Ask a salesperson how they became skilled at their job and you might hear: “I just learned what works.” They didn’t study a textbook or memorize a script. They learned from years of watching, trying, and adjusting. That’s implicit intelligence in action—the unspoken knowledge we accumulate through experience.
That’s not how AI learns.
Until now, AI systems have been built to rely on explicit intelligence. Think KPIs, forecasting data, and codified rules—measurable information we can feed into AI models to draw rational, logical outputs. But humans aren’t always logical and we don’t work solely from explicit knowledge. We learn by doing, sensing, and pattern matching in ways we often can’t explain.
This matters now because organizations are on the cusp of a new phase in AI adoption. Agentic systems are AI-driven workflows that automate tasks while acting autonomously on our behalf. To do so effectively, AI must think and behave like us, which means learning like us too.
LOOK UNDER THE HOOD
Implicit intelligence shows up every day—at home and at work, whether we notice it or not. It’s how a procurement lead instinctively knows when a vendor’s offer is soft. Or how a customer service rep picks up on frustration in a client’s tone even before the conversation begins. These insights aren’t pulled from a dashboard. They’re felt by humans.
There are countless other examples—all of which require us to get under the hood of what it means to be a successful human worker. Knowing how to behave in a meeting without being told. Sensing when a team needs reassurance versus a push. Even knowing when to speak and when to listen. These behaviors reflect a type of intelligence that most organizations aren’t capturing and that few AI models are trained to recognize.
IF THIS, THEN THAT
The challenge for leaders is getting this knowledge out of their workers’ heads and into their AI systems. That requires shifting from data collection to knowledge capture. That means looking beyond numbers to understand how decisions are made and why. It means documenting the “if this, then that” instincts that live in people’s innate understanding of their job and role.
Take pricing strategy. An AI tool might know that if a key commodity drops 15%, a related product should be adjusted by 12%. That knowledge is based on data points built up over many analytical exercises. But if a human analyst knows something intuitively about the organization, market, or customer, their implicit understanding will be factored into the decision too. Perhaps 12% becomes 10% or 15%.
Another avenue for leaders to explore is behavioral observation. AI can begin to learn from watching the way humans act in context. By analyzing how different people respond during interactions with certain customers, suppliers, or colleagues, AI can tailor responses to the idiosyncrasies of that relationship and style.
CULTURE COUNTS
For every employee who embraces AI as a time-saving superpower, there’s another who sees it as surveillance—just as one person will happily shout to a voice-activated assistant when she needs to search for a fact, while another refuses to let her phone listen at all. People are different. Organizational cultures are different too.
That’s why capturing implicit intelligence requires more than tech. It demands trust, transparency, and a clear ethical compass. Leaders must consider how their teams feel about the idea of AI observing and mimicking their work, and where they draw the line between assistance and intrusion.
HOW TO CAPTURE IMPLICIT INTELLIGENCE
Either way, foundational AI capabilities are now table stakes. What will set organizations apart in 2026 and beyond is the proprietary data they own and the knowledge layers that data creates. This includes the unique contexts of their business—the information that exists not only in their dashboards but also in their people and their behaviors.
Here are three steps to get started.
1. Codify patterns
Focus on the workflows you most want to automate. What decisions underpin them? Who holds that knowledge and how do they apply it? Use interviews, journey mapping, or observational research to surface patterns. Then build the scaffolding to codify these patterns using ontologies, knowledge graphs, and domain language models.
2. Create ethical boundaries
Knowing where AI doesn’t belong is just as important as knowing where it does. For instance, I don’t ever want a robot telling me I’m seriously ill, how my child is doing in school, or why I’m not getting a loan. Some decisions will always require human empathy. Ethical boundaries matter.
3. Prepare your teams
Understand your workforce’s comfort levels and let that dictate the pace of change. Build opt-in paths. Make transparency a core part of your AI deployment. And keep revisiting as you go.
AI’S NEW FRONTIER
As we move into the agentic era and AI shifts from a tool to a teammate, implicit intelligence will be the key to building AI systems that reflect not just what we know, but how we know it.
That doesn’t mean letting AI negotiate with AI or manipulate itself unchecked. Without human oversight and our ability to compromise, agentic systems can enter feedback loops that delay or derail decisions rather than resolve or progress them.
Instead, success comes from judgments that merge explicit, high-quality system data with the implicit human knowledge hiding in plain sight. That’s AI’s new frontier. And it’s how organizations will gain an edge long into the future.
Traci Gusher is EY Americas AI and data leader.