Gaming
Intelligence
for AI training
Pleiada is at the forefront of turning pure gameplay into Gaming Intelligence — structured, curated, commercially licensed data, built specifically for AI training.
Get in Touch →The Missing Link Between Games and AI.
Just as Business Intelligence turned raw enterprise data into decision-making fuel, Gaming Intelligence turns gameplay into model-training fuel — all aggregated, governed, and delivered at scale.
We aggregate gameplay from the people who create it every day — our player community, gaming guilds and e-sport teams — and license it to AI labs and data brokers who need rich, domain-specific training data that goes far beyond what the open web can offer.
Our model aligns incentives on both sides: players earn recurring revenue from gameplay they already produce, and AI developers gain access to a structured, legally compliant data source built for the complexity modern models demand.
Inside the Dataset
Gameplay and its raw signal, side by side.
Every session ships frame-synced with its raw signal — camera pose, keyboard and mouse inputs, and semantic in-game actions, one row per video frame. Sessions carry camera data, input data, or both. Structured for training, not scraped from the web.
Canonical Format
One normalized format, ready for any pipeline.
Every session is delivered as a per-frame table (CSV or Parquet) synced to its video — one row per frame, identical column conventions across every title.
Camera-to-World matrix
Every camera frame carries the full 4×4 pose matrix in a single, consistent world convention — ready to convert downstream into whatever pose format your training pipeline expects.
Camera intrinsics
Pinhole intrinsics with every camera session — focal lengths and principal point in pixels, derived from each title's field of view and resolution. Extended distortion models supported.
Input data
Per-frame active keys, mouse buttons and high-rate mouse deltas — plus semantic in-game actions like "Sprint", "Jump" or "Reload". A per-session key-binding map ties raw keys to the player's actual settings.
Data you can trust.
Not scraped, not synthetic, not approximated — captured at the source and verified before it ships.
The same match, every point of view.
Multiplayer sessions are captured as synchronized per-player streams — one session per player, sharing a single world clock. Built for multi-agent training.
Frame-accurate, and proven.
Camera, input and video are cross-correlated before delivery — every session ships with its alignment verified at the single-frame level.
Played by humans, licensed for AI.
Every hour is played by a consenting human contributor — first-party capture, anonymized contributor IDs, full chain of custody. No scraping, no synthetic loops.
