How CEOs are preparing their customers and employees to bring AI agents into the workplace

Companies are investing heavily in AI-powered agents and looking toward a future where chatbots may be the first communication a customer has when they want an answer or want to buy something.

In October, walmart announced an agreement with OpenAI that will enable shoppers to discover and purchase products without leaving ChatGPT. The retailer now has an AI agent in its app that can respond and recommend products.

During Walmart’s November earnings call, CEO Doug McMillon said agent AI will be one of the growth drivers for the company’s e-commerce business. He said the technology “helps people save time and enjoy shopping more.” Walmart announced in January that customers will soon be able to use Google’s artificial intelligence assistant, Gemini, to more easily find and buy products from the retail giant and its warehouse club, Sam’s Club.

These investments in agent software also help employees send emails, summarize notes, and improve overall productivity. All of this puts additional pressure on companies to ensure that this approach actually works for all stakeholders.

At the communications software and service provider’s recent annual customer conference. CalixCEO Michael Weening asked a room of broadband service provider executives if anyone didn’t have enough to do.

“And we turned up the lights and no one raised their hands. We asked some of them if they were sitting around lazily, with nothing to do, waiting for their jobs to be taken away. There were no hands,” Weening said. “The message I always hear from people is, ‘There’s too much to do.’ So my message was how do we make time to do more and how do we add capacity to grow?”

In October, Calix rolled out an AI agent across the platform used by the broadband service provider’s customers. This includes agents helping marketers generate offers for subscribers, customer service representatives improving troubleshooting, directing subscriber questions and interactions to the right people, field technicians helping automate diagnostics and optimizing installations, and more.

From Wiening’s perspective, this should be a welcome support, but he acknowledges that messages from executives at big technology companies about agent AI leading to layoffs are scaring many.

Artificial intelligence is cited as the reason for more than 55,000 layoffs nationwide in 2025, according to data from Challenger, Gray & Christmas. This includes layoffs at major employers such as Amazon, Microsoft, and Salesforce.

That trend is amplified by technology and AI sector executives touting the potential for AI to eliminate jobs across multiple industries. Earlier this week, Anthropic CEO Dario Amodei said in an essay that AI could have a broader shock to the labor market than any other technological advance, eliminating jobs in some industries. “Rather than replacing a single job, this technology functions as a ‘substitute for common human labor,'” Amodei wrote.

All of this is leading to a decline in worker sentiment towards AI. A Mercer poll conducted in January 2026 found that 40% of employees are concerned about losing their jobs due to AI, compared to 28% in 2024.

Wiening said a major concern is the “demonization and anomaly” arising from executives’ messages about AI, which will distract from the technology’s potential. “Agent AI is purely a workflow, and every task in the workflow is an agent,” he said.

Instead, companies need to strive to demonstrate how AI agents are “new teammates that help us do our jobs better,” he said.

There are several ways to try to soften the introduction of AI agents. When Calix rolled out its technology across the platform, it took steps to turn agents into what Weening calls “really non-aggressive, very friendly, Teletubby-like characters.”

“From my perspective, they’re becoming part of your workforce. You consider them part of your team,” he said.

In fact, some companies are starting to include AI agents in their total workforce. Consulting firm McKinsey currently has 25,000 personalized AI agents and 40,000 human employees, according to data shared by the company’s global managing partner during a recent live recording of the podcast “All-In” at the CES trade show in Las Vegas.

This is a similar message shared internally by Weening, whose Calix was an early adopter of Microsoft’s Copilot AI companion.

“My view is that using Copilot ubiquitously has the benefit of data protection, but more importantly, it shows our entire workforce that we are serious about innovation,” he said.

Weening said the company has built more than 700 employee-created agents. Calix also identified 40 workflows that the company believes would have a significant impact on productivity if improved with AI. The company’s IT team then formalized those tools and rolled them out across the organization.

“Are all those people agents changing the world? No, it could be something as simple as a tool to write emails faster, but at least they’re playing with it,” Weening said.

Weening said that all communications about AI make it clear that a balance needs to be struck when it comes to risk and speed, especially when it comes to critical data.

“Some people are very concerned, and others are so focused on acting quickly that they don’t realize the risks,” he said. “I think a lot of people are struggling to find the right way to do that. I think it comes down to having very clear guidelines around protecting data, whether it’s customer data or how our partners manage our data.”

Weening said he acknowledged that jobs will be affected by these new agent-based AI tools. “I heard this saying the other day, and now I keep repeating it: 80% of jobs change by 20%, and 20% of jobs change by 80%,” he said.

In short, his internal message is that these tools will empower employees to take on new tasks as Calix continues its growth trajectory. AI productivity improvements will not continue to double employee growth, but it will still increase exponentially.

“We’re in a stage of disillusionment with AI right now, where everyone is asking, ‘Where’s the ROI?'” he said. “What we have to build is a change mindset within our companies that embraces AI, looks at it pragmatically, and evolves the status quo.”

“We are making great progress in this regard, but it will accelerate at a very fast pace in 2026,” he added.

From the perspective of Everest Group CEO Jimit Arora, there have been several enterprise-level systems that have helped transform the way business is done, from systems of record like ERP, engagement systems like CRM, to insight systems that start leveraging insights and data.

In his words, AI agents are part of a new category of “execution systems.”

“That’s when you use the combination of deterministic machine learning, AI, generative AI, and agentic AI as they are currently defined, and that’s where the value comes in,” Arora says.

Although experiments are well underway toward that future, Arora said he would still refer to this moment as “pre-drug.”

“We don’t have real agency with our agents yet. We’re building agents that can take action, but there are differences,” he said. “We have gained autonomy in a sense, but we have not given them agency.”

But Arora said he expects efforts to do just that to begin in 2026, particularly in the “three biggest use cases” for true agent AI: the software development lifecycle; Service desk applications within human resources, IT, and finance functions. and customer experience.

But as these efforts move forward, Arora said companies need to ensure they avoid what he calls “PTSD”: “process debt, technology debt, skill debt, data debt.”

“If you have the right data and you try to identify a broken process, you end up making the broken part even bigger,” he said. “You also need the right skills, because applying yesterday’s skills to tomorrow’s problems won’t work. And through all of this, technology can be the easy part.”

Still, Arora said he has a warning for CEOs and executives expecting big results from agent AI next year.

“I like to take inspiration from the cloud. AWS came out in 2006, Google Cloud Platform came out in 2008, and Azure came out in 2010. It took 15 years for public cloud adoption to reach 50%,” he said. “True unlocking will happen in the next three to five years, but we should see meaningful progress. We need to think of this as a capital investment project. That’s when the real unlocking will happen. If we don’t, we’ll be stuck in the valley of incrementalism, pilot purgatory.”

Bruno Ghicardi, co-founder of information technology company CI&T, said that when building his own agent AI, he prefers structures that “gradually give autonomy to the agents in a system where there is a level of supervision that you can define when you actually withdraw supervision.”

Mr. Guicardi used the example of automated client responses. Initially, someone would review every response created by the AI ​​before the response was sent, but over time, once a response was deemed acceptable, that person would review the response less and the AI ​​could automatically submit the response.

“We think this is a way to build confidence,” he said. “The key is to build systems that gain control and trust to be autonomous.”