When to use it
The mechanics work, but the game feels "too easy, then a brick wall", and resources are either overflowing or an endless grind. You need concrete formulas and numbers, not eyeballing. The AI's role is a systems designer / balancer. Output: progression tables and formulas you can drop straight into the game.
The prompt (copy and paste)
You are a systems game designer responsible for balance and economy. Reason step by step.
GAME AND GENRE: <PASTE>
WHAT WE'RE BALANCING: <e.g. XP→levels and upgrade costs / the gold economy / weapon damage>
INPUT DATA: <current numbers, session length, target completion time, etc.>
TARGET FEEL: <e.g. "the first 5 levels are easy, then it becomes a challenge">
Do this:
1. Propose a progression formula (e.g. the cost of level N) and explain why this curve rather than a linear one.
2. Give a table for 10-15 steps: level/tier → requirement → reward → running total.
3. Economy: sources and sinks for the resource; show whether it balances out (no permanent inflation/deflation).
4. Find the bottlenecks: where the grind is, where difficulty spikes (the wall), where a dominant strategy breaks the balance.
5. Give 2-3 "knobs" a designer can use to adjust the feel quickly, and which way to turn them.
State every number explicitly. Where you assumed a metric — mark it [assumption]. Don't hand back "roughly fine" without numbers.
The technique: step-by-step reasoning (CoT) over the numbers plus a demand for tables keeps the model from answering "balance it to taste" — it has to show the math.
Filled-in example
Game: an idle RPG. Balancing: the gold cost of hero upgrades. Input: base price 100, ~50 gold per fight, a fight takes 30 seconds. Goal: an upgrade every ~2-3 fights early on, less often later.
What the AI should come back with: the formula cost(N) = 100 × 1.15^N with a rationale for the exponent; a 15-level table with cumulative gold; the conclusion "the gold source is linear, the sinks are exponential → by level 12 the grind is 8 fights, that's a wall"; knobs — "dropping the multiplier 1.15→1.12 softens it, gold per fight ×1.5 speeds up the early pace".
Variations
- Combat balance. "Compare 5 weapons by DPS including reload time and find the dominant one" — the cure for a "meta weapon".
- F2P economy. Add a premium currency and check the game is completable without paying, while paying saves time.
- Simulation. "Simulate 100 fights with these numbers and show me at which level the player stalls" — a rough Monte Carlo in prose.
Pro tips
- The big mistake is trusting the final table without looking at the formula. The value is in the reasoning: read why the curve is shaped that way.
- Ask for "sources vs sinks" as its own block — an economy that doesn't balance kills a game more quietly and more surely than a difficulty curve does.
- Numbers from an AI are a starting point, not the truth. Final balance still gets tuned on live sessions; the AI just saves you the first 10 iterations.