An honest look at ChatGPT and other AI chatbots as workout planners: what studies found, what they do well, where they fall short, and how to get a better plan from them.
Yes, ChatGPT can make a reasonable starting workout plan, especially with a detailed prompt. Studies from 2024 and 2025 found AI-generated plans followed general guidelines and improved with more detailed input, but experts said they need fine-tuning before use. The bigger limit comes after week one: a general chatbot doesn't see your logged sets or effort unless you paste them in, so week-to-week progression stays generic.
Asking ChatGPT for a workout plan is now one of the most common ways people start lifting. You type "give me a 4-day plan to build muscle" and get a neat table in seconds. It is free, fast, and usually looks convincing.
So is it any good? Researchers have tested exactly this, and the answer is more nuanced than "yes" or "no."
It can write a decent first draft. In a 2024 study, Washif and colleagues asked GPT-3.5 and GPT-4 to design 12-week resistance training programs for hypothetical 20-year-old intermediate lifters, then compared them with established guidelines. ChatGPT produced structured programs with three 4-week phases, sensible frequencies, rep ranges, and rest periods. The authors concluded the plans "likely require additional fine-tuning" and that ChatGPT may serve as a tool for writing a draft program, with human expertise needed to get the most from it.
A 2025 study by Havers and colleagues had 12 coaching experts rate muscle-building plans from GPT-4 and Google Gemini. Plans were highly reproducible when the same prompt was repeated, GPT-4 was rated higher on several quality criteria, and more detailed prompts produced higher-rated plans.
ChatGPT is good at building a sensible plan structure, explaining training concepts, and adapting a plan to constraints you give it. More specifically:
For someone who has no plan at all, a well-prompted chatbot plan is a big step up from wandering around the gym.
ChatGPT falls short mainly on progression, because it does not see your actual logged sets unless you paste them in.
The main limits show up after the plan is written, when you actually start training.
Give it detail. Because more information produced better plans in the research, a strong prompt should include:
Then ask it to explain its choices, and check them against basics like those in the American College of Sports Medicine's 2009 position stand on progression: train regularly, increase load in small steps once you exceed your target reps, and manage fatigue.
A general chatbot is best at writing and explaining plans; a logging AI coach is best at recording what you actually lifted and setting next session's numbers from it.
| Need | General chatbot | Logging AI coach |
|---|---|---|
| Write a first plan | Good with a detailed prompt | Good |
| Explain training concepts | Strong | Good |
| Remember every set you logged | Only what you paste in | Stores every set |
| Next-session weights from your actual data | Generic unless you supply history | Calculated from your log |
| Effort tracking per set | Only if you describe it each time | Built in |
| Progress charts | Not from your real history | From your log |
| Form checks | No | No |
Neither one replaces a qualified coach for technique or injuries. For that comparison, see our guide on AI coaches vs. personal trainers.
Yes, as a starting point, then pair it with a log. The plan itself matters less than what happens over the next 12 weeks: showing up, recording what you did, and adding reps or weight at the right time. Whatever tool you use, write down every working set. Our guide on how to track your workouts shows what to record, and double progression shows how to turn those records into next week's numbers. If you would rather have an app build and update the plan for you, see our roundup of the best workout planner apps. For free tools that build a plan for you, see our list of the best free workout plan generators.
Alexandros is a personal lifting coach you talk to in a chat, currently in early access. Like a chatbot, you talk to it in plain English. The difference is that it is built around your log: describe a session, like "db bench 4 × 10 with the 50s, last set was a grind," and it records every set with an effort rating, from very easy to max effort, or reps in reserve. Every number in its replies is computed from sets you actually recorded.
It then tells you what to lift next time using double progression: add weight at the top of the rep range with reps to spare, back off after a ground-out set, and deload after three flat sessions, with realistic jumps like 5 lb per hand on dumbbells or 2.5 lb plates per side on a barbell. It keeps a long-term memory of injuries and goals that you can review in Settings, and answers "How has my squat progressed?" with a chart of top weight, estimated 1RM, or volume. Like ChatGPT, it cannot watch you lift.
It is good for a first draft, especially with a detailed prompt. Studies found its plans followed general guidelines but needed fine-tuning by someone with training knowledge.
Only if you give it your data. A general chatbot does not keep a structured record of your sets, so you would need to paste in your training history each time you want advice based on it.
Plans for healthy adults are usually conservative. If you have an injury or a medical condition, check the plan with a qualified professional.
Include your goal, days per week, session length, experience, current weights, equipment, injuries, and how you want to progress. Research found more detailed prompts produced higher-quality plans.
Not for form coaching or injury work. AI tools are strongest at planning, logging, and applying progression rules consistently.
