SwimGPT

On tech and swimming

There is a solid argument for using no technology while swimming.
At one end of that spectrum, we are skinny-dipping.

There is also a solid argument for using lots of technology while swimming.
At the other end of that spectrum, we are Michael Phelps, using every available resource to optimize training and pursue Olympic gold.

Most of us swim somewhere in between. We can dial the technology up or down depending on what we want from the water that day.

I enjoy the full spectrum.

My evolution in swimming technology began with a simple Timex Ironman watch.

The first time I tried a practice swim to see whether I should enter a triathlon, I did not own a smartwatch. I clicked the stopwatch on my Timex and pushed off.

Five minutes later, I was gassed.

I didn’t need the watch to tell me that, but it did tell me exactly how long it had taken. I then took my pulse and used the watch to calculate my heart rate.

And so it began.

A few years later, when I started seriously learning swimming technique, Terry Laughlin’s Total Immersion introduced me to the Finis Tempo Trainer.

Now I had an audible pacer. I could place a small beeper under my swim cap, set it to a particular stroke rate, and measure my efficiency by combining three pieces of information: stroke rate, strokes per length, and time.

A new world opened.

I could swim stroke-rate pyramids. I could experiment with a slightly faster rhythm and notice when my stroke began to fall apart. I was receiving feedback during every workout.

Then, for my birthday, my wife got me waterproof headphones and a waterproof case for my iPod.

Swimming with music.

Amazing.

Apple and Garmin eventually introduced smartwatches that could record mountains of data from every lap: pace, distance, heart rate, stroke count, stroke rate, efficiency, and more.

Then came Form goggles, which put that information directly in front of my eye while I was swimming.

The first time I used them, the goggles gave me immediate feedback on my head position. I made a small correction and swam faster that same day.

In open water, the goggles can help me sight toward a distant target and swim a straighter line across dark water, where there is no black line beneath me and no lane rope beside me.

Each technology gave me a new kind of feedback.

The Timex told me what had happened.
The Tempo Trainer gave me a rhythm to follow.
The Garmin recorded what I was doing.
The Form goggles gave me information while I was doing it.

This year, I began experimenting with a different kind of swimming technology.

Artificial intelligence.

In 2025, I read Geoff Woods’s The AI-Driven Leader. The book helped me understand AI not merely as a place to ask questions, but as a thought partner.

The distinction matters.

A search engine gives me information. A thought partner helps me examine a problem. It can ask questions, consider my circumstances, challenge an assumption, organize what I know, and suggest an experiment.

I began using that approach with swimming.

It has been an absolute blast.

To be clear, I would choose my Dedham Masters coaches over AI for swimming guidance 100 times out of 100.

I would choose the hard-earned wisdom of Paul Newsome, Brenton Ford, Adam Walker, Terry Laughlin, and the many other swimmers and coaches I have learned from over anything generated by a machine.

AI cannot swim.

But a book cannot swim either. Neither can a podcast, a YouTube video, a stopwatch, or a pair of goggles.

They are tools. Their value depends on how we use them.

My experiment began when I created an AI project devoted entirely to my swimming. I called it SwimGPT.

I gave it information about my swimming history, current pace, technique challenges, training schedule, injuries, goals, and the ideas I had learned from coaches and books. Then I started uploading screenshots from my Garmin after workouts.

At first, I was simply curious about what it might notice.

After one recent 2,200-yard workout, it identified the structure of the session from my lap data. It noticed that I had repeated several 100s in the 1:37–1:39 range and later held nearly the same pace across three 300s.

The overall average pace did not look remarkable. The pattern underneath it was more encouraging.

The analysis suggested that I was learning to repeat a faster pace without giving up much stroke efficiency. It also noticed that each of my four fast 50s was quicker than the one before it.

That sounds positive, but SwimGPT offered a different interpretation: perhaps I had not swum the first one fast enough.

Instead of celebrating the fastest final repetition, it suggested trying to “own” the target speed from the beginning.

That gave me a specific experiment for my next swim.

This is where AI becomes interesting to me.

It does not simply display the data. My watch already does that. It helps me have a conversation with the data.

I can ask why the 300s matter. I can question its interpretation. I can explain that my heart-rate readings in the pool are unreliable. I can add what the workout felt like, which no watch can know. I can ask it to compare this session with an earlier one or help me design the next experiment.

The conversation keeps moving.

This is not the same as having a coach on deck. A coach can see what the data cannot: the dropped elbow, the mistimed breath, the fatigue entering my stroke. A coach also brings judgment developed through years in the water and through watching thousands of swimmers.

SwimGPT occupies a different role.

It is part library, part notebook, part analyst, and part endlessly available conversation partner. It helps me gather what I have learned from human experts, connect it with my own training, and turn it into something I can try.

A book is written for many readers.

SwimGPT feels more like a book that can ask me questions and rewrite itself as I learn.

It also holds far more context about my swimming than any generic training plan could. It knows that I am an adult-onset swimmer. It knows my current CSS pace, my preferred stroke rate, my injury history, my endurance background, and the dozen swimming problems I am most interested in solving.

Recently, I asked it to help me think through a 24-week experiment in becoming faster.

It did not hand me a magical answer. It helped me define the problem.

What kind of speed am I trying to develop? How much intensity can I absorb while keeping my shoulders healthy? How should strength training fit around swimming? Which measurements will tell us whether the experiment is working? What should change when Masters swimming resumes in the fall?

The plan will not be perfect. AI can misunderstand data, make faulty assumptions, and deliver advice with more confidence than wisdom. I remain responsible for deciding what makes sense, comparing its suggestions with trusted sources, listening to my coaches, and paying attention to my body.

I am still the swimmer.

But I now arrive at the pool with better questions.

That may be the greatest contribution technology has made to my swimming. It has not reduced swimming to numbers or automated the experience. Used well, it has made me more curious, observant, and intentional.

And then I put the technology away.

I push off the wall. The water takes my weight. The noise disappears.

No watch, goggle, coach, book, or artificial intelligence can swim the next length for me.

That part remains beautifully human.