The trust layer
for AI.
Helping people know when to trust AI,
and what to do when they can't.
Scroll
AI delivers polished answers, forecasts, and recommendations in seconds, with the same confidence even when the evidence is weak. Producing an accurate output is no longer the hard part. It is knowing when to rely on it, when to challenge it, and how uncertainty should change the next step.
Much of machine learning has focused on prediction. Less attention has gone to causal impact, uncertainty, trust, and how decisions should be made when the evidence is incomplete. That is where illuminAI is focused.
AI gives fluent answers with the same confidence whether right or wrong. Knowing when not to answer is still one of the hard parts.
Many systems optimise for a reward without carrying uncertainty through to the decision. Better forecasts do not automatically mean better decisions.
As AI systems pass work between models, tools, and people, the hard question is when to trust an output, when to challenge it, and when to stop.
illuminAI is built on a research-grade statistical machine learning foundation, focused on making AI confidence, uncertainty, and decision-making measurable.
The underlying work treats trust as contextual. It's inferred from evidence, behaviour, and outcomes.
In practice, this means asking these questions: what may happen, how wrong could we be, what changes when a specific action is taken, and how much to believe a specific person, group, customer, agent, or system relative to similar contexts.
Estimating what is likely to happen with high accuracy and granularity.
What changes when an action is taken.
How wrong we may be in this specific context, not on average.
How much to believe a source relative to similar cases. Inferred and earned over time.
Founded by Mitch Prevett · Actuary and AI researcher · Sydney