AI evaluation & applied research
Better evidence
about AI.
FSI Platforms is an independent research and engineering company exploring how to evaluate AI systems in practice.
Our work
Evaluation through
experimentation.
We develop test environments, measurement tools, and experiments to understand how AI systems behave—and what their behavior means for the people using them.
AI evaluationSimulationApplied research
Get in touch ↗Selected projects
Research in practice.
MINT
Economic test environments for AI agents. We study how agents’ decisions affect client payoffs and market outcomes, using responsive simulations and controlled comparisons.
AI evaluation · Experimental markets
AURA
Run-time assurance for autonomous systems moving from test to operational use.
What it costs to prove autonomy works ↗