The human advantage in collaboration: reading situations, not just data sets

The question that any enterprise comms professional deploying an agentic AI workflow should be asking is simpler— are the existing AI tools doing the job of serving everyone effectively?

Sometimes (ok, more often than that) I feel like I’m either a broken record or an ostrich with rigor mortis—stuck with my head in the sand about limitations of AI—not its capabilities, but its real-life limitations. Right at the top of the list is the fact that many people—including many in Gen Z—simply hate it. They use it, but they hate it. But I digress. Many years ago, a friend and client in the off-site airport parking business commented that the business he was in had nothing really to do with parking or real estate. I was surprised but wanted to hear what business he thought he was in. He opined: his business was customer service. His isn't the only one: When you think about it, whether an operation is a nail salon or hardware store or dental practice, it’s really about customer service. 

Is "my business is a customer service business" still true in the age of technology dominance? Customer service has obviously lost some of its luster simply as a result of perceived AI-driven efficiencies. But I also hope that the AI-agentic contact centers in the world don’t replace the human experience, particularly when a consumer has a problem that doesn’t fit the mold. A recent podcast/article in The Verge tackles this question in a different way, questioning the differences between what the author calls “software brain” and “legal brain.”The question that any enterprise comms professional deploying an agentic AI workflow should be asking is simpler— are the existing AI tools doing the job of serving everyone effectively?

In author Nilay Patel’s piece, he contends software brain occurs when “you see the whole world as a series of databases that can be controlled with the structured language of software code.” The creation of large databases that are subject-driven (such as the database that companies like Uber have created to analyze and support its customers) is a perfect example of software brain — but only so long as the fields in the databases accurately reflect actual customers, and as long as those customers get the service that they require. That is, so long as the request and the AI-generated responses fit in the right boxes.

How different is the concept of “legal brain?” Not terribly different. According to Patel’s theory, lawyer brain involves “thinking in the structured language of statutes and citations” that can make things happen in the real world. Many of the biggest names in reasonably effective AI-driven world would be very happy and successful if reality could truly be flattened to fit into nice little boxes like giant 3D Excel spreadsheets. But as good lawyers know — at least I hope they do — very little in life fits into nice little boxes. The answers to challenging questions are rarely black or white. It’s all about the gray, otherwise known as ambiguity.

When I raised this subject with Christine Pedigo Bartholomew, Professor of Law at the University of Buffalo Law School, she made two very astute comments. First, she said, "What I try to teach my students is that ambiguity is the law. Always has been. Good lawyers do what good customer service does: they read the situation, not just the rule, then find the answer that actually works for the person in front of them. AI cannot do that—at least not yet." She continued commenting on the increasing use of AI within the practice of law: "My students will practice law with AI. That's fine: maybe even good fiscally for the profession. What I worry about is whether future lawyers will know when to put [the AI tools] down. The client sitting across from them won't fit neatly into any system. Learning to handle that moment: that's what true lawyering is that AI cannot provide."

What makes us human is our ability to reason with conscience. AI tools simply don’t have that capability. Since LLMs do not have context, every decision made relying on those tools can be seen as “history-free”— or at least free of context beyond the LLM’s experience. Context is truly what separates our brains from LLMs. We can consider a list of facts—particularly those that are not routine—and use the AI tools, combined with consciousness and comprehension of the non-conforming program to solve that problem. That is, the numbers, out of context, are limited by the model’s experience. If the challenge doesn’t fit the pattern that the LLM recognizes, it may not know how to resolve it. AI tools may get closer as their sophistication evolves upward, but what makes us human is our ability to read a situation as more than a set of yes/no questions. Agentic-driven contact centers can meet the needs of most customers most—or at least much—of the time. Lawyers who think only in terms of statutes and contract language without considering facts, circumstances and the climate of the issue at hand will find themselves at a marked disadvantage. And that’s way before they use AI tools like Claude or Chat GPT to do their work for them. 

According to well-known AI consultant and prognosticator Blair Pleasant, President and Principal Analyst of CommFusion, “In addition to reading the situation and reasoning with conscience, AI can’t apply common sense or empathy. There may be situations and special circumstances that call for exemptions and different ways of handling a situation, which require the human touch. The best AI can do is pass an interaction to a human, who can reason with conscience to do the right thing in that specific situation. The machines may be able perform tasks faster and more efficiently than humans, but it does so without the actual humanity that is often needed or desired.”

While the Verge framing was “software brain” versus “legal brain,” I think that’s the wrong question. While a legal approach to a problem may be different from a software approach, in fact, the analysis may be very much the same. To me, the questions that entities who rely on agentic AI should be asking are simpler—namely, are the existing AI tools doing the job of serving customers collectively? Are those customers who reach the contact center returning to spend more money with us, or is the AI interaction more annoying than it’s worth? To me, those are the questions.

Origially published iin No Jiitter on May 18, 2026