The hidden cost of Fin AI

Intercom’s Fin AI is without a doubt one of the industry leaders for agentic customer experiences. However, if you haven’t had the opportunity to work with it before, you might not fully understand how crucial content and knowledge management is to its entire engine. In fact, knowledge is Fin AI’s only engine.  

You can’t draw water from an empty well, as the saying goes. 

What this means is that without real human writers on the backend feeding content into the machine and training it through all the available features, the chatbot will stagnate. If you have not taken a deep dive into agentic experiences with your enterprise yet, this might seem exaggerated. I assure you that this is a fact. It was something my team lived week-by-week. 

Stale content means the AI bot resolution rate drops, while constant input and updates keep it steady. Intercom’s CSMs will always counsel all new customers to keep up with content creation for the best chatbot output. 

Intercom’s Fin 3 launch blog reported that across 6,000+ customers, the industry average AI resolution rate sits at 66% (and let’s be honest; human Support Agents can exceed a 66% resolution rate easily). Top performers who exceed 80% AI resolution rate are distinguished by the fact that they are using the features available to them to the fullest. This translates to active knowledge management, and an understanding of how to train the bot to deliver the best results possible. You need a team behind the scenes who understands how the gears are turning, and what keeps them oiled. 

So, this is where we are; this is what I know now. If your company is investing in Fin AI or any agentic experience (including building and integrating your own), you need to be prepared to invest more money, not less, in humans who understand technical content, knowledge management, and AI optimization.