- \“ai can only go off of the data you give it… you have to do the work yourself because it’s only going to give you the results if you feed it the right info.\”
- \“the problem with ai is people expect it to do all the work for you… if you treat ai like a google engine and just follow its instructions without thinking, then yes, you are going to get dumber.\”
- \“you’re pretty much your own coach at this point. ai is just a tool helping you… it’s not going to tell you to wake up, go to the gym, track your foods, or drink your water—it’s really only there as a resource.\“
the creator shares how he used ai (chatgpt and claude) as his marathon coach to improve his half-marathon time by over 15 minutes, emphasizing that ai is a powerful tool but only when fed comprehensive data and treated like a real coach with daily updates. he explains the pros and cons of both ai platforms, recommending chatgpt for its memory retention despite claude’s more personalized responses, and stresses that users must maintain accountability and critical thinking rather than blindly following ai suggestions. the key to success is the continuous cycle of planning, executing, tracking data, feeding it back to ai, and adjusting—while always double-checking the ai’s work and following your own intuition.
What are the crucial points in this article or video that make it iconic, ideas I want to remember for the rest of my life?
- data quality determines output quality - any tool, system, or coach can only work with the information you provide; comprehensive, honest input is essential for meaningful results.
- tools don’t replace accountability - technology can guide and inform, but execution, discipline, and personal responsibility remain entirely human responsibilities that cannot be outsourced.
- critical thinking must accompany automation - using powerful tools without questioning their output or applying personal judgment leads to dependency and diminished capability rather than enhancement.
the creator wants to demystify ai as a coaching tool and show that when used thoughtfully—with proper data input, consistent updates, and critical thinking—it can be a powerful resource for athletic training, while emphasizing that ai is only a tool that requires human accountability and judgment to be effective.
- athlete profile building: creating comprehensive personal data (age, height, weight, training frequency, schedule, injuries, past performance) as the foundation for ai coaching
- dual ai system: using claude for big-picture program design and chatgpt for daily coaching updates and tracking
- feedback loop cycle: plan → execute → track data → feed back to ai → receive feedback → adjust (continuous improvement cycle)
- holistic training structure: integrating running programming with nutrition protocols (carb loading), recovery routines (mobility, ice baths, saunas), and rest periods
- screenshot and document all workout data (paces, splits, times) from fitness tracking apps
- provide ai with initial 2 weeks of baseline training data before requesting pace targets
- update ai daily with run performance and weekly with comprehensive summaries
- create a living document to maintain training history when using claude (due to chat expiration)
- ask ai for mobility routines, carb loading protocols, calorie targets, and recovery schedules
- double-check ai’s programming suggestions and question inconsistencies (\“are you sure?\”)
- feed chatgpt the program claude creates, then summarize weekly progress back to claude
- treat ai like texting a real coach—update on missed workouts, diet changes, injuries, and schedule disruptions
How was this video or article relevant to my current life? Did it answer a specific question, enlighten me on a topic, etc.
this is directly applicable to where I am in running. I’ve been using AI for content and vault work but hadn’t seriously considered it as a training coach — especially given I don’t have a coach and I’m building zone 2 base. the key takeaway: AI coaching only works when fed detailed daily data (paces, splits, HR from Coros), not vague updates. the accountability cycle (plan → data → AI feedback → adapt) is essentially what I already do with the vault, applied to running. and benny’s comparison between ChatGPT and Claude is useful — Claude being more personalized but lacking memory retention is exactly the limitation I’d run into.
- how can we balance the efficiency gains of ai coaching with the accountability and motivation that human coaches provide?
- what types of athletic training or personal development are best suited for ai assistance versus those requiring human expertise?
- as ai becomes more integrated into personal training, how do we maintain critical thinking skills and avoid over-reliance?
- what is the optimal amount of data to provide ai—when does more information improve results versus create noise?
- how might ai coaching democratize access to personalized training for people who cannot afford human coaches?
- what are the long-term psychological effects of using ai as a primary accountability partner versus human relationships?
- chatgpt (openai) - ai platform with memory retention across chats, $20/month
- claude (anthropic) - ai platform with more personalized responses but chat expiration limitations, $20/month
- garmin/coros fitness tracking apps - for capturing workout data (paces, splits, heart rate)
- sony 11mm f1.8 lens - aps-c lens mentioned for vlogging (not directly related to ai coaching but shows creator’s equipment)
- post-run data dump — after every run, paste the Coros summary (distance, pace, HR zones, splits) into the AI coach chat alongside a one-sentence qualitative note. no data = no good coaching.
- weekly AI training review — on Sunday, give the AI a full week of run data and ask: what pattern do you see? what should I change this week?
- log run in the daily note with Coros data included; don’t just write “did 5km.”
- set up a dedicated AI training thread (ChatGPT for memory retention) and paste in my current running baseline data: weekly mileage, target race, current zone 2 pace
- test one week of daily AI coaching updates using Coros data and compare the quality of the plan it outputs
- research whether Coros has any export format that makes it easier to paste data into an AI prompt
- define my current running goal specifically: target race, target time, current fitness level — give this to the AI as context
- compare ChatGPT vs Claude for this use case after a 2-week trial and pick one