Third-generation chatbots flip support economics

The big change is that setup work moved from the buyer to the model. Earlier bots only worked after a support team spent weeks writing decision trees, mapping intents, and maintaining canned replies. Third generation bots start with the help center a company already has, then combine those docs with customer context and past support history to answer in plain language, which is why they can get to meaningful resolution rates without a long implementation tax.

• This is why the unit economics change so sharply. A company no longer pays people to hand code hundreds of flows before seeing value. It can point the bot at public docs from tools like Notion, Zendesk, or a help center, start resolving tickets quickly, and pay per successful resolution instead of per seat.
  1 sacra 2 sacra
• The real product is not just retrieval from docs. Strong systems also pull in user attributes, order status, plan tier, and prior conversations, then decide whether to answer, ask a follow up, or hand off to a human. That is what turns a doc reader into an actual support agent.
  2 sacra
• This shift pressures traditional help desks differently. Intercom can bundle the bot with the inbox and human workflow, while Gorgias was already closer to usage pricing through ticket based plans. AI native players like Sierra and Decagon push further, selling a replacement for outsourced support teams rather than another software seat.
  3 sacra 4 sacra 5 sacra

The next step is from answering questions to taking actions. As these agents get better at using backend systems to process refunds, update accounts, and handle voice calls, the help center becomes the training data for a broader digital worker. That pushes customer support software toward outcome pricing, fewer human reps, and a tighter link between support, retention, and sales.