Intro to AI Balancer: How TRU Uses Artificial Intelligence to Perfect Card Balance
Intro to AI Balancer!
Hey folks!
I'd like to introduce myself! My name is Xace, and I'm the Marketing and AI Engineering Lead here at Two Robots Studios. Recently, the TRU Team and I have made some amazing progress with our first highly accurate AI Card Balancer! This represents a huge milestone for Two Robots: Unleashed because…
…for the first time, we can actually SEE what’s happening in thousands of real games — not just guess.
For years, balancing a competitive card game has been part science, part art, and part “okay maybe this is too strong, let’s nerf it a bit and hope nobody complains.” Every studio goes through that cycle – patch - feedback - patch again – because humans just can’t watch every interaction, every weird combo, or every edge-case synergy players discover.
This problem is especially harsh in TCGs, where imbalance slowly erodes player trust over time.
But Two Robots: Unleashed is something different.
We’ve built the industry’s leading AI Card Balancer because we prioritize long-term trust.
Let me officially introduce you to our AI Balancer.
What the AI Balancer Actually Does (In Simple Terms)
Our AI Balancer digests raw gameplay telemetry. Every move, every card, every damage tick, every buff or debuff, every recursion — everything is now under the eyes of the system.
It reads:
- cards played each turn
- powers triggered
- ultimates used
- damage events
- status effects (shock, corrosion, poison, shields, etc.)
- recursion / scrapyard interactions
- actual win/loss outcomes
- player behavior and decision patterns
And then it converts all of that into a massive dataset measuring impact, not just usage.
This distinction matters:
A card isn’t strong because people use it — a card is strong because it changes the outcome of games.
The AI measures exactly that.
Why This Is So Important for TRU
Two Robots is built around assembling robots, combining effects, and generating wild synergies. That means the game is inherently complex — more complex than Chess and impossible to manually balance with human eyes alone.
Some cards barely do anything on play but quietly swing entire matches. Others only shine in niche synergies. Some explode in power when duplicated. Some look mediocre until you analyze how often they turn losing states into winning ones.
Humans miss that. The TRU AI does not.
With the balancer, we can finally identify:
- which cards consistently decide matches
- which parts overperform in specific archetypes
- which cards break when duplicated
- which powers need tuning
- which ultimates end games too reliably
- which “weak” cards are secretly sleeper-strong
This lets us create a meta that’s healthy, fun, skill-based, and E-Sport friendly — without the wild balance spikes players hate.
How the AI Calculates “Card Strength”
The brain of the operation – where raw gameplay turns into balance decisions that actually make sense.
The algorithm uses a mix of:
- actual win/loss correlations
- impact contribution during matches
- per-turn effectiveness
- long-term effect influence
- repetitive-play performance
- matchup-weighted scoring
- Wilson confidence intervals (to prevent tiny sample sizes from lying)
- normalization across thousands of games
So instead of:
“Power Gloves has 7 wins and 3 losses… maybe it’s good? Do we nerf it?”
We get:
“Power Gloves consistently contributes dynamic value by drawing extra cards. Played 76% of the time when drawn, with an 84% win rate. Impact Score: 8.76/10.”
Way clearer. Way more scientific.
What This Means for Future Patches
The AI Balancer allows us to:
- nerf problem cards with precision
- buff underperformers
- identify rank-specific issues
- detect toxic combos early
- keep the meta fresh
- dramatically speed up patch cycles
- build trust with transparent balance decisions
Instead of gut-feeling patches, we’re moving toward data-driven balancing — exactly what any modern, E-Sport-ready card game needs.
Concluding Thoughts from Xace
This project is honestly one of the coolest things I've ever worked on at Two Robots Studios. It blends everything I love — AI, engineering, competitive systems, and game design — into a tool that shapes long-term fairness, competition, and fun for our players.
We're already using it in internal testing, and the accuracy is overwhelming. This system will help our dev team move faster, balance smarter, and push TRU into a new era of competitive integrity.
More updates soon — and yes, I’ll be sharing some spicy analytics once we gather enough play data!
We’re just getting started. 🚀🤖