get escalated straight to a person? Can do!

Have more complex situations that come up rarely, but do come up and can be repetitive? - Create snippets, perfect!

We have consistently been keeping a 70%+ resolution rate with the work we've done, and it's been a huge help as our little level 1 support agent, for a very complex software.

Our fin involvement rate is also very high at 90% - there's very few situations where you can't use Fin in some way.

New staff also frequently use Fin co-pilot when learning, and this can be connected to a variety of different internal and external sources, which makes it a one-stop shop for getting the information you need at your fingertips.

- You can't create different 'bots' for the different Brands you add into intercom.

- There are quite a lot of options to look at - while intuitive once you look into it, it can get a bit overwhelming. I have the benefit of being one of the first people to be able to play with Fin in Australia, so I've adapted as Fin enhancements have grown, but I could imagine how this may be a lot to go over as a newbie - but the staff at intercom are super helpful and I'm sure the implementation to new orgs would be just as amazing :)

- There's no way to teach Fin CoPilot the same way you can with Fin Agent, because each question and answer is only visible to the person using it, it's hard to make sure the information being provided is accurate. It would be handy to have some kind of permission-based option that allowed reviewing these to avoid incorrect education; though in saying that, everyone should always be double-checking information!

Managing fin could be someone's (or two peoples') full time job, don't let AI fool you into believing it's set and forget, you do need to keep things updated, you need to stay ontop of content to make sure it's still relevant, Fin is only as good as the information and time you put in.

It's giving my support team the opportunity to work on more complex queries, freeing them up from the easy, repetitive questions which are still important to users to have answers to, but can be time-consuming.

Fin also allows our team to work on more things outside of support, or further upskill themselves so they are more equipped to answer the more complex queries that come in.

Since its implementation, Fin has been crucial in reducing work overload by providing solid answers to customers' inquiries, triaging and solving simple questions while directing the most complicated ones to the Support team improving the response and resolution times accordingly.

I have not found a feature that I would say I dislike. So far, so good!

It has been solving from simple to somewhat elaborate inquiries and this impacts my day to day workload by bringing the most complicated topics to my inbox.

We are a small team with only 2 people handling support. Since FIN leverages our help center content and recent team member conversations, it has allowed us to have our small team focused on more critical issues and proactive support. We recently got a customer CSAT survey for an issue FIN resolved that said, "the AI bot was the most helpful I've ever used!” Customers are able to get answers immediately vs. waiting for a team member or having to go through a decision tree for answers. It was really fast to set up and I've been able to make adjustments along the way. It's also helping us keep our content up to date by making suggestions of areas that need improvement or are missing information. Previously, we would do periodic audits of our help center, but with FIN, we can now look at what had to be routed to a team member and FIN can tell us areas it didn't have information to leverage.

You have the option to turn on multiple functions of FIN, for example you can just turn on Chat or Email or both. We have both turned on, however occasionally a customer might go into a loop where they chat in and FIN tells them to email support, which isn't the best experience. I've been able to provide additional guidance to FIN to try to avoid this, but there are definitely some trial and error things you'll need to do to ensure the customer experience isn't impacted.

FIN is solving customer issues for us based on our help center content. This is allowing our small team to focus on more critical areas and proactive support, which ultimately will help us with customer retention.

Fin is great at deflecting our current customers inquiries by offering great responses from our robust help articles

Fin sometimes does not understand human language nuances.

Deflection rate has soared to over 40%