AI Adoption Starts with People

What OpenAI's Atlas Announcement Really Teaches Business Leaders About AI Adoption

Part 1 of a three-part executive series

Artificial intelligence is changing rapidly, but successful adoption still depends on the same principles that have guided every major technology shift over the past three decades. In this series, we explore how organizations can move beyond AI experimentation and build lasting business value.

---

Earlier this month, OpenAI announced that it would discontinue Atlas, its standalone AI-powered desktop browser, and instead incorporate many of its capabilities directly into the ChatGPT desktop application. Unsurprisingly, much of the discussion surrounding the announcement centered on Atlas itself.

Had OpenAI changed direction? Did the product fail to gain traction? Was the AI browser concept simply ahead of its time?

Those are interesting questions, but we believe they're the wrong ones.

The Atlas announcement isn't really about a browser. When the tool first arrived, we explored its potential in ChatGPT Atlas and the Future of Intelligent Browsing. Looking at its consolidation today reveals a much broader lesson about how organizations adopt new technology.

Even one of the companies leading today's AI movement appears to have reached a familiar conclusion: people don't necessarily want another application to learn. More often than not, they want the applications they already rely on to become more capable.

That distinction is more important than it might first appear.

At Synergy, we've spent nearly thirty years helping organizations navigate waves of technology change. During that time, we've watched businesses adopt company intranets, SharePoint, Microsoft 365, workflow automation, customer portals, mobile applications, cloud platforms, and now artificial intelligence. Every one of these technologies promised meaningful improvements, and many have delivered exactly that.

Yet we've also seen the same challenge repeat itself.

Projects rarely struggle because the technology isn't capable. More often, they struggle because organizations underestimate how difficult it is to change the way people work. Technology evolves quickly. Human behavior evolves much more slowly.

That's what made the Atlas announcement so interesting.

Rather than continuing to invest in a separate browser experience, OpenAI chose to bring those capabilities into an application millions of people already use every day. Whether Atlas itself succeeded or failed is almost beside the point. The bigger takeaway is that integration often creates more value than another destination.

For organizations beginning their own AI journey, that's a lesson worth paying attention to.

Technology Changes. Human Behavior Doesn't.

Every major technology wave has promised to transform the way organizations work—and many of them have.

The internet changed communication. SharePoint redefined document collaboration. Cloud computing reshaped infrastructure. Microsoft Teams transformed workplace communication. Workflow automation streamlined repetitive business processes.

Today, artificial intelligence represents the next major shift.

Each of these technologies solved real business problems. Yet many organizations struggled to realize their full value—not because the software wasn't capable, but because adoption proved far more difficult than deployment. It's a challenge that helps explain why 95% of generative AI projects fail, and why successful AI initiatives require much more than selecting the right platform.

Deploying technology is largely a technical exercise. It involves selecting software, configuring infrastructure, integrating systems, and training users.

Adoption is different.

Adoption asks people to trust new processes, develop new habits, and sometimes rethink the way they've worked for years. Better software alone rarely accomplishes that.

We've seen this pattern play out across organizations of every size and industry:

• Beautiful intranets that employees rarely visited because asking a colleague remained faster.

• SharePoint deployments that became little more than document repositories because governance and information architecture were never fully developed.

• Customer portals with excellent self-service capabilities that failed to reduce support calls because customers still preferred speaking with someone directly.

• Workflow automation initiatives that digitized inefficient manual processes without first questioning whether those processes should exist in their current form.

• AI pilots that generated excitement during demonstrations but quietly faded because they required employees to interrupt their existing workflows.

The common thread isn't the technology.

It's the assumption that people will naturally change their behavior simply because a better tool exists.

History suggests that's a risky assumption. Successful digital transformation has always been about aligning technology with the way people already work—not expecting people to redesign their work around new technology.

Artificial intelligence will be no different.

AI Doesn't Need More Destinations. It Needs Better Workflows.

In many ways, Atlas reinforces a lesson that organizations have been learning for years.

People don't necessarily want another destination. They want the places they already work to become more capable.

Think about the average employee's day. Most people already move between Outlook, Microsoft Teams, Microsoft 365 applications, line-of-business systems, ERP platforms, CRM systems, browsers, and a handful of specialized applications unique to their role.

Every additional application competes for attention. Every additional login introduces friction. Every additional interface requires learning. Every additional destination creates another place where information can become disconnected.

Individually, those inconveniences may seem small. Collectively, they create unnecessary complexity.

That's why we encourage clients to begin AI conversations by focusing on business processes rather than AI platforms. Instead of asking,

"Where can we deploy AI?"

we encourage leadership teams to ask a different question:

"Where do our employees lose time every day? How can AI remove that friction without forcing them to completely change the way they already work?"

That shift changes the conversation. AI stops being another technology initiative and becomes a business improvement initiative.

And that's where organizations begin realizing meaningful value.

The Atlas announcement may have been about a browser on the surface, but its broader message is about adoption. The most successful AI initiatives won't ask employees to spend more time moving between applications. They'll make the applications employees already rely on more intelligent.

In Part Two, we'll build on that idea by introducing the Four Layers of Successful AI Adoption: a framework for evaluating AI readiness before technology decisions are ever made.