The Atlantic CEO Nicholas Thompson just revealed he's been secretly training with a custom GPT that knows his every workout, race history, and training pattern. The former WIRED editor built his AI coach by uploading years of Strava data and race results, using it to answer questions like "Should I delay my 5-mile workout to Thursday?" The revelation comes as Thompson releases his new book about running, showcasing how AI is quietly revolutionizing personal fitness optimization beyond the typical corporate applications.
Nicholas Thompson has been running to work for years, but now he's got artificial intelligence running his training plans too. The Atlantic CEO just dropped a bombshell during a WIRED podcast interview - he's been using a custom GPT as his personal running coach, feeding it everything from workout logs to race histories to optimize his performance.
"I have a custom GPT," Thompson revealed to WIRED's Katie Drummond. "I've uploaded all my workouts. I've uploaded all my previous races. I tried to upload all my Strava logs. I've given it as much information as I can." The system works as his virtual coaching staff, answering questions like whether to delay workouts or adjust training schedules based on his complete athletic profile.
The AI coaching revelation comes as Thompson releases "The Running Ground," a book documenting his journey from decent recreational runner to holding the American 50K record for his age group. But it's his practical use of AI that's catching attention in tech circles, especially as companies rush to find real-world applications for large language models beyond workplace productivity.
Thompson's custom GPT represents something different from the typical ChatGPT interactions most people know. By uploading years of personal training data, race results, and performance metrics, he's created what amounts to a personalized fitness advisor that understands his specific patterns, limitations, and goals. "I don't have a coach I work with, so I have an AI coach. I use it all the time," he explained.
The approach mirrors what elite athletes get from human coaches - pattern recognition across vast amounts of performance data, contextual decision-making about training adjustments, and personalized advice based on historical performance. But Thompson's doing it with AI at a fraction of the cost and with 24/7 availability.
This isn't just about running faster. Thompson's marathon times dropped dramatically in his 40s, from consistent 2:40-something finishes to a blazing 2:29 - improvements that typically don't happen as runners age. While he credits elite Nike coaches for much of that breakthrough, his AI system now helps maintain and optimize that performance level.
The fitness AI space is heating up as wearable devices generate massive amounts of personal health data. Apple and Google are investing heavily in health AI features, while startups like Whoop and Oura are building more sophisticated analytics platforms. But Thompson's approach shows how individuals can create their own AI coaching systems using existing tools and personal data.
"The most genuinely useful thing it was good for was analyzing years of notes," Thompson said about AI's role in his book writing process. But his training GPT goes beyond analysis - it provides actionable coaching advice based on comprehensive understanding of his athletic profile.
The custom GPT approach raises interesting questions about the future of personalized AI. Instead of one-size-fits-all models, Thompson's created a highly specialized system that knows his specific needs, constraints, and goals. It's the kind of personalized AI that tech companies have been promising but haven't quite delivered at scale.
For runners and fitness enthusiasts, Thompson's system offers a blueprint for creating AI-powered training assistants. The key is feeding comprehensive historical data to create context for the AI's recommendations. "I will often ask it questions, like 'Should I delay the 5-mile workout to Thursday?'" he explained, treating the AI as a knowledgeable training partner rather than just an information retrieval system.
The timing is perfect as AI capabilities expand and personal data becomes more accessible. Fitness tracking devices now generate detailed metrics about sleep, heart rate variability, training load, and recovery. That data, combined with AI's pattern recognition abilities, creates opportunities for the kind of personalized coaching that was previously available only to elite athletes.
Thompson's success also highlights AI's role in democratizing expertise. Elite running coaches charge hundreds of dollars per month and have limited availability. An AI coach that understands your complete training history, responds instantly to questions, and costs pennies per interaction could make high-quality coaching advice accessible to recreational athletes.
The approach extends beyond fitness. Thompson also used AI to analyze transcripts and notes while writing his book, demonstrating how custom AI systems can serve multiple aspects of someone's professional and personal life. It's a vision of AI as a comprehensive personal assistant rather than just a workplace tool.
Thompson's custom GPT coaching system represents the early stages of truly personalized AI that extends far beyond workplace productivity tools. By feeding his complete athletic history into a specialized model, he's created something closer to an AI training partner than a generic fitness app. As wearable devices generate increasingly detailed health metrics and AI models become more sophisticated, we're likely to see more individuals creating their own specialized AI advisors for everything from fitness to nutrition to career planning. The real innovation isn't the AI itself - it's how Thompson has thoughtfully curated and structured his personal data to create genuinely useful, contextual advice. For athletes and fitness enthusiasts, his approach offers a roadmap for leveraging AI to achieve the kind of personalized coaching that was once reserved for elite competitors.
