Data Doesn’t Make the Call: What the Brewers Can Teach Us About Human Intervention in Voice AI

On October 15 at Dodger Stadium, a team with a $146 million payroll finished off a team spending $339 million. And it wasn’t close. It was a sweep. 

The Dodgers led in every one of those four games. The Milwaukee Brewers came back every time. Now they’re headed to their first World Series in 44 years, and when Game 1 opens in Milwaukee they’ll face the Tampa Bay Rays, another club that has made a habit of beating teams with deeper pockets. 

After the clincher, Brewers manager Pat Murphy put it in one line: “They may have a bunch of Hall of Famers, but we have a bunch of gamers.” 

I’ve been selling conversational AI for a long time, going back to speech recognition and intent engines years before anyone was talking about LLMs. That quote stuck with me because it gets at something every CX leader should be thinking about: Data can guide the play, but people still need to make the call. 

Here’s what I want you to notice. The Brewers didn’t get here by ignoring analytics. Milwaukee has a sharp, analytically minded front office. They got here because their manager knows exactly when the data should guide the decision and when a human needs to make it. 

That’s the same question voice AI is forcing on every customer service operation: When should the AI step aside and let a person take over? 

Analytics Gets You Through 162 Games. Feel Wins in October. 

Murphy has been clear about this. Data is incredibly helpful over a grueling 162-game season. But in the postseason, he manages with a feel for the game. Sometimes that means leaving a pitcher who’s dealing in to face a tough opposite-handed hitter, instead of going to the bullpen just because the script says so. 

Your contact center has its own 162-game season. Password resets. Order status. Billing dates. Appointment changes. AI should own a big share of that volume, and it gets better at it every quarter. 

But every customer journey also has postseason moments. The homeowner filing a claim after a fire. The small business owner whose account just got frozen. The patient staring at a bill they don’t understand. Those are the moments that decide whether a customer stays with you, and they’re the moments where a script isn’t enough. 

Customers already know this. Gartner’s 2026 customer service research found that 87% of customers believe companies using generative AI for customer service must give them a way to reach a human agent. 

That’s an important distinction. Customers aren’t rejecting AI, the same way Brewers fans aren’t rejecting analytics. What they reject is getting stuck: a manager pulling a pitcher in the middle of a gem because a spreadsheet said so, or a voice bot that won’t let them out of the loop. 

So the goal shouldn’t be maximum containment. It should be successful resolution with the least amount of customer effort. 

Find out where AI can create value without compromising the customer experience. 

Explore LiveNexus AI Review & Roadmap

Don’t Let the Metric Become the Manager 

One of my favorite stories from this Brewers run: the analytics staff brought Murphy data on “pull air percentage,” meaning how often hitters lift fly balls to their pull side. His answer? “Don’t bother putting that in.” 

Conventional wisdom treats ground balls as low-value outcomes. Murphy saw a blind spot. In 2026, MLB hitters batted .205 on pulled grounders and .415 on grounders hit the other way. Since Murphy took over in 2024, the Brewers have led the majors in wins, stolen bases, singles, opposite-field hits, and ground-ball hits. They won the pennant with 154 home runs, the fewest by a pennant winner since the 2015 Royals. 

“It’s like golf,” Murphy said. “Sometimes you need the wedge.” 

In voice AI, containment rate is our pull air percentage. It’s a useful number. It’s a terrible manager. If every escalation counts as a failure, you’ll end up with a system that works hard to keep customers away from people, and that’s exactly the wrong outcome in healthcare, insurance, and financial services, where a wrong answer costs a lot more than a frustrated caller. 

The fix is to make human intervention intentional instead of reactive. Good managers know before first pitch which matchups they’ll never hand over to the model. Good CX leaders should too. 

Routine requests, simple transactions, and predictable workflows can go to AI. Sensitive conversations, complex decisions, regulatory issues, and emotionally charged situations get a human. Set clear boundaries on what the AI is allowed to decide, and build the escalation paths into the workflow from day one, not after the first bad call. 

Human intervention shouldn’t be the exception you struggle to accommodate. It should be part of the game plan. 

The Handoff Is a Pitching Change, Not a Do-Over 

Murphy still keeps his own old-school scorecard in the dugout, full of shorthand, grading every at-bat in real time. His reads usually line up with what the front office values. He just wants the data delivered in the language of a baseball mind. 

That’s what a good handoff looks like. When a reliever jogs in from the bullpen, he already knows the hitter, the count, and what’s been working. Nobody makes him start the inning over. 

Your customers deserve the same. Someone who spends several minutes explaining a problem to a voice AI system shouldn’t have to repeat the whole thing after being transferred. Good orchestration carries the context forward, gets the customer to the right expertise, and gives the agent what they need in a form they can act on immediately. 

That takes coordination across technology, customer journeys, operational processes, and the human capacity behind them. It also changes how you keep score. Beyond containment and automation rates, look at whether customers are reaching resolution, how well escalations are handled, and whether the overall experience is getting better. 

A well-timed pitching change isn’t a knock on the starter. And a smooth handoff isn’t a failure of automation. It shows the experience was designed to adapt. 

Build an AI strategy around real customer journeys, operational readiness, and measurable outcomes. 

Build Your AI Roadmap

It’s About Who You Have in the Room 

When Murphy took the job, he insisted on defining how the team would play. “We’ve got to be different,” he said. “And it’s all about the who we have in the room.” Or, even more directly: “It’s not about numbers. You need people in the room like Sal Frelick.” 

Small-market teams can’t buy their way out of mistakes. They need an operating model where the analytics and the people actually work together. 

Most companies aren’t the Dodgers of AI spending. The good news is you don’t have to be. You need a clear view of where automation fits into your broader customer experience strategy before you scale it. 

That’s the purpose of LiveNexus AI Review & Roadmap. Built on more than 26 years of customer experience expertise, the engagement looks at your existing workflows, identifies the high-value automation opportunities, assesses operational readiness, flags governance gaps, and pinpoints where human involvement remains essential. 

What you get is a prioritized roadmap tied to your business objectives, your customers’ needs, and what it will take to execute in the real world. 

Because putting voice AI in place is only part of the job. Building an operating model that knows when to automate, when to intervene, and how to deliver consistent outcomes is what wins. 

Hall of Famers or Gamers 

As voice AI gets more capable, it will be tempting to measure progress by how many conversations the technology handles on its own. That’s a bit like measuring a ballclub by its payroll. 

The organizations that lead in customer experience will keep a better scorecard: Did every interaction reach the right outcome? 

That means getting past the AI-versus-human debate, the same way the Brewers got past the analytics-versus-gut debate. The front office and the dugout aren’t competing. They’re on the same team. 

The best voice AI strategy isn’t the one that takes humans out of the game. It’s the one that knows exactly when to put them in. 

Ready to move from AI experimentation to a more intentional customer experience strategy? 

Get Started with LiveNexus AI Review & Roadmap