Here's the whole system, with worked numbers.

Step 1: Set calories (the only thing that changes your weight)

Weight change is driven by energy balance. A practical starting point:

• Maintenance ≈ bodyweight in kg × 33 (or lbs × 15)
• Fat loss — subtract 300–500 kcal from maintenance
• Muscle gain — add 200–300 kcal to maintenance

These are estimates. Track your weight for two weeks; if it isn't moving the way you want, adjust by ~200 kcal and reassess.

Step 2: Set protein (the priority macro)

Protein preserves and builds muscle, and it's the most satiating macro — both critical in a deficit.

• Target: 1.6–2.2 g per kg of bodyweight (0.7–1 g per lb)
• Cutting? Stay at the high end.
• Each gram of protein = 4 kcal.

An 80 kg person aiming for 2 g/kg eats 160 g protein/day (640 kcal).

Step 3: Fill in fat and carbs

Both matter; neither is evil. A sane default:

• Fat: 0.6–1 g per kg of bodyweight — hormones and vitamin absorption need it. Each gram = 9 kcal.
• Carbs: everything left — they fuel training. Each gram = 4 kcal.

Higher-carb splits suit people who train hard and often; higher-fat splits suit people who prefer that eating style. Adherence beats optimization — pick the split you can sustain.

Worked example: 80 kg, fat loss

Daily targets: 2,200 kcal, 160 g protein, 64 g fat, 246 g carbs.

Turning targets into meals

Targets don't cook dinner. The practical layer:

• Anchor each meal on protein — 30–50 g per meal across 3–5 meals hits a 160 g target without heroics.
• Repeat breakfasts and lunches — variety at every meal is the enemy of adherence. Rotate 2–3 options.
• Plan the week, not the day — one weekly plan plus a matching grocery list removes daily decision fatigue. This is the part FitDrake automates: it converts your calorie and macro targets into a week of meals that fit them.

When to adjust

Recheck every 2–3 weeks against the scale trend (weekly average, not daily readings):

• Cutting but weight flat → drop ~200 kcal, mostly from carbs or fat, never from protein.
• Gaining too fast (>0.5% bodyweight/week) → trim ~150 kcal.
• Training performance tanking on a cut → shift calories toward carbs around workouts.

Key takeaways

• Calories control weight change; protein (1.6–2.2 g/kg) protects muscle; carbs and fat fill the rest by preference.
• An 80 kg person cutting lands around 2,200 kcal — 160 g protein, 64 g fat, 246 g carbs.
• Adjust every 2–3 weeks based on the weekly weight trend, and automate the meal-level planning so adherence stays easy.

Why your body stops changing without it

Muscle growth and strength gain are adaptations to stress. Lift a weight your body already handles comfortably and there is nothing to adapt to — you maintain, but you don't improve. This is why the person doing the same three sets of ten with the same dumbbells for a year looks the same as they did a year ago.

The fix isn't training harder in one heroic session. It's training slightly harder than last time, repeatedly, for months.

Five ways to overload (weight is only one)

Adding weight to the bar is the most obvious progression, but it's one lever among five:

1. Load — add weight. The classic. Works best on compound lifts (squat, deadlift, press, row).
2. Reps — same weight, more reps. Going from 8 to 12 reps with the same dumbbells is real progression.
3. Sets — add a set. Total weekly volume per muscle is one of the best predictors of growth.
4. Density — same work, less rest. Doing the same session in 45 minutes instead of 60 is a harder session.
5. Execution — fuller range of motion, slower eccentrics, stricter form. Ten deep squats beat fifteen quarter-squats.

A practical rule: progress reps first, then load. Work within a rep range (say 8–12), add reps each week until you hit the top of the range on all sets, then add weight and drop back to the bottom.

A sample four-week progression

Dumbbell bench press, rep-range method, 3 sets with 10–12 rep target:

Small, boring, relentless. That's what working programs look like.

The mistakes that break it

• Adding weight too fast. Form degrades, reps get partial, and the "overload" becomes fake. If you added 10% and your range of motion shrank, you didn't get stronger — you changed the exercise.
• Never deloading. Fatigue accumulates. Every 4–8 weeks, an easier week (reduce volume ~40–50%) lets adaptations consolidate. Plateaus and nagging aches are usually the signal.
• Not tracking. You cannot progress from numbers you don't remember. Log every session — weight, reps, and how hard it felt. This is exactly the bookkeeping an app should do for you; FitDrake tracks it and prescribes the next increment automatically.
• Chasing soreness. Soreness measures novelty, not productivity. The goal is progression in the log, not pain the next morning.

Key takeaways

• Progressive overload — gradually increasing training demands — is the non-negotiable driver of muscle and strength gains.
• Load, reps, sets, density, and execution are all valid progression levers; reps-then-load is the simplest system.
• Track every session, progress in small increments, and deload every 4–8 weeks.

This post explains what data actually drives that personalization, how weekly adaptation works in practice, and where AI planning genuinely outperforms static programs.

What data an AI plan is built from

A template asks one question: "What's your goal?" A personalized plan starts from a much wider picture:

• Goal and timeline — fat loss, muscle gain, strength, or general fitness, and how fast you want to get there.
• Training history — a first-year lifter and a fifth-year lifter need different volume, intensity, and exercise complexity.
• Equipment and environment — full gym, dumbbells at home, or bodyweight only. A good plan never prescribes a machine you don't have.
• Schedule — three 45-minute sessions and five 90-minute sessions are different programs, not the same program compressed.
• Body metrics — weight, measurements, and how they trend over time.

FitDrake collects this once at onboarding, then keeps the plan honest with weekly check-ins.

How weekly adaptation works

Static programs assume every week goes to plan. Real life doesn't cooperate. Adaptive programming closes the loop:

1. You train and log — completed sessions, skipped sessions, and how hard the work felt.
2. You check in weekly — a short prompt about energy, soreness, adherence, and progress.
3. The plan updates — next week's volume, exercise selection, and intensity shift based on what actually happened, not what was supposed to happen.

Missed two sessions? The plan redistributes work instead of piling it on. Crushed every set? Progression accelerates. Reported nagging shoulder discomfort? Pressing volume comes down and movement selection changes.

Why this beats a template

Templates fail for a predictable reason: they are written for the average person, and almost nobody is average. Three concrete advantages of adaptive planning:

• Progressive overload is managed for you. The single biggest driver of results is doing slightly more over time. An adaptive plan tracks what you did and prescribes the next increment, so you're never guessing.
• Plateaus get a response. When progress stalls, the plan changes — volume, exercise variation, or intensity — instead of repeating the same week and hoping.
• Consistency survives real life. The best program is the one you can actually follow. A plan that reshapes itself around a busy week keeps you training instead of quitting.

Where humans still matter

AI planning is a tool, not a replacement for judgment. See a physician before starting a new program if you have health conditions, and see a qualified coach for sport-specific technique work. What AI does well is the unglamorous middle: tracking, adjusting, and programming week after week without fatigue or bias.

Key takeaways

• AI workout plans are generated from your goals, history, equipment, schedule, and metrics — then adapt weekly.
• The feedback loop (train → check in → adjust) is what separates them from static templates.
• The biggest practical wins: automated progressive overload, plateau responses, and plans that survive missed sessions.