Before you automate anything, you need to know what actually matters. Here is why visibility comes before AI deployment, and what happens when you skip that step.
Not every ball bounces
Every few years, a new technology wave arrives promising to solve everything. Right now, it is Agentic AI. Every headline, keynote, and vendor seems to be telling the same story: AI agents will transform operations, automate decisions, improve customer experiences, and unlock massive productivity gains. Maybe they will. But after spending years helping organizations navigate technology shifts, I have noticed a pattern: the companies that see the biggest returns are not the ones that adopt the newest technology first. They are the ones that know exactly what problem they are solving before they deploy it.
I'm often reminded of what former Coca-Cola CEO Brian Dyson called the “Glass Ball Theory”: not everything on your plate carries the same weight. Some things are rubber balls. Drop them, and they bounce. Others are glass balls. Drop them, and they shatter. The challenge is not identifying that both exist. The challenge is knowing which is which.
Every day, every team and every organization needs to identify the handful of things that absolutely cannot break and focus relentlessly on those first. When you do not, noise wins. And nowhere is that more apparent today than in the race to adopt AI.
Inspired by a concept often associated with former Coca-Cola CEO Brian Dyson, the Glass Ball Theory is a decision-making framework for identifying what truly cannot fail in an organization. Rubber balls can be dropped and recovered. Glass balls, once dropped, shatter. In the context of AI strategy, the theory argues that healthcare organizations should identify their highest-impact operational gaps before deploying automation. Without that diagnostic foundation, organizations risk automating the wrong processes, investing in the wrong tools, and missing the revenue and patient access opportunities that matter most. The most successful AI strategies start not with technology selection but with operational visibility.
AI has a clarity problem
The current AI landscape is full of ambitious promises. Every platform claims to automate workflows. Every company claims to have intelligent agents. Every demo looks impressive. But there is an uncomfortable question many organizations are skipping:
What exactly should AI be fixing?
I have had countless conversations with healthcare leaders over the past year who know they need an AI strategy. They know patient expectations are changing. They know operational pressures continue to grow. They know automation can help. What they do not always know is where their biggest opportunities actually exist.
That is not a technology problem. It is a visibility problem. You cannot automate what you cannot see. You cannot optimize what you have not measured. And you should not deploy AI across processes that are not fully understood. The most important step in any Agentic AI journey is not automation. It is clarity.
Start with the diagnosis, not the prescription
One thing I have learned watching the AI market evolve is that everyone suddenly has an answer. The advice comes quickly and confidently: automate scheduling, replace the call center, speed up collections. The problem is that healthcare is not a playground for experimentation.
My first hard stop when evaluating any AI provider is simple: can they meet the compliance, security, governance, and reliability standards healthcare demands? If the answer is no, the conversation ends there. The right infrastructure is HIPAA-native and carrier-grade, built for reliability, call quality, and data integrity. Certification across HITRUST, SOC 2 Type II, PCI DSS, and ISO/IEC 42001 matters. So does alignment with the NIST AI Risk Management Framework. Healthcare organizations need more than innovation. They need trust.
But compliance alone is not enough. My second hard stop is whether an AI provider can actually help identify where the problem lives before recommending a solution. Too many vendors start with the answer. Very few start with the diagnosis. That distinction matters.
You may think you have a scheduling problem when the real issue is patient acquisition. You may assume your call center is underperforming when the actual revenue leakage is occurring during collections. You may believe appointment bookings are the biggest opportunity while patient retention is quietly costing you far more. If you do not know where your glass balls are, you will likely automate the wrong thing.
Steve Jobs understood this long before AI existed
One of my favorite business examples comes from Steve Jobs’ return to Apple in 1997. At the time, Apple had dozens of products, initiatives, and priorities competing for attention. Jobs famously simplified everything down to a handful of core products. He was not rejecting innovation. He was creating focus. He understood something many leaders still struggle with today:
Scale follows clarity.
When leaders can clearly see what matters most, resources follow. Innovation accelerates. Results compound. The same principle applies to AI strategy in healthcare. Organizations that rush to automate everything often end up automating complexity. Organizations that first identify their most important operational gaps create far better outcomes and do so with significantly less wasted investment. In healthcare, where every dollar matters and operational disruption has clinical consequences, that discipline is not optional. Focus first. Scale second.
Healthcare organizations are drowning in signals they cannot see
In healthcare, the challenge is not a lack of data. It is the exact opposite. Calls, appointments, patient interactions, scheduling outcomes, revenue cycle performance, collections, patient retention: the volume of information is overwhelming. Yet many organizations still rely on manual reviews, small call samples, isolated reports, and anecdotal feedback to understand performance. That creates blind spots.
A missed appointment does not seem significant on its own. An abandoned call does not look urgent in isolation. A scheduling breakdown at a single location may appear contained. A collections issue may never surface until months later. But when these issues happen hundreds or thousands of times across multiple sites, they become significant revenue leakage and patient access challenges.
The problem is that most leaders never see the full picture. They are forced to react to whatever problem is currently the loudest.
See the full picture before you automate anything
The most valuable thing any AI partner can do before recommending automation is help an organization see what it is currently missing. That starts with analyzing 100% of patient interactions across every location, not a small sample reviewed days after the fact. It means surfacing the patterns that manual QA programs never reach: the missed revenue, the access barriers, the conversion gaps, the retention challenges hiding in call volume that no one has time to review.
The goal is not a report. It is an answer to the question that should precede every AI investment:
Where are your glass balls?
When leaders have that answer, something changes. Resources stop chasing the loudest problem and start flowing toward the highest-impact opportunity. Automation investments produce measurable returns instead of expensive lessons. The organization moves from reacting to leading. That is the value of starting with visibility: not a report, but the confidence to know you are solving the right problem before you invest in solving it.
Healthcare AI strategies fail most often not because the technology is wrong but because the problem is wrong. Organizations that analyze only 1 to 3% of their patient interactions through traditional QA are making automation decisions based on an incomplete picture. The remaining 97% of interactions contain the signals that reveal where revenue is actually leaking, where patients are actually disengaging, and where the highest-impact automation opportunities actually exist. Seeing 100% of interactions before deploying AI agents is not a luxury. It is the diagnostic step that determines whether the investment delivers measurable outcomes or expensive complexity.
The most successful AI strategies begin with visibility
I believe Agentic AI will create enormous value for healthcare organizations. But the winners will not be the organizations that automate first. They will be the organizations that understand their operations first.
Identify the glass balls. Protect them. Prioritize them. Then scale.
In a market full of AI vendors eager to tell you what to automate, the most valuable conversation starts somewhere different. Before deploying Agentic AI, before redesigning workflows, before automating scheduling, collections, or patient engagement, there is one question every healthcare leader should answer:
What problem am I really trying to solve?
That question, answered honestly and with full visibility into your operations, is where every successful AI strategy begins.
Frequently Asked Questions
What is the Glass Ball Theory and how does it apply to AI strategy?
Why do most healthcare AI deployments underperform expectations?
How do you identify your glass balls before deploying AI?
