Winner Takes Most in AI Support
Intercom's $250M/year AI bet
This is a scale game disguised as a product launch. If one support platform handles enough chats, it learns where AI fails, which answers need hard rules, which customer attributes matter, and which model mix gets the best answer at the lowest cost. Intercom was built around the chat surface itself, and by 2024 had grown from $249M in 2022 revenue to $343M, giving it both message volume and installed workflow to train Fin inside a full help desk rather than as a stand alone bot.
-
The moat is not just the model. Intercom’s deeper advantage is that the bot can read help docs, see account context, hand off to a human, learn from the reply, and update future answers inside one system. That makes every resolved chat useful training data for the next one.
-
The market is likely to compress around a few platforms because pricing shifts from seats to outcomes. Intercom moved Fin from $1.90 to $0.99 per resolved issue and described it as near cost pricing, with the long term margin coming from surrounding software. That is how scale leaders can underprice smaller wrappers.
-
The main alternatives reveal where winner takes most stops. Zendesk has huge distribution and can bundle AI into its installed base, while Gorgias is stronger in Shopify centric ecommerce and already fits usage based pricing better. So the likely consolidation is by workflow and customer segment, not one vendor taking all support.
The next phase is a land grab for the system that owns both automated resolution and the human fallback path. As AI handles more of the easy tickets, the winning platforms will be the ones that turn each remaining human reply into product improvement, lower unit cost, and higher resolution rates, then use that loop to pull more of the support stack onto the same platform.