42 Kilometers, 1 AI Coach, and a Whole Lot of Curiosity

Running a marathon was never part of my long-term plan. Over the years, I had completed a few half marathons and enjoyed them — but 42.195 kilometers? That felt like another universe.

Then in March 2024, something unexpected happened: I won a last-minute spot in the sold-out Haspa Halbmarathon in Hamburg — just four weeks before the race. I wasn’t particularly fit. But I decided to give it a go.

Four weeks later, I crossed the finish line in 1:47h. And suddenly the thought hit me: If I can do that with minimal preparation... what’s possible with a real plan?
So I signed up for the 2025 Haspa Marathon. The goal was set. Now I needed a way to train that actually fit my life.

Coaching Apps? Books? Nope. Let’s Try AI.

Hiring a running coach wasn’t an option — too expensive. And I had tried pretty much everything else before: apps from Adidas, Nike, and Apple; marathon blogs; even “How to Run a Marathon in 40 Days” plans.

The problem was always the same: too generic, too rigid, and not tailored to my life. I’d start, get annoyed, and give up.

Professionally, I work in tech and use ChatGPT almost daily — mostly like a coding partner. One day it clicked: Why not ask it to help me train for a marathon?

My AI Training Buddy

I explained my situation to ChatGPT: my running pace, past experience, goal time (3:45), and most importantly, my weekly limits — just two runs per week, mostly evenings, plus other sports and family life.

ChatGPT responded with a custom weekly plan that actually made sense. After each session, I checked in with my feedback — and it adapted. It gave me context, explained training theory, and kept me motivated when energy was low.

I also had some serious disruptions:

  • A muscle injury in February

  • A flu outbreak in my family in March

  • A last-minute ankle twist two weeks before the marathon

Each time, ChatGPT stayed cool. It offered a recovery approach, swapped runs for mobility sessions, and encouraged me not to panic. It even helped with fueling and hydration strategies for long runs, and guided me through introducing new shoes.

It wasn’t just a coach. It became a calm voice when things got tough.

Setbacks, Laughs, and Unexpected Highlights

My back injury early in the training cycle was a real scare. I thought that might be the end of my marathon plan. But ChatGPT told me to step back, heal, and return gradually — and that worked.

When my family got sick and I couldn’t train, it reminded me I had already built a base. That helped me stay confident and mentally grounded. We even switched to mobility work to keep the habit alive.

The best moment? Probably post-race. I asked why I didn’t hit my goal of 3:45 (I ran 3:52), and ChatGPT gave me a clear, nerdy breakdown: probably glycogen stores empty around km 26, maybe hydration gaps. No drama. Just analysis. I laughed — it was like getting a race debrief from a seasoned coach who never panicked.

Race Day — No Nerves, Just Flow

The final week was all about rest and prep. The usual questions came up: Did I train enough? Will the ankle hold up? But I went into race day with calm energy. Those last “talks” with my AI coach helped.

I arrived early, warmed up, and stood at the start line with a smile. The first few kilometers were slow due to the crowd, but I found my pace and began to enjoy the race — the people, the cheers, the city vibe. Hamburg was glowing.

Then came kilometer 26. The wall. I had prepared for it mentally, but it still hit hard. I slowed down and focused on getting through one kilometer at a time.

What kept me going? Knowing my family was waiting at km 41. Seeing their faces — proud, cheering — gave me the last push. I crossed the finish line tired, but not crushed. I was done. And it felt amazing.

What I Learned — and What’s Next

This journey taught me that training with a coach — even an AI one — can make a massive difference. ChatGPT brought structure, context, and calm to a process that can feel overwhelming.

I’ve already continued using it for new goals. I’m preparing for a bouldering trip and working on grip strength. And my next half marathon is coming up, with a new goal: under 1:45.

What I learned most?
→ Trust the process.
→ Stay flexible.
→ Share your progress — even with a machine.
→ And don’t overthink setbacks. Adapt and move on.

Final Thoughts

This marathon wasn’t just about the distance. It was about learning a new way to train — one that worked for me.

No app subscription. No expensive gear. No coach yelling "No excuses."

Just me, my goal, and an AI assistant that quietly got me to the finish line.

Would I do it again?

I already am.

0
Subscribe to my newsletter

Read articles from Stephan Hoelscher directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Stephan Hoelscher
Stephan Hoelscher

Born in 1981 in Ludwigsfelde, Germany, I grew up in a working-class family and was always encouraged to be curious and seek out new experiences. During my 11th school year, my teacher introduced me to Java programming and I discovered a passion for software development. After finishing high school, I started studying computer science at the University of Potsdam but dropped out before completing my degree. Because of my wife I went on to Hamburg to work for several software development companies, eventually restarting my studies in 2010 in Business Informatics and earning my bachelor's degree in early 2015. One of my proudest professional achievements was setting up a payment gateway for some large clients in the German public transportation market. On private software development I challenged myself with a personal project, building an app to fill a gap in Facebook Pages post list feature. I believe that the future of software development lies in the use of AI and I want to stay ahead of the curve by being an early adopter of these new tools. If others asks me about a statement in the field I want to encourage them always to be curious, question patterns, and never be afraid of change.