Why 95% of enterprise AI fails—and why success is closer than you think
Every week, the headlines trumpet a new rollout of enterprise AI agents. Yet behind the scenes, most of these projects stall out before they ever deliver results, often failing in unexpected and costly ways. In fact, according to a recent report from MIT, as many as 95% of enterprise AI projects fail.
For Kevin Chung, Chief Strategy Officer at WRITER, the problem is rarely the technology itself—but the people, processes, and assumptions behind it. “Ninety-five percent of enterprise AI projects don’t fail because of the tech,” Kevin says. “They fail in adoption—the last mile is where transformation either happens or stalls.”
On the debut episode of MindMakers, Sendbird CEO John Kim sat down with Kevin Chung to unpack why AI success is so elusive—and why most enterprises are closer to ROI than they realize.
Fixing the last-mile problem in enterprise AI
Kevin has yet to see an AI project fail because the model wasn’t powerful enough. The breakdown happens when it’s time to embed AI into workflows, empower employees, and scale. But he points out that this “last-mile” failure often starts much earlier with misconceptions about what it takes to succeed.
“A lot of people mistakenly think you just adopt AI, buy the software, and it will work,” he says. “But AI doesn’t work off the shelf for enterprises. It’s pretty tricky.”
The challenge isn’t just technical, it’s organizational. Teams are expected to learn a new industry, build solutions, and keep pace with a rapidly evolving technology, all at once. Meanwhile, it’s frustrating to spend hours configuring components and crafting the right prompts, only to see new use cases pop up every month,
Kevin likens the experience to drinking from a firehose: overwhelming and unmanageable. Faced with a flood of tasks, decisions, and the attendant uncertainty—“Is it time to pivot? Was any of this worth it?”—even motivated enterprises risk losing momentum and watching projects go under before adoption. He encourages organizations to focus on change management rather than just in-house solutions, as success ultimately hinges on the people who believe in the mission and advance it.
“Number one, you gotta get your people empowered. Number two, you actually have to start promoting the people. And promoting in a sense of [not just] giving them the tools, but actually finding those champions who can be your AI advocates, right? The AI leaders in the organization who are gonna effectively drive forward those initiatives.”
— Kevin Chung
How to choose an AI agent platform that works
Treat AI like a new hire, not software
One of Kevin’s biggest lessons: companies that thrive with AI treat it like a new teammate, not just another tool. “You onboard it, you train it, you assign it a champion, just like you would with a new hire,” he says.
This shift in mindset is critical. Tools can be ignored, but teammates get integrated, developed, and supported. By treating AI like a member of the workforce and allowing trust to grow over time, enterprises are far more likely to see adoption stick.
The friction between AI and humans is real. A recent study finds that 41% of Gen-Z and Millennial employees report sabotaging their company's AI strategy. Kevin understands the pushback. If AI doesn’t directly work for what employees want, they’re naturally going to resist, stick to old tools, and divert from adoption in favor of their own strategy.
41% of Gen-Z and Millennial employees report sabotaging their company's AI strategy.
— Inc.com study
This is why building trust is vital. “You can’t just throw AI over the wall and expect people to use it,” Kevin says. “Trust is built when you’re hands-on with your teams, showing them how AI supports their goals and then scaling from there.”
2 major pitfalls to dodge when converting to AI customer service
From pilots to productivity
Another reason AI initiatives fail is that they never get beyond the pilot phase. In fact, a recent RAND report shows that 80% of AI pilots never make it to production—a failure rate nearly double that of traditional IT projects.
Kevin acknowledges that enterprises juggle a myriad of moving parts and considerations as they try to prove concepts. But at a certain point, he says, it’s time to stop tinkering and start scaling.
“We’re seeing the [AI] leaders move beyond experiments into hundreds of deployed AI agents,” Kevin says. “That’s where the ROI shows up—tens of thousands of hours reclaimed and teams freed up for higher-value work.”
One of those leaders is Ryder, a global logistics and transportation company. Despite early hesitancy, Ryder pushed forward into scaled deployments. Today, they’ve launched hundreds of custom vertical AI agents, proving what’s possible when adoption sticks.
AI agents built for specific use cases are more likely to create value compared to generic AI
For Kevin, Ryder is proof that enterprise AI can work when it’s done right. Pilots can validate potential, but until organizations appoint AI champions and scale deployment, they’ll never realize the full benefits.
5 key questions to vet an AI agent platform
Building trusted AI through partnerships
For most enterprises, the barrier to AI success isn’t whether the technology works—it’s whether they can trust it while navigating toward transformation. Kevin stresses the importance of partnerships with technical AI experts, which are becoming more abundant and effective every day. “You really do need a good guide to get you there,” he says.
This is because AI isn’t plug-and-play for enterprises. Every deployment has to be customized to the organization’s workflows, and even within the same industry, no two companies are alike because they have different processes.
That’s why WRITER takes a consultative approach, working closely with enterprises. “We understand their business at the deepest level. We do the use case mapping. We’re like a part of their team in many cases. That’s the way we’ve been able to be successful,” Kevin says.
“We understand their business at the deepest level. We do the use case mapping. We’re like a part of their team in many cases. That’s the way we’ve been able to be successful.”
— Kevin Chung
By handling the heavy lifting, WRITER frees internal teams to focus on what they know best: change management, adoption, and empowerment. This way, enterprises can chart a course for change without losing momentum.
Build lasting customer trust with reliable AI agents
A proven playbook for trusted enterprise AI
At a time when most AI projects are failing, Kevin Chung and WRITER have established a playbook that delivers reliable results. The difference isn’t cutting-edge models or massive datasets—it’s well-supported people and processes. When motivated employees have both the mission and the backing to push forward, AI projects gain the traction they need to succeed.
Kevin’s playbook comes from recognizing a familiar pattern in the AI space, one he first observed during his career in enterprise technology, including mobile payments at Google and cloud storage at Dropbox. The pattern? Savvy enterprises didn’t try to build the groundbreaking technology themselves. Instead, they waited until it was enterprise-ready and then focused on integrating it effectively. He realized the same principle applied to AI.
WRITER’s own evolution mirrors this lesson. “We started as a niche translation tool,” Kevin recalls. “But enterprises needed more—they needed AI that could work across functions. That’s how WRITER grew into a full-stack AI agent platform.” By stepping in to handle the technical complexities of full-stack AI implementation, WRITER enables organizations to play to their strengths and position themselves for lasting success.
In the end, rather than trying to reinvent the wheel, Kevin encourages organizations to focus foremost on change management and leave the rest to expert partners. Upholding this conventional enterprise dynamic in the evolving AI space has enabled WRITER to turn numerous AI projects into operational solutions that deliver real results. Kevin knows others can do the same.
Catch the full conversation with Kevin Chung on MindMakers to hear more lessons from WRITER’s AI playbook—and discover how your company can be part of the 5% that succeed.
Want to learn about Sendbird’s enterprise-ready AI agents? You can contact sales.
