"AI agents will use existing software 100 times more than people, making a shift in software business models to headless inevitable."

Aaron Levie (아론 레비), CEO of collaboration software company Box, highlighted this in a recent post on social media platform X (Twitter). He said in an era when AI agents use software more than people do, software can be used far more than it is now.

In software, headless refers to software operating without a user interface (UI). Existing software was structured for people to use while looking at a screen, but in a headless setup it runs only through APIs or commands, without a display.

Existing software business models set fees based on the number of seats within a company, meaning the number of users. Levels of software use were also limited by how much work a person can process in a day. For this reason, most software was not used as much as its actual processing capacity.

Agents can change this structure. Levie said, "AI agents handle multiple tasks in parallel 24 hours a day and link work across multiple systems. They process entire sets of contracts at once rather than reviewing them one by one, run 10 times more marketing campaigns, and speed up customer onboarding where bottlenecks formed at stages requiring human involvement."

He also said, "Systems of record such as Salesforce can be used more than 100 times through agents. They can be used more for customer targeting and sales automation, and can also be used to convert documents into structured data to automate other workflows."

Levie said software business models will also shift in this environment. He forecast that a structure will take hold in which seat-based fees are charged for people and usage-based fees for agents, with API and agent consumption emerging as new revenue sources.