July 17, 2026 · ← All posts

How to Turn a ChatGPT Workout Plan Into a Program You Can Actually Track

Asking ChatGPT or Claude for a workout plan has quietly become one of the most common ways people start training. It makes sense: the plans are personalized, free, and often genuinely decent. The problem shows up on day one at the gym — your "program" is a wall of chat text. You're scrolling mid-set to find what's next, you have no record of what weight you lifted last week, and by week three the thread is buried and the plan is dead.

The fix isn't to stop using AI for programming. It's to treat the chat as the planning tool and move the result somewhere built for execution. This guide covers all three steps: getting a better plan out of the AI, checking it before you trust it, and the four realistic ways to track it — including options that have nothing to do with us.

Step 1: Prompt for a plan worth following

Most bad AI workout plans come from thin prompts. "Make me a workout plan" produces a generic 3-day full-body template. Give the model the same intake a coach would ask for:

• Goal — strength, muscle, fat loss, a race, or a mix. Be specific: "add 30 lb to my squat" beats "get stronger."
• Schedule — days per week and how long each session can run. A perfect 6-day plan you can't attend is worse than a decent 3-day one.
• Equipment — full gym, dumbbells only, or bodyweight. This is the #1 cause of unusable plans.
• Experience and history — how long you've trained and your current working weights on main lifts, so prescribed loads aren't fantasy.
• Injuries or constraints — anything a movement should work around.
• Progression — explicitly ask: "include how to progress each exercise week to week, and when to deload." This is the single biggest thing AI plans omit unless asked.

One more that pays off later regardless of how you track: ask for structure. "Format the plan with clear day headings, and list each exercise as sets × reps with rest times." Structured output is easier for you to read — and easier for any app or spreadsheet to ingest.

Step 2: Sanity-check it before you trust it

Language models write confident plans, not verified ones. Before committing 8 weeks, check five things:

• Weekly volume per muscle. For most people, roughly 10–20 hard sets per muscle group per week is the evidence-backed zone. AI plans sometimes assign 30+ sets of pressing while giving legs six. Count it.
• A progression rule exists. If nothing in the plan tells you when to add weight or reps, it's a workout list, not a program.
• Rest is real. At least one full rest day; hard sessions for the same muscles aren't back-to-back days.
• The exercises exist. Occasionally an AI invents a plausible-sounding movement or mislabels one. If you can't find a demo of it, swap it for the standard version.
• Loads are relative, not absolute. Good plans say "start at 70% of your 5-rep max" or reference the numbers you gave. If it prescribes "185 lb" out of nowhere, adjust before day one.

If the plan fails a check, don't discard it — tell the model what's wrong ("this has 26 weekly sets of chest and 8 of back — rebalance it") and it will usually fix it on the next pass.

Step 3: Pick how you'll actually track it

Here are the four realistic options, honestly compared.

Option 1: Notes app or paper

Paste the plan into Apple Notes, or write the week on paper. Works when the plan is short and static — three days, same exercises weekly, you just need a reference. Breaks when progression matters: nothing reminds you what you lifted last week, rest timing is on you, and editing a text blob mid-workout with sweaty hands is miserable. Fine for week one; rarely alive by week six.

Option 2: A spreadsheet

The classic. Ask the AI to output the plan as CSV, import to Google Sheets, add columns for actual weight and reps. Works when you enjoy the control — you can build any progression logic you want, and serious lifters have run spreadsheets for decades. Breaks on friction: phone spreadsheet editing between sets is fiddly, you maintain the formulas, and there's no rest timer, exercise demo, or history view. If you already love Sheets, this is a legitimate answer.

Option 3: HevyGPT (if you use Hevy)

Hevy, a popular gym logger, offers a custom GPT that can save a routine directly into your Hevy account. Works when you're a Hevy user and you're happy generating the plan inside their custom GPT specifically. Limitations to know: it's tied to ChatGPT (not Claude or Gemini), you generate through their GPT rather than pasting a plan you already have, and what transfers is the routine — day-by-day scheduling across a multi-week program is still on you.

Option 4: Paste it into Kiron (that's us)

This exact problem is why Kiron exists, so here's the pitch — after three options that don't involve us. Kiron takes the plan you already have — from ChatGPT, Claude, Gemini, or a coach's PDF text — pasted as-is, messy formatting and all. Its parser reads the days, exercises, sets, reps, and rest periods, maps each movement to a library of 1,400+ exercises with demos, and turns the plan into scheduled workouts on a calendar. Three concrete things it handles that the options above don't:

• Any assistant, any format. It parses pasted text — you're not locked into one AI or one custom GPT.
• Multi-phase programs. A 12-week plan with base/build/peak phases becomes sequenced phases on your calendar, each starting when the last ends.
• Cardio and strength in one place. Interval runs, bike sessions, and lifts live in the same program — relevant if your AI plan mixes both.

Honest caveats: the parser needs an actual plan (days, exercises, sets or durations) — it can't structure "just tell me what to do." And after parsing you get a review screen to confirm exercise matches; a weird custom movement may need a manual pick. Kiron is a paid app with a free trial, so if a notes file or spreadsheet already works for you, keep it — the upgrade matters when progression tracking and scheduling are the things falling through the cracks.

The bottom line

AI is genuinely good at writing training plans and genuinely bad at being your training log. Prompt with a coach's intake, audit the volume and progression before you start, and then move the plan out of the chat window into whatever you'll actually open at the gym — paper, Sheets, Hevy, or Kiron. The plan that gets logged is the plan that works.

Try Kiron free — paste your plan, and it's a program.