The Playbook

You are not building a fitness app.

You are building: a real-time decision engine for food

If outputs feel like advice, you lose
If outputs feel like decisions, you win

Your wedge order:

Macro correction (primary entry)
Damage control (retention engine)
Fridge mode (support layer)

Everything revolves around: "When your day goes off plan, get one exact next move in seconds."

DECISION ENGINE (NON-NEGOTIABLE)

Every output must follow:

Identify the problem, what is wrong in their situation
Apply constraint, what they ate or what they have
Force correction, remove, reduce, replace, or add ONE thing
Output Format: Next move: [specific action] Why: [specific reasoning tied to their situation]

QUALITY BAR (BURN THIS INTO YOUR BRAIN)

Bad: "Try eating more protein"

Good: "You had carbs without protein, that’s why you’re hungry again. Add 30g protein in your next meal."

REDDIT TARGET THREADS (CORE GTM WEAPON)

You will repeatedly use these threads:

Macro correction (PRIMARY)

Damage control (HIGH URGENCY)

Fridge mode (SUPPORT)

COMMENT STRATEGY (THIS IS LIFE OR DEATH)

Never say: "I built something"

Always:

Diagnose "You’re not actually overeating here, you just have no protein anchor after that meal."
Explain "That’s why hunger spikes later."
Ask "Want me to give you a quick next move for your case?"

14 DAY EXECUTION PLAN

DAY 1

Build MVP (2–4 hours max)

Single web page:

3 buttons (macro, messed up, what to eat)
text input
output (next move + why)

Then:

Post 10 HIGH QUALITY Reddit comments
Not generic. Diagnose each case.

Goal: Get 3–5 replies

DAY 2

Reply to every comment

Send DM:

"Based on what you said:

Next move: X
Why: Y

Try here: [link]"

Track:

who clicks
who replies again

DAY 3

Improve outputs

Look at failures:

where did it feel generic?
where no "aha"?

Fix logic manually

Post 10 more comments

DAY 4

Introduce “aha enforcement”

Every output must:

reveal something they missed
simplify confusion

If not then rewrite

Talk to 5 users manually

DAY 5

Start continuity

If user returns:

"Last time you had no protein anchor. Did you try fixing that?"

Now product feels alive

DAY 6

Content starts

Post 2 short videos:

"why you're always hungry after this"
"this one mistake breaks your diet"

DAY 7

Evaluate:

Do any users return?

If NO: Your outputs are weak

Fix decision engine, not UI

DAY 8

Push damage control mode harder

Find binge posts

Respond with: calm + correction (not judgment)

DAY 9

Fridge mode integration

Use logic:

NOT: "make this meal"

BUT: "given what you ate + what you have, do this"

DAY 10

Add light feedback loop

"Was this helpful?"

Track yes/no

DAY 11

Identify repeat users

Message them:

"Want faster responses? I can prioritize you"

Build early core group

DAY 12

Add soft monetization test

ONLY if repeat usage exists

"Want unlimited daily fixes?"

Stripe link

DAY 13

Double down on best use case

If macro correction wins: focus ONLY on that

Ignore rest temporarily

DAY 14

PMF CHECK

You only win if:

users return without prompting
users ask again
users say "this helped me decide"

If not: Your outputs are still advice, not decisions

FRIDGE MODE (FINAL CLARITY)

Fridge mode is NOT:

meal generator

It IS:

macro correction with constraints

Example:

Input: "I have eggs, bread, cheese. Ate pasta earlier"

Output: Next move: Skip bread, do 3 eggs

Why: You already had carbs, this balances protein

WHAT NOT TO DO

Do not:

build full app
add tracking
add dashboards
give multiple options
sound like ChatGPT

FINAL RULE

Your entire startup depends on this:

Does your output feel like a decision?

If yes: people return

If no: they leave

DAILY MINIMUMS

Every day:

10 high-quality Reddit replies
3–5 DMs
5 real outputs tested
1 improvement to decision logic

Miss this = no progress