Gaming is our data engine.

We use competitive gameplay to observe human behavior under pressure — turning millions of in-game decisions into data for behavioral AI.

A measurable environment for human behavior at scale.

Competitive games repeatedly place people in situations with limited time, limited information and direct consequences. Every match produces a continuous record of how players perceive, decide, coordinate and adapt under pressure.

  1. 01Capture

    Actions

    Every movement, input and reaction leaves a data trace. We can observe continuous behavior — not only kills, scores or final results.

  2. 02Reconstruct

    Context

    The same action can be correct or disastrous. We reconstruct what the player could see, hear and know when the decision was made.

  3. 03Measure

    Outcomes

    Every decision changes the game state. We compare what happened before, during and after to see how actions shaped the outcome.

The platform behind the data.

Rankacy.com is our competitive Counter-Strike 2 platform. It turns raw match data into performance analytics, personalized coaching and shareable video — all powered by the same underlying data layer we use to understand player behavior.

>12 MIL

players analyzed across
our gaming datasets

Including more than 80,000 Rankacy.com platform users.

>4.5 TB

New gameplay dataprocessed every day

>5 MIL

Counter-Strike .dem filesprocessed to date

>1200 TB

Game data representing>5,000 years

From demos to decisions.

GPRT — the General Pretrained Rankacy Transformer — is our proprietary AI model for understanding and predicting player behavior in Counter-Strike 2. It processes complete match demos — every moment, every shot, every decision — to explain what is happening now and how players and teams evolve over time.

parameters
75M
encoder blocks
8
training instances
160M
GPRT architecture showing player data transformed into team-state and player-timeline tokens, processed by a transformer encoder and multitask prediction heads

Tools built to understand why.

Predictions tell us what is likely to happen. Our proprietary research tools reveal what a player could perceive, how they reacted, and the context behind each decision.

Perceive

Visibility
Engine

Reconstructs exactly when an opponent entered a player’s field of view and how much of them was exposed.

Counter-Strike player partially hidden behind a wall, with a translucent outline showing the body area detected by the Visibility Engine
Why it matters

It lets us measure decisions from the true visual stimulus — not from the shot, damage, or kill that followed.