A startup born of an old question.

Discernia began at a golf driving range. Learning the sport as an adult, the founder built himself a system that analyzed his own swing on video and organized his progress in his instructor's voice. What was a personal project became GolfEvo — and showed that the same architecture served a bigger question.

If AI can observe, measure and teach a movement, it can do the same with a decision. From there came the simulations with agent-characters for executives and board members, and Beatriz: an individual AI teacher, built to teach you to do — not to do it for you.

We call what we do by that name because AI is both the subject and the table. You learn about AI with AI itself — in a simulated board meeting, on a whiteboard that comes to life on the spot, in a debriefing that describes what the person did.

What sets us apart is not the technology, which the whole market has. It is the standard: capability is demonstrated, not declared — and what matters is what stays when the AI is switched off.

Everyone measures what a person does with the AI on. We measure what is left when it is switched off. For that we publish the LUMEN standard openly: a standard that belongs to no system it evaluates, including our own, and that invites validation and refutation.

We are an education and technology startup, founded in São Paulo in 2026. Small by design, with a single person on the team so far — and with every product used first by the person who made it.