AI Agents Aren’t Killing SaaS – They’re Killing the Old Subscription
In my earlier post on how AI agents might be “killing SaaS,” I argued that the real risk isn’t software disappearing—it’s value shifting away from thin UIs toward data, workflows, and domain intelligence. This same lens helps explain why so many people now feel “done” with subscriptions, even as software revenue keeps compounding.
On the surface, the story looks bearish: households and IT teams are hitting hard subscription fatigue, running audits to cull overlapping tools and dormant seats. Boards are asking why they’re paying enterprise prices for products that are basically glorified forms on top of cloud storage. Yet if you zoom out, overall SaaS and cloud spend continue to rise; what’s changing is investor and buyer tolerance for blunt, per‑seat, all‑you‑can‑eat pricing models that ignore how work is actually performed.
AI agents act as an accelerant. When an agent can execute end‑to‑end workflows across CRM, support, billing, and docs, the number of human “users” stops being a meaningful unit of value. A single agent might replace dozens of marginal seats while generating far more output. In that world, charging by seat feels arbitrary, while charging for jobs completed, tasks automated, or resources consumed feels natural. That’s why we’re seeing a steady shift toward hybrid pricing—smaller base subscriptions plus usage‑, token‑, or outcome‑linked components.
For investors, the takeaway is not that subscription is dead, but that subscription‑only is losing its edge. The future compounders will be SaaS platforms that (1) own clean, auditable systems of record, (2) ship agent‑native APIs and domain‑specific copilots, and (3) blend recurring revenue with metered, “jobs‑done” pricing that lines up with how AI actually creates value.
