Skip to main content

Tech R&D

We dream in \(p(y\mid x,\mathcal D)=\int p(y\mid x,\theta)\,p(\theta\mid \mathcal D)\,d\theta\)

Then we apply it to games: turning frontier uncertainty research into production systems studios can actually use.

We do frontier research where uncertainty, taste, and cultural dynamics aren’t edge cases. They’re the whole problem. Three pillars, each pushing a different boundary of what’s possible in game intelligence.

Intelligence-in-Flow

How should humans and models share reasoning in creative environments where there’s no stable ground truth? We’re researching trust calibration, temporal cognition, and joint reasoning models that know when to defer to intuition versus model evidence.

What this becomes: Embedded reasoning UX and decision-support primitives teams use inside production workflows.

Related briefs

  • SATE An online architecture for topic segmentation and tracking with bounded memory and bounded revision.
    Read More

Behavioral Intelligence

Can we map player cognition and cultural dynamics to market outcomes with real predictive power? We’re building models for motivational states from telemetry, cultural diffusion through networks, creative-market fit forecasting, and the feedback loops between communities and development decisions.

What this becomes: A foresight layer for production and marketing, powered by proprietary behavioral data and models.

Computational Intelligence Architecture

How do you run explainable probabilistic reasoning over open, changing creative systems in real time? We’re working on dynamic graph inference, automated ontology induction, system-dynamics simulation for counterfactual scenarios, and continuous self-calibration under drift.

What this becomes: The core infrastructure behind live reasoning, forecasting, and strategy across the Iridae stack.

Related briefs