Enterprise AI has moved beyond experimentation and the challenge now facing organizations is whether it can deliver value consistently across an entire business. While pilots often produce promising results, scaling AI exposes weaknesses that technology alone cannot solve. As Aicha Gersing, Senior Vice President of Customer Success for Agentforce and Data 360, puts it, “AI agents are amplifiers and not fixers. They scale what already works, but they can also scale what doesn’t work.”
That distinction sits at the heart of enterprise AI adoption. Success depends less on deploying sophisticated models than on strengthening the operational foundations beneath them. Data quality, disciplined workflows, organizational trust, and thoughtful role design determine whether AI accelerates performance or simply creates faster chaos.
Why Enterprise AI Stalls After the Pilot Phase
Launching an AI pilot is relatively straightforward because companies can tightly control the data, workflows, teams, and success metrics involved. Scaling is fundamentally different. As adoption spreads across departments, organizations lose that level of control and long-standing operational problems become impossible to ignore.
The real blockers most of the time are not technical,” says Gersing. “It’s data quality, workflow discipline, and organizational trust.” Many organizations discover that the systems feeding AI have never been as reliable as they assumed. Customer records may be incomplete, processes may vary across teams, and employees may not consistently follow established workflows. Rather than correcting those issues, AI magnifies them. According to Gersing, some of the biggest failures occur when organizations automate broken processes instead of fixing them first. “Most scaling failures are companies that automated a broken workflow and got faster chaos instead of faster results.”
Start Where the Business Already Has Structure
While AI has broad potential, not every business function offers the same path to success. Gersing believes customer support provides one of the strongest starting points, because it combines structured data with repeatable processes that AI can execute reliably. “We were able to deploy AI agents that resolved the majority of customer interactions within minutes, accelerating processes that previously took hours or days.” The impact goes well beyond productivity gains. Rather than simply assisting employees, AI can reduce large volumes of repetitive work altogether.
Support functions such as ticket routing, first-response generation, refund workflows, and other highly structured tasks represent what Gersing estimates to be, “about 60 to 70% of the enterprise agentic opportunities today.” These lower-risk applications allow organizations to build confidence while creating measurable business value. Once trust has been established, companies can gradually expand into more complex customer interactions and decision-making processes.
Redesigning Work Around Humans and AI
The greatest transformation may not come from the technology itself, but from how organizations rethink work. As AI supports routine knowledge retrieval and standardized decision-making, traditional specialist roles begin to evolve. Gersing describes the emerging employee as “a specialist generalist,” someone who combines account management, technical troubleshooting, implementation guidance, and customer success capabilities, while AI provides expertise exactly when needed.
The result is fewer handoffs, stronger customer relationships, and more time focused on high-value interactions rather than administrative tasks. “What changes are the narrow specialist roles that are built around holding a single piece of knowledge,” she explains. “As AI retrieves and synthesizes information reliably, these roles evolve into general capabilities.”
This evolution also requires thoughtful leadership. Employees often worry that helping build AI systems will ultimately replace them. Gersing believes that concern should be addressed directly, because the people expressing the greatest hesitation are often the organization’s deepest experts. “The exact people that you need to help you build these systems are frequently the ones who feel most threatened,” she says.
Instead of positioning AI as a replacement, leaders should demonstrate how it supercharges employees to spend more time creating value while removing repetitive work. “This makes you dramatically better at the work where you’re already creating the most value.” For Gersing, successful AI adoption is ultimately a human operating model. The opportunity lies in redesigning roles, incentives, and career paths around a new division of labor between people and intelligent systems.
Trust Must Be Engineered, Not Assumed
Allowing AI agents to make decisions requires confidence that their outputs are both accurate and reliable; organizations must establish clear performance thresholds before expanding autonomous AI use. Systems performing below acceptable accuracy levels simply create additional work through exception handling instead of reducing it.
“It is the ‘trust but verify’ approach,” she says. While many organizations focus heavily on context engineering to make models relevant to customers, “correctness engineering (ensuring the output is trustworthy) is what really helps improve your AI agents.” Building confidence internally is just as important as earning customer trust. Consistent verification creates the foundation for broader adoption across the enterprise.
AI’s Greatest Impact Will Be Human
Five years from now, many of today’s headline-grabbing AI capabilities will simply be standard business practice. Customer support agents will manage increasingly sophisticated business-to-business interactions, allowing employees to focus on problems that demand creativity, judgment, and innovation. “What you will see in the next five years is a transformation of entire organizations and really rethinking what is the role of a human alongside agents.”
The long-term opportunity extends far beyond large enterprises. She sees AI becoming a practical tool for small businesses, neighborhood shops, and local service providers, helping people spend less time on repetitive work and more time creating value. Ultimately, enterprise AI adoption is not about deploying more technology. It is about strengthening the business that technology serves, earning trust through measurable outcomes, and designing organizations where humans and AI each contribute what they do best.
Follow Aicha Gersing on LinkedIn for more insights on AI adoption, customer success, and digital transformation.
