Welcome to the Seven-Figure Dumpster Fire

It seems obvious that blindly tacking AI onto core strategy is a questionable bet, and yet here I am, feeling obligated to write about it. It's one of those duh-type statements that still somehow isn’t duh enough, especially as we’ve witnessed the repercussions of more than one big-bet AI investment gone wrong.

Even the biggest players aren’t exempt from poor decision-making, and reality has finally come home to roost: those pitching the “AI will completely replace humans in the contact center” theory have since pulled a full 180, claiming that AI was intended to enhance human agents all along. But their publicly traded shares are still catching up to the messaging switch.

Putting the (AI) Cart Before the Horselink to this section

Certain high-profile AI initiatives suggested you could dramatically reduce your customer service headcount, and sales and satisfaction would just keep going up and up. Lots of companies bought in. Many are now staring at the wreckage of seven-figure projects that can’t even begin to justify their expenses.

They’ve treated AI as a rip-and-replace solution, turning it into a monolithic “it can do it all” type of system. They’ve colored wildly outside the lines of its actual capabilities, asking it to do too much too soon, with no guardrails or alternatives.

What Is Working: Coloring Inside the Lineslink to this section

The prevailing messaging over the last few years has been that we should accept that AI will replace humans entirely. But we’re now seeing that the companies that have been most successful with their approach skipped the aspirational fanfare. Instead, they:

• Started with a small, precise set of use cases
• Stayed within the boundaries of those use cases
• Built seamless handoffs for the moment a customer needed a real person

And, importantly, these companies held on to their people. They decided that there’s an alternative to cutting the workforce in the name of AI efficiency. Take a look at ServiceNow, where they deflected 75% of cases with AI while still not reducing headcount.

They relied on AI to identify and communicate the right signals, and they trusted their people to turn those signals into outcomes. Bots handled the repeatable and predictable, and humans handled the nuanced and relational.

I’m not here to make some Luddite argument about how AI needs to go away. AI can work well alongside humans rather than instead of them, and we need to make that our north star as we figure out our AI strategies.

Change Is Unavoidable, but the Human-Robot War Isn’tlink to this section

Rather than showing them the door, AI can enable us to change people’s roles for the better. The agent who spent their day resetting passwords and reading scripts might instead become:

• A high-value closer
• An empathetic escalation specialist
• An upsell conversation pro
• And definitely, still, the person customers get transferred to when the AI doesn’t work

What could this look like in practice? Let’s use United Airlines as an example. I usually call United if I have a problem; I'm either going to miss a flight or need a different one, etc. In a few years, AI will be good enough that a bot can replace a human in solving these situations on its own.

When my initial problem gets resolved, I'm ready to hang up. But imagine the system is set up to be smart enough to bring a human into that interaction. The bot knows that I’ve recently been on the United website, browsing flights to Bora Bora. And so the system connects me to a person who's actually been to Bora Bora in the last six months.

And they say, "Hey, Bryan, I see you've been looking at Bora Bora a lot; I was in Bora Bora four months ago. Let me tell you what it’s like: where I stayed, where I ate, the tours I did.”

Suddenly, the person who used to be busy solving the same problems over and over becomes a salesperson who's upselling my wife and me to a Tahitian vacation. The agent—the human—is now a revenue-generating powerhouse. They have a different role at United than they used to, but they’re more useful than ever.

Optimism, Not Doomerismlink to this section

To clarify, I don’t think all implementations are doomed to fail because AI is some human-hubris-driven tower of Babel or some such thing.

But the monolithic "does everything" approach does carry real risk. It’s the nine-in-one shampoo approach to AI solutions. It says it can clean your hair, unclog your sink, and fix your car, and it works as a great accompaniment to a rotisserie chicken. But really, all you’re left with is a bottle of junk that doesn’t do anything particularly well.

So for those of us looking at the inevitable conjunction of AI and the product roadmap, there’s something to learn from the cautionary tales in the market. Know what your AI is for, stay in those lanes, and enable your humans to do what only humans can do.

Bryan Martin

Co-Founder & CTO, 8x8

Bryan Martin is Co-Founder and Chief Technology Officer at 8x8, where he leads the company's technology vision across AI, cloud communications, and platform architecture. One of the original architects of 8x8's cloud infrastructure, Bryan has shaped the company's technical direction from its early days as a VoIP pioneer to its current position as a Communications Intelligence Platform. He holds 137 U.S. patents spanning semiconductors, video processing, computer architecture, and communications — and brings that same builder's instincts to how 8x8 approaches AI: specific use cases, clean handoffs, and outcomes that hold up under scrutiny. Bryan received Bachelor's and Master's degrees in Electrical Engineering from Stanford University.