What My Glucose Taught Me (and How AI Helped Me Listen Better) Part of the “Applied Intelligence: My Life With AI” Series

Editor’s Note:

This is part of an ongoing personal series on how I’m using AI to optimize my life. While not directly about tennis, it reflects the same mindset I bring to athlete development and performance architecture: observe the data, find the pattern, and build a plan that fits the individual.
—Duey

The Setup: Two Meals, Two Walks, One Pattern

Over the past few months, I’ve been watching my blood sugar. Not casually, but with intention.

Most days, I eat twice:

• Midday Meal: A heavier plate between 11:00 AM and 1:00 PM

• Evening Meal: A large salad with protein (chicken, tuna, or shrimp) between 5:00 PM and 7:00 PM

Almost without exception, I walk within 30 minutes after eating. I don’t usually walk in the morning unless the Texas heat makes it the best option, but those post-meal walks have become a cornerstone of my plan.

Why? Because the data told me that when and how I move matters just as much as what I eat.

The Problem: Mornings

Despite eating nothing since the night before, my blood sugar often started creeping up in the early morning hours. Not a spike — just a slow, steady rise.

At first, I thought it was the coffee. Then I blamed stress. Eventually, I remembered something I’d learned years ago: Lantus doesn’t last 24 hours in everyone.

In my case, the coverage was solid… until around 3 or 4 AM. After that, my liver took over, my cortisol joined the party, and my Dexcom graph started climbing.

The Pivot: A Split-Dose Experiment

This morning, with a BG of 108 mg/dL, no food since last night, and only black coffee planned, I decided to test a new approach:

• ✅ Splitting my Lantus dose (18 units AM / 18 units PM)

• ✅ Holding off on fast-acting insulin unless BG rose above 130–140

• ✅ Tracking everything for the next 7 days

This wasn’t a doctor’s directive — it was an informed experiment based on my own patterns, rhythms, and data.

To keep it consistent, I built a custom tracker that logs:

• Lantus and Lispro timing

• Metformin doses

• Meals, walking times, and morning blood sugars

The Insight: Know Your Body, Then Build the Plan

None of this came from a template. My CGM didn’t just give me numbers — it gave me a story. And with the help of AI, I organized that story into something actionable.

I learned:

• Coffee without walking in the morning can nudge BG upward

• Afternoon Lispro is most effective when paired with movement

• Evening Lantus was wearing off before I needed it most

This wasn’t trial and error. It was trial and attention — amplified by AI’s ability to help me spot patterns and build structure.

How AI Fit In

This post — and the protocol it describes — evolved through conversation with AI. I shared my data, refined my goals, and used AI to:

• Create a personalized 7-day glucose and insulin tracker

• Develop a Dexcom note template for my morning routine

• Reflect on what I’d learned by helping me draft this article

The biggest lesson? AI is only as useful as the questions you ask. If you’re willing to stay curious and consistent, it becomes a true collaborator.

Final Thoughts

This isn’t a prescription. It’s a personal case study. But whether you’re managing blood sugar, a training plan, or any performance goal, the formula is the same:
Observe → Find the pattern → Build the plan → Refine.

And yes — AI can help you do it better.