Your SaaS Stack Was Priced for Humans. Now What?

Gartner says $234 billion in enterprise SaaS spending is at risk by 2030. Not because the software stopped working. Because it was built assuming humans would be the ones clicking through it.

I’ve been in enough AI strategy conversations with clients to know this question has come up more than once: what happens to our per-seat licenses when an agent does the work instead? We spend an hour mapping which processes to automate, which headcount to reskill or retarget, which workflows to hand off to an agent. Then someone asks about the ten software subscriptions tied to those workflows, and the room gets quiet. If an AI agent is completing tasks by calling APIs directly and never loading the UI, the link between “active users” and “software revenue” breaks. Gartner calls this agentic arbitrage, and the term is useful. The logic is simple: agents don’t need seats.

Our teams are deploying agents in client environments right now. And I can tell you what the SaaS invoice doesn’t say: it doesn’t say “this license assumes a human will touch the product 40 hours a week.” It assumes usage. When agent-driven usage replaces human-driven usage, some vendors will adapt their pricing model and some won’t. The ones who don’t are going to have a difficult renewal conversation with procurement teams who have now done the math. For enterprises, this is an optimization opportunity. For SaaS vendors building on per-seat economics, it’s an existential question they’re probably not ready to answer yet.

Before you scope your next AI agent, add one question to the project brief: which software subscriptions does this make redundant, reduce, or renegotiate? It’s not a gotcha. It’s a real number. And it belongs in the business case.