From AI Hype to Agentic Impact
- Vivienne Wei
- Dec 27, 2025
- 3 min read
Why Leadership Will Decide Who Wins the Next Decade
In 2025, artificial intelligence is no longer a moonshot. It’s table stakes. Every major company has an AI strategy, a handful of pilot projects, and an ambitious executive sponsor. Yet few have moved from promise to performance. Most are still treating AI as an experiment instead of an engine.
A recent MIT Technology Review report found that while 95% of organizations have adopted AI, only 5% have scaled it across the enterprise. We’ve reached the stage where success is about building the kind of organization that can harness it.
Agents Are Products, Not Projects
Many leaders still approach AI as a project: something to launch, measure, and close out. But AI agents don’t operate like projects. They evolve.
At Salesforce, we saw this firsthand. Early in our AI journey, a 1% escalation rate from AI to human agents looked like a triumph. “We were literally high-fiving each other,” recalls Bernard Slowey. But when we reviewed the conversations, we discovered many customers were stuck in frustrating loops. The agent was trying too hard to solve everything on its own.
The fix was counterintuitive and effective: we increased human handoffs to about 5%. That small change driven by CX expertise and human judgement improved satisfaction scores, proving that hybrid beats either alone.
The insight ran deeper. Accuracy alone isn’t enough when AI acts on behalf of people. Empathy, even when simulated, drives trust and better outcomes. As my colleague Flavio Lopez wrote,
“Agents aren’t deliverables. They’re meant to learn, adapt, and grow with the business, just as products do.”
In fact, this is why at Salesforce we have created the FDE (Forward Deployment Engineer) role. The FDEs are meant to work in the "Agent as a Product" organization within our customers and not just simply assigned temporarily in a standard Project fashion.
That’s the shift leaders must make: Stop sponsoring projects. Start building products.
The Role Every CEO Should Be Creating
Every enterprise needs an Agent Manager, a role that bridges human leadership and digital intelligence. This is a product and strategy role: someone who owns the roadmap, defines success metrics, and determines when a digital teammate is ready to act autonomously.
When an agent says, “I’m not able to do that yet,” that isn’t failure. It's feedback. It means your people trust the system enough to explore its limits. That’s the moment when adoption becomes culture.
The best Agent Managers operate like miniature CEOs: balancing ambition with governance, learning with accountability, and scale with ethics.
Start Small, Win Big
Every CEO I meet asks the same question: “What’s the first AI use case that actually pays for itself?” The answer is simple: start where repetition meets frustration. Look for a workflow that consumes time but adds little strategic value.
That’s exactly where we began at Salesforce. In customer service, our digital labor learned to deflect routine inquiries and free up human agents for higher-value work. The results were clear:
5% fewer support cases
80% faster response times
84% of inquiries resolved by AI Agents
The key wasn’t automation for its own sake. It was empathy. The agents improved because they learned to say, “I’m sorry.” AI is amplifying people and taking care of repeatable tasks.
The Economics of Agentic AI
Agentic AI isn’t just a technological leap. It's a structural shift in how enterprises grow and allocate capital.
Goldman Sachs projects that by 2030, AI agents will represent more than 60% of the enterprise software market, expanding the total opportunity by as much as 45%.
Boston Consulting Group (BCG)’s analysis echoes this, showing that leading AI adopters outperform peers with 50% higher revenue growth and 60% greater shareholder returns, largely because they focus their transformation on people, not platforms.
This is what I call the Compounding Agentic Enterprise: Humans create. Agents execute. Agentic AI system keeps learning. It’s an economic flywheel that runs continuously and scales trust.
The Takeaway
The job of the modern CEO is to design a collaboration model where humans and digital teammates can both reach their highest potential. If you’re leading a company today, resist the temptation to chase every new AI headline. Your competitive edge come from deploying agents where they move the needle most.
De-risk your operations.
Assign an Agent Manager.
Start with one workflow that matters.
The future won’t belong to companies that build AI projects. It will belong to those that productize intelligence and lead it responsibly.
Comments