When AI Agents Close the Deal, RevOps Still Has to Pay (and Build For) the Humans

When AI Agents Close the Deal, RevOps Still Has to Pay (and Build For) the Humans

ElevenLabs went from $0 to more than $600 million in annual recurring revenue in 41 months. Carles Reina, the company's first revenue hire, has described a detail that matters more than the growth curve itself: when an AI agent closes revenue on an account, the human account owner still gets paid commission on it.

That detail was shared in a recent conversation with Reina covered by SaaStr, and it points to a design choice every RevOps and founder team building AI-assisted sales motions will eventually have to make.

The Real Lesson Isn't the AI, It's the Compensation Design

Paying a human commission on revenue an AI agent generated looks inefficient on paper. In practice, it is a deliberate way to prevent a predictable failure mode: sales teams that quietly slow-walk or distrust automation they see as a threat to their own quota. By keeping account owners compensated on agent-assisted revenue, the incentive shifts from resisting the tooling to using it well.

Reina's team paired that structure with an aggressive quota and uncapped commission model, which only works if the underlying targets, territories, and account scores are built on solid analysis rather than guesswork.

What This Means for RevOps Leaders Building a Hybrid Motion

• Compensation design.Comp plans need to explicitly account for agent-assisted revenue, not just human-originated deals, or they will quietly punish the reps who adopt the tooling fastest.

• Attribution modeling.Attribution has to separate what the agent actually did (research, scoring, first outreach) from what the human did (relationship-building, negotiation), or the data behind future comp decisions will be wrong.

• Quota and territory design.Quota-setting needs account-level data, not top-down targets, especially when agents are changing how much pipeline a single rep can realistically carry.

The Analytical Work Hiding Behind Every AI-Plus-Human Sales Motion

The headline is always the AI. The actual lift, in every version of this story, is analytical: account scoring models that decide which accounts an agent should work first, TAM and segmentation work that defines territories fairly, and attribution frameworks that make a comp plan defensible once agents are doing part of the selling.

This is the layer that determines whether a hybrid motion scales cleanly or creates comp disputes and rep churn six months in, and it is rarely where an early-stage RevOps or founder team has spare capacity.

Three Questions to Ask Before You Redesign Comp Around AI Agents

• Can you currently separate agent-assisted revenue from human-originated revenue in your CRM data, or would you be guessing?

• Does your quota model account for the fact that agent-assisted reps may realistically carry a different pipeline load than before?

• Do you have account scoring or TAM segmentation rigorous enough to defend the comp decisions you are about to make?

Get the Analytical Backbone Right Before You Scale the Hybrid Model

Building the account scoring, attribution, and quota models behind a hybrid AI-and-human sales motion is exactly the kind of focused analytical work that does not need a full-time hire to get right. Tell us what you need and Gratia will match you with a growth and RevOps analyst in days.