Intercom wants to help companies help their customers quickly, efficiently, and reliably. At Pioneer, its AI customer service summit last week, the AI-first customer service company announced some significant updates to Fin, its AI customer service agent. Intercom's Chief Product Officer, Paul Adams, shared what's new and what's improved.
In March 2023, Intercom announced Fin, its AI customer service agent. It started out supporting information queries, but the company made continual updates that saw the agent become much more. Adams said that resolution rates grew by 52% a year later when Fin 2 was announced. Fin 2 brought what Intercom describes as human quality service - Adams said it acted like the best performing customer service representatives - and was able to connect to third-party systems. The Fin CoPilot, an AI assistant that supports customer service agents, was also announced at the same time.
Today, Adams says, Fin's average resolution rate across Intercom's 6,000 customers is 66%. But we should expect to see that rise with Fin 3.
As you'd expect, Fin is custom-built for customer service. Intercom talks about a Fin Flywheel with four distinct stages: Train, Test, Deploy, and Analyze. Keep these stages in mind as we talk about what's new.
As part of the Train stage of the flywheel, Fin 3 includes Procedures, which enables companies to set Fin up for complex queries. With procedures, Fin is trained on everything human reps do.
There are four parts to creating procedures:
The second big update involves simulations. This is part of the Test stage of the flywheel, providing a new automated testing capability. Adams said it can fully simulate a customer conversation from start to end.
You can simulate a procedure by selecting the segment or audience, adding questions, and providing context. Once you do that, you can debug and edit the procedure until it works as expected. The AI assistant can help with simulations, offering suggestions and implementing changes in a single click.
There's also a library of simulations. If you hit "run all", you can test all your procedures and find out if they are working correctly.
The third part goes with the Deploy stage of the flywheel. Fin can now be deployed to Slack, allowing direct replies in a channel, which should help resolve issues faster. You can also deploy Fin to Discord. Discord is often used by high-growth crypto and gaming companies to support customers.
Several enhanced capabilities and insights were also announced.
Voice gets an upgrade with more power and customization, leading to an improved voice experience. Along with enhanced guidance, customization, improved testing, and deployment, you also get comprehensive transcripts and summaries.
Coming to the fourth stage of the Fin flywheel, Analyze, there's another big update. Earlier this year, Intercom added analytics tools, but new updates further enhance those analytics. For example, now you can see reasons for a CX Score, so you can dig deeper into what actually resulted in getting that score. The Topics Explorer uses AI to analyze and organize conversations into topics and sub-topics (which you have full control over). Plus, there's a Topic Trends Report that lets you get ahead of emerging issues.
Finally, AI-powered suggestions get an update, including identifying duplicates or contradictions in content and offering suggestions to improve the content. It also learns from what you accept or reject. It also works with Zendesk and Salesforce, offering one-click suggestions for data, actions, and guidance changes.
To give some market insights into the AI transformation in customer service, Nick Clark, Partner and Director at the Boston Consulting Group, took the stage at the Pioneer summit. He says that every year, customers spend 14 billion hours on the phone or chatting with customer support. To help improve these experiences, 90% have invested in AI, and 85% plan to invest more in the future.
However, only 28% are achieving profit and loss (P&L) impact in service. So, it's clear AI is having an impact, but it's taking time. The most significant part of AI transformation efforts, however, is not in the algorithms or the tools and technology, but in people and process change. New structures, new ways of working, and new definitions of success are required.
In his work with companies that are finding success, he says five things stick out as critical:
Clark argues traditional customer service is frustrating because companies are dealing with limited resources, but AI changes that by opening the door for pre-emptive and proactive service. There are three questions customer service leaders need to ask to inspire their teams to come up with new support systems and models for customer service, he suggests:
Most people dislike interacting with customer service because it takes so long and can be very frustrating. If bringing AI into the mix helps transform that experience for the better, then I'm all for it.
The new capabilities for Intercom's Fin will go a long way to helping customer service organizations create better experiences across all channels. But it's Intercom's vision for what comes next that is even more interesting. That story is coming next.
