AI may have saved my life recently, and I do not say that lightly.

Earlier this year, I had what felt like a persistent calf cramp for about 5 days. It was tender, a little swollen, and getting worse. I initially thought it was a muscle issue. My chiropractor treated it that way as well when I went to see him.

After the pain kept getting worse over the next two days, I decided to pivot. I checked in with the AI health agent I had built for myself.

The AI Agent’s Diagnosis

Nearly a year earlier, I had uploaded my medical records, lab work, medications, visit notes, and genetic data into that system. I had not built it because I expected a crisis. I built it because I train Vistage members and their companies on how to use AI effectively and create practical AI agents, so I applied the same approach to my own health.

The instructions I gave the agent directed it to synthesize my records, generate ranked diagnostic possibilities, identify specific tests to request, and provide concrete next steps in urgent situations.

When I checked my symptoms through that agent, it raised deep vein thrombosis, or DVT, as a serious possibility. It also pointed me to the key diagnostic step: I needed an ultrasound.

That mattered because DVT involves a blood clot in a deep vein, usually in the leg. Symptoms can include calf pain, swelling, tenderness, warmth, and discoloration. The dangerous part comes when a clot breaks loose and travels to the lungs, causing a pulmonary embolism, which can become life-threatening.

I had not even considered DVT. I did not know enough about the pattern of symptoms. Later, when I shared the story with people close to me, I learned that my wife’s grandfather died from a pulmonary embolism. So did the mother of one of my wife’s close friends. That made the risk much more real to me.

I messaged my primary care office, described the constant calf pain and swelling, and raised the DVT concern. They advised me to either schedule an appointment or go to urgent care.

A Second Opinion

That sounded reasonable at first. The AI agent pushed me to think through the operational reality of that advice. My doctor’s office could evaluate me, but it could not perform the ultrasound. Urgent care could assess me, but most urgent care centers do not have the ultrasound capability needed to diagnose or rule out DVT. Duplex ultrasound, which uses sound waves to evaluate blood flow in the veins, is the standard diagnostic test for DVT.

Ultimately, I made a different decision. I went to the ER, despite the hassle of waiting for hours in the waiting room.

And I’m glad I did. The ultrasound found 4 blood clots in my left leg.

The doctors seriously considered admitting me to the hospital. They even inserted a pair of IV ports in preparation. After consulting with specialists, they fortunately sent me home on blood thinners.

I am stable now. I caught it in time. But this could have ended very differently if I had delayed or followed the wrong diagnostic path. If one of those clots had broken off and traveled to my lungs, I could have faced the same outcome as my wife’s grandfather or the mother of her friend.

Connect, Contextualize, and Clarify

The lesson for executives goes well beyond health.

The real value of AI does not come from asking it to “be smart.” The value comes from providing enough context, clear instructions, and a practical role in your decision-making process.

In this case, the agent did 3 things well.

• First, it connected scattered information. My symptoms, medical history, medications, and risk factors existed in different places. The agent could consider them together instead of treating my calf pain as an isolated complaint.

• Second, it expanded the diagnostic frame. I saw a cramp. My chiropractor saw a muscle issue. The agent raised a high-consequence possibility I had missed.

• Third, it clarified the next best step. It did not just say, “talk to a doctor.” It helped me identify the specific test that mattered and the setting where I could get it fast enough. And it helped me advocate for myself and follow the right course of action, despite the advice of my primary care doctor.

That same pattern applies inside small and midsize companies.

A well-built AI agent can help a leadership team connect dispersed information, surface overlooked risks, and move faster toward the right next step. It can review meeting notes, customer feedback, policies, financial reports, sales data, employee concerns, and operating procedures. It can help executives ask better questions before a crisis becomes visible.

But leaders need to avoid a common mistake. A generic chatbot interaction rarely delivers the same value as a well-designed agent.

A practical agent needs context, role clarity, and instructions. It needs access to the right information. It needs boundaries around what it should analyze, what it should recommend, and when it should escalate. It should help leaders reason more effectively, not replace human judgment.

I have seen this repeatedly in my consulting work. In one small company, the leadership team struggled with customer churn. The sales team blamed price, the service team blamed unrealistic expectations, and operations blamed poor handoffs.

We built a simple AI workflow that analyzed customer emails, CRM notes, onboarding documents, and cancellation reasons. The system did not magically solve churn. It did something more useful. It identified recurring failure points that leaders had missed because the evidence sat across multiple departments. Once the team saw the pattern, they changed the onboarding process, improved handoffs, and reduced avoidable churn.

That is the executive opportunity. AI can become a second brain for the organization, but only if leaders treat it as an operating system for better decisions rather than a novelty tool for writing faster emails.

For personal use, I strongly recommend that every executive set up a private AI assistant for health, finances, travel, insurance, estate planning, and other areas where fragmented information can lead to costly mistakes. For company use, leaders should start with a few high-value agents tied to recurring decisions, risk detection, and operational bottlenecks.

Start Small and Be Deliberate

You do not need to make this complicated. Start by choosing one domain where better information could materially improve outcomes. Gather the relevant documents. Write clear instructions. Tell the agent what role it should play, what context it should consider, what outputs you want, and when it should recommend escalation.

My health agent did not replace doctors. It helped me advocate for the right care faster. That distinction matters.

The best use of AI for executives follows the same principle. Do not use it to avoid expert judgment. Use it to reach better questions, better escalation points, and better decisions sooner.

In my case, that difference may have saved my life.