“VCs wanted us to build an LLM. We said no” | John Sabino (LivePerson)

"AI is not the answer to poor CX. AI is not the lever for cost out. It helps you on cost out and efficiency, but it should be used to elevate the experience."

Sep 19, 2025

John Sabino stepped into LivePerson as CEO facing a perfect storm: financial distress, harsh investor criticism, and a company that had "lost its way" trying to build the world's best LLM. Within weeks, major investors were telling him his strategy wouldn't work and that the company wasn't mentioned in critical industry reports.

"You have to have the wherewithal and the courage to say, I'm not doing that. I'm going to try this first."

His response was to refocus on LivePerson's core strength - 30 years of orchestration expertise - while pioneering what became an industry standard: Bring Your Own LLM (BYOLLM).

"Everybody thought I was nuts when we started with that. And now that is the industry buzzword. I'll be blunt. I think we started that."

Why 95% (or is it 60%?) of AI Pilots Fail

The MIT study showing 95% of AI pilots fail. We think it’s closer to 60%, but either way AI pilot failures validates what John has been saying all along. Companies are making a fundamental error in their approach to AI implementation.

"A use case is not a process. A use case is not a full customer journey. A use case is not a solution. That is what many people miss."

The failing pattern:

AI companies grab isolated use cases
They wow boardrooms with resolution rates
They ignore orchestration, multi-channel needs, and escalation paths
Customers get frustrated when complex issues can't be resolved

The Human + AI Formula

John's framework for CEOs navigating AI transformation focuses on strategic augmentation rather than wholesale replacement:

"Don't fire all your humans. Those are the people that should handle complex cases where the package went to the wrong location, where the payment was wrong."

The opportunity lies in creating premium tiers of service. High-net-worth banking customers and serious gamers will pay for white-glove human support. Meanwhile, automate the "high caloric, low value tasks" that make up 45% or more of inquiries.

The Partnership Imperative

Drawing from his GE Digital experience, John believes survival in the AI era requires acknowledging what you can't build:

"Partnerships matter. Unless you're one of the big three, you're just not going to have the resources to build everything for everybody."

LivePerson's partnership strategy today includes multiple LLM providers - through a BYOLLM approach. For John, this was validated when DeepSeek emerged "out of left field" and could immediately integrate with LivePerson's platform.

Pricing in the Age of AI

John's approach to monetizing new AI capabilities rejects the "fail fast and free" mentality:

"I don't believe in doing a pilot for nothing because if it's free, it doesn't have value or a basis to establish value."

His framework:

Start with 3-4 early access customers
Charge professional services fees initially
Understand value creation through real usage
Then determine pricing model (seats, volume, or outcomes)

The goal is outcome-based pricing, but you need data first:

"In order to price any metric or unit that you're going to use to establish value, you need to know the outcome that you're driving."

Leadership lessons from crisis

John's military background (Ranger School, Airborne) shaped his crisis leadership philosophy:

"You want people in your organization that have the grit, that know how to take the objective with less. If they fail, they dust themselves off and say 'next one, let's go.'"

His framework for CEO decision-making:

Always consider shareholder value
Answer to your customers
Take care of employees
Make financially responsible decisions
Never shy away from tough choices

The Gartner Magic Quadrant Journey

Responding to investor criticism about not being in key reports, LivePerson made a concerted effort to improve their position. The achievement wasn't just about targeting the report: