The Company
Origin Foundry
Origin Foundry builds software infrastructure for autonomous simulation systems.
The company's focus is simulation intelligence, agent execution systems, and experimentation platforms. Its flagship product is Biomata.
Mission
Infrastructure first.
Autonomous simulation is one of the most generative ideas in software: systems that contain their own cause and effect, agents that form and act on beliefs, environments that evolve without scripts. The research potential is real. The application space, in games and modeling and synthetic environments, is real.
But the infrastructure to do it rigorously has lagged. Systems stitched together from game engines, academic simulators, and custom scaffolding lack observability, reproducibility, and the ability to scale. Researchers can't trust their results; developers can't ship reliably.
Origin Foundry builds that missing layer. Biomata is not a research toy or a game AI toolkit. It is production infrastructure: designed for rigor, built for scale, and open to the full range of simulation applications.
Principles
How we think about simulation.
Observability is not optional.
A simulation you cannot see inside is a simulation you cannot trust. Every layer of Biomata is built to be instrumented, traced, and inspected. Cognition is not a black box.
Reproducibility enables science.
If two runs of the same configuration produce different results, you have a stochastic system, not an experiment. Deterministic execution modes are a first-class primitive.
Modularity allows comparison.
You can only understand a cognitive component if you can vary it in isolation. Biomata separates memory, decision-making, perception, and action into independently composable modules.
Scale is a design requirement.
Infrastructure that collapses at 500 agents is not infrastructure. Biomata is designed for production-scale simulation from the ground up.
Integration, not replacement.
Biomata sits inside existing pipelines. The SDK, engine adapters, and open protocol are not afterthoughts. Most users already have a game engine or research environment. Biomata connects to it.
Simulation is not the goal.
The goal is knowledge, or game quality, or robust AI systems. Simulation is the method. Biomata stays oriented toward that end: what matters is what you learn or build, not the simulation itself.
Positioning
What Biomata is not.
Biomata is not just a game AI toolkit. It is infrastructure that happens to serve game developers, among others.
Biomata is not just a research simulator. Its production scale and integration capabilities matter as much as its experimental rigor.
Biomata is not an LLM wrapper. Agent cognition in Biomata is a structured execution model. Language models can be cognitive components within that model; they are not the model itself.
Biomata is not agent orchestration middleware. Orchestration is one subsystem. The scope is a complete simulation stack: world, agent, cognition, observation, evaluation, and scale.
Build with Biomata.
Explore the platform or reach out to talk about what you're building.