Stop Token Maxing — Start Outcome Maxing

Hey, everyone. Welcome back here to Techstrong TV. My next guest is Lexi Reese.

Lexi is the CEO of a company called Lanai, and we are happy to have her on here. It's her first time on Techstrong TV. Let's welcome her.

Welcome, Lexi. It's nice to have you here. Thank you, Alan.

Who cannot be excited with that level of enthusiasm? Well- I'm so excited to be here. Thank you.

I've been in tech 35 years, startups mostly, and the thing I learned, and I'm sure you know this as a CEO too, people feed off of passion. Mm-hmm. Right?

When you could exude passion, people feed off of that, and they get passionate. So it's something I learned starting up or co-founding a bunch of companies. Passion's important.

Always. If you're not passionate about what you do, find something else. Anyway, Lexi, speaking about my career, let's segue to your career.

Talk to me about what you've done in your career leading up to becoming CEO here at Lanai. Yeah. So I actually co-founded and I'm CEO of Lanai, but I have spent my career building infrastructure products.

That was after I actually started in documentary film, and I did a film about adolescent sex workers living in Managua who- Mm ... had gotten to that point in their young lives because they didn't see a way to help their families make money. And I think the through line of my career has been how do I help people who are in situations like that find a way to make money without having to turn to something so dramatic?

Went to business school. I worked at American Express, at Google, and then at a company called Gusto, and in all of those- You know it well ... I was building sort of financial rails, payroll rails, a lot of quiet infrastructure that changes how people work or get paid or market themselves, and Lanai is my next chapter, which I know we're going to talk about.

Absolutely. Early on, I don't think it was here at Techstrong, might've been the company before that I started, I think we were a Gusto customer. Thank you.

With Gusto. Yeah. Gusto.

Yeah. Small businesses do it with Gusto. It was when we were really tiny.

It was a small startup. Actually, it was the predecessor to Techstrong, MediaOps. Okay.

Yep. So this goes back to, I'm going to guess 2013, 2014, something like that. Yeah.

I would've been there. I started in around that 2015 time when they had- Yeah ... about 10 million in revenue, and I left when we had about half a billion.

Yep. Yeah. Crazy.

Great story. So where did the passion for Lanai come from? What drove that?

Yeah. So I had spent a year after I was at Gusto with an investor that had invested in Gusto. And the chairman of that firm, it's called General Catalyst, the chairman of that firm is Ken Chenault, who was the CEO of American Express back when I had been there.

So talk about following your passion. Mm-hmm. We talked about that, but also following people that you really- Yeah ...

respect and have high integrity. And with Ken at General Catalyst, I spent a year studying the future of work. I was already intrigued about how AI was going to change work, and this was 2021, '22 timeframe.

And I kept asking business owners, given the advent of AI, let's imagine it's five years from now, how is your workplace going to change? And what I kept hearing was people talking about how AI was going to change their products, but I didn't hear how people were contending with how much AI was going to change their workforce. And now, fast-forward, every workplace has two workforces.

They have the human one they can see, and they have the AI one that they really can't, as it relates to doing work in their enterprise. So we created Lanai because I had a passion for, if you can't see it, how are you going to make decisions about what AI should do and what humans should do that are going to be durable and that frankly, you're going to be proud of? com or you can follow me on LinkedIn at Lexi Reese, to see all of AI-mediated workflows, to steer investments about them, and to scale AI in your workforce in a way that you're really proud of.

I love it. I think you're right. A lot of people focused on what AI is going to do for my products, not how AI is going to morph my workforce.

Yes. It's interesting. Just this weekend, we were talking off-camera about my weekend, but I was watching TV.

Was it Netflix or Peacock? I saw this documentary- Mm-hmm ... about a guy around AI.

Him and his wife were about to have their first child, and he was worried that his child would not have a job when they grow up. Yeah. But it was amazing.

He was able to get Modi and Sam, all the CEOs of Anthropic and OpenAI to actually come on here and talk about these things, as well as the founder from Google DeepMind, too, one of the founders. Sure. So it's pretty interesting.

But certainly, AI's impact on the workforce has become front and center. Right. That and data centers- Yes ...

are the two big hot buttons around AI today. Right. You guys recently did your 2026 AI Labor Report.

Well, let's first ask, what do you mean by AI Labor Report exactly? Yeah. What we mean is AI is already doing real work inside of companies, and Alan, you know this, but most firms still can't tell what it did, what it produced, or what it cost until after they get the bill.

And that creates- Mm-hmm ... a business problem, not just a technology one, because companies are using AI like labor, but they're budgeting for it like software. And we wanted to give some real evidence on exactly how much labor AI is doing.

I love it. And by the way, I'm probably going to take that saying you just said. Say it again for us.

They're using AI like labor, but budgeting for it like it's software? Yeah. That's exactly right.

All right. And I get it. I get it.

The whole phenomenon of token maxing, which I understand as a CEO why we encouraged it. Some of us are still encouraging it maybe, but why we need maybe a different sort of lens to look through this. Yeah.

We talk about the end goal was never to token max. That just sounded fun, I guess. Mm-hmm.

And all people had-- Let's go back. 2023, OpenAI hits 100 million users. By the end of that year, it's a boardroom conversation, no matter what business you're in.

" Use more AI. " Mm-hmm. And what we started Lanai to do is to be able to have people outcome max, because that's the real goal.

Ultimately, as a CEO, I'm indifferent as to what choice I make to get something done. Do I use a person to do it? Do I use technology?

If I'm a CEO, I probably care that I have employees who are gainfully employed this generation and the next. So I don't mean indifferent in that way. I just mean, given options, I want to choose the technology or the person, or the combination of technology and person, that drive the biggest impact for the business.

Right. Get the job done. Get the job done.

So I think there was, and we go through this in tech sometimes, there was just sort of a la-la thinking of if people token max, good things will follow. And as it turns out- Mm ... that is definitely not always the case.

So what we do know, and when you have the CEOs of Microsoft and Uber, who have much bigger budgets than most of us do, saying, "Wow, these costs are astronomical. I need to make major changes," you know you have to switch your operating model. So Techstrong is not Microsoft or Uber, in case you were wondering.

But we went through our own little thing here where, look, I really pushed the entire company to go agentic. I wasn't crazy about it. We set up some guardrails and platforms and so forth, but I didn't care whether you used OpenClore or one of the variants of that, or now you have the Codex and these things in Coworker or Copilot or Perplexity Computer.

Go use whatever agent you like. Mm-hmm. Just make sure you follow our guidelines on securing it as best you can.

Use it as much as you want. I don't care. The only thing I ask is every week, give me a report of what you're doing, what worked, what didn't work, what'd you learn.

Right. And how many people are at Techstrong? We're about 25 people only.

Right. So you can get through with that show and tell approach for only so long because you're small, and it probably isn't even now scaling very well. So here's the lesson I learned.

First of all, no matter what I do, some people pay lip service to the CEO and are using an agent, but they're not really using the agent. Mm-hmm. They're not getting anything out of it.

Mm-hmm. I've got a few rock stars whoreinvented so much of the workflow of our company and automated it using these agents, and we have incorporated those into our workflows and our day-to-days. Mm-hmm.

Out of 25 people, I'd say five people- Right ... probably did 80%, and I was one of the five, to be honest. Yeah.

But we probably represent 80, 85% of what we did with the agentics, and that's okay. Right now, about a month ago, I started looking at, okay, who's spending what here? And changing our plan.

I can't let everyone just use every agent they want. We've got to pick our agents, so to speak. Yeah, totally.

Alan, what I would say is, not to cut you off, but I just- Yeah ... feel passionate about this point is, it's so understandable why- You know, I'm probably right in the center of the- Yeah ... bell curve, right?

I mean- And most companies actually think they're measuring AI. We studied 200 C-level executives in businesses a thousand and above, okay? So not- Mm-hmm ...

exactly on target but- No ... it rhymes with what has happened in smaller businesses, too. It's just on a smaller scale.

People think they're measuring AI, but they're actually measuring usage. They can count prompts, licenses, active users, but they can't show what business result came out the other side. And even if they know, they don't know that question of, was it 5% of people who produced that result, or are we evening out people's fluency in order to get consistent results over time?

And that's a huge deal. You can imagine, the bigger the company, CFOs don't fund on vibes. Yeah.

If AI spend can't be tied to revenue and margin and productivity or cost savings, it starts to look like a discretionary expense. It starts to look like what it was, which is people having fun experimenting. An experiment.

That's exactly it. " I get it. But Lexi, I've been in technology almost 35 years.

You and me both, brother. Yeah. So we've seen these scripts before.

What does the Mandalorian say? " This is the way. First, we just run hog wild into stuff, not thinking, and then at some point, someone says, "Wait a second.

This has to make economic sense. " And then things get real. We've seen it.

It happened with the internet. It happened with the dotcoms. It happened with the cloud.

This is not new. Maybe it's on a scale that's bigger, right? Yeah.

I think that's a valid point. However, here's where I think this is very different and why I studied it and founded a company- Mm-hmm ... on it.

I think the previous generations of technology automated processes, or they automated- Yes ... things. This is automating cognition, and it actually is doing real work.

The fact that people can't prove it yet doesn't mean it's not editing and coding and drafting and having a real impact on work- Yeah ... that then gets sent to customers that is completely invisible, both on the org chart and on the P&L. That is what is different.

We have never had a situation where software expense was masquerading as labor- Yeah ... totally invisible to any system of record, and being produced and seen in media by the case of viewers who are seeing content that sometimes or sometimes not hasn't been checked by a human. And that, I think is, Alan, the place where I would agree, the dot dot dot of your question is, is this real hype, or is it real?

I think what's hype right now is how impactful truly autonomous agents are. And in our study, 200 of 200 companies said they have zero fully autonomous agents. Zero.

And that- I don't blame them. I can't imagine letting it run wild. Well, however, it doesn't mean they don't believe that these-- And I say AI workforce, and I think of it as AI workers, and so usually in organizations, you say level one, level two, level three.

Mm-hmm. And I think AI introduces A levels. So you have A level one, which is assisting a human at their various levels, and you have A level two, which is truly autonomous.

And in between that is humans. The workflow is being redesigned to either manage or edit or check the work of their-"autonomous coworkers," and that's an operational challenge. And so what this report shows is, wow, okay, I'm not denying AI is doing a lot of work, I just don't budget it well, and I am not able to predict what it's doing in my org chart.

And this is where it becomes really interesting when you get back to the question of what every leader, when they're really being honest, says. They're making decisions because they're pressured to show margin changes. So they're making decisions that are, as has been well reported, in some cases, hollowing out those entry-level jobs, meaning their kids' jobs, and they don't feel great about that.

They know that that's not necessarily the right long-term decision, but they're getting pressured to do that. And if I cut a bunch of CX workers or sales workers and I put agents in, at least for this quarter, my P&L's going to look better because software seems cheaper than a human. But what is really interesting is that now that you have had people like Satya Nadella saying, "Actually, these software expenses are getting very close to human labor costs," now you have people- A little more ...

" Like, "Let's go back-" Yeah ... " And that's where we obviously hope people will use Lanai, because Lanai is built expressly for this. See it, budget for it, account for it in your org chart.

Let me give you two examples. We were talking off camera. I had HVAC issues this weekend, or this last week.

Oh, boy. The first company I called, a very nice guy named Patrick answered the phone. Mm-hmm.

He started taking my information to have a rep call me, and about halfway through my conversation, I realized that Patrick was an AI. Uh-huh. A very well-spoken, competent sort of AI.

Mm-hmm. And I said, "Look, I'm just setting an appointment. " And so he made an appointment with me.

It made an appointment with me. " Mm-hmm. " Mm-hmm.

At 9:30, I hadn't received a call. I called the company. " Yeah.

I said, "But no, he confirmed it," blah, blah, blah. " Mm. So Lexi, to me, they don't have a human doing nighttime dispatch, so they've saved labor costs.

But how much have they really lost? " I called another HVAC company and had them there, so they lost, in this case, a $25,000 job. Yes.

So Alan, perfect example. Large financial services company did the exact same thing. They laid off a big portion of their call center representatives, and we helped them follow the long-term consequence of that after they had gotten so many calls about the equivalent of that.

People were calling to say, "I've had fraud on my card," and the AI was not recognizing that it was fraud on the card. So in the quarter they announced it, in the quarter they booked it, post-severance and go-costs. Okay, this financial services company, which was under pressure, fintech, they looked good.

And then the next quarter they had high- I'm so glad to ... high dissatisfaction. The following quarter, they had a ton of churn.

Alternatively, a company that was working with us before they made that decision did it in an AB control group, which is really the only way you can show real return on investment. So I'm introducing an agent into this group that's going to have supervision by humans, and I'm not introducing anything into control group B. And we're using Lanai to see and steer the investment across the whole set of goals, not just the financial goal.

And lo and behold, their org shape changed, their design changed. Now it is humans working together with AI, and AI is doing certain things on their own that they have gotten comfortable with, they can do without changing CSAT. And I think that's my big message to anybody listening.

When you announce we're going to use more AI or else, or we're going to token max, that's not really the point of AI. The point of AI is we are going to produce our product or service better for our customers, and we see a world where human and AI operating together is the future. But we- I agree ...

" We will not- Mm-hmm ... scale anything to our customers unless both of those things hold true. And imagine the difference in the morale of an employee group that is given the peace of mind by their employer that, yes, your work is going to change, but I am committing to you as I am committing to my customers that it will get better or we will not scale it.

And I have only seen one CEO do that, and I cannot say who that is. I hope I can say who that is. But say that and believe that, and I will...

Again, I wish I could, because that is a company I would bet on all day. I love it. Lexi, I'd love to talk to you about this all day, but we've got people in the waiting room for our next one.

But listen, I love what you're doing with Lanai. Did we mention the website? Yeah.

com. I'm Lexi Reis, and wherever we can help you see and steer and scale high-impact AI investments, we would love to support you because this software is in fact a lot like labor. Come back on soon.

Let's continue this discussion. All right. Good luck with the HVAC.

Thank you. All right. We're going to take a break.

We'll be back in just a moment.