Incumbent SaaS Repackaged as Agents
Ayan Barua, CEO of Ampersand, on infra for AI agent integrations
The important shift is that customer support software is moving from a tool an agent uses, to a worker that does the job itself. Intercom did not win Fin by inventing support from scratch. It repackaged a help desk, knowledge base, inbox, and human handoff system into an outcome priced agent, which is why it now looks strategically similar to AI native entrants like Sierra and Decagon despite coming from older seat based SaaS roots.
-
Older SaaS already owned the hard parts that matter in support, the ticket history, help articles, routing rules, inbox, and human escalation path. Adding an agent layer turns that installed workflow into a conversational worker instead of making customers rip everything out and start over.
-
The business model also gets rewritten. Intercom layers Fin on top of its existing platform and charges per resolved ticket, while Sierra and Decagon were built around outcome pricing from day one, at roughly $0.99 to $1.50 per resolution instead of charging mainly by human seat count.
-
This is why the market feels convergent. Sierra behaves like an AI BPO replacement with autonomous chat and voice. Decagon goes deep on complex support actions and integrations. Intercom bundles the same agent behavior inside an incumbent help desk. Different starting points, same destination, software that resolves the issue directly.
Going forward, the winners in support will be the companies that combine agent performance with deep system access and clean human handoff. That favors incumbents that can convert their installed base into agent traffic, and also favors AI native players that can move faster into voice, backend actions, and full BPO replacement.