About Supreme Mind
Supreme Mind simulates the expert on the other side of a matter, so counsel can prepare against them before the first report is served. It is a solo-founded company in New York. This page is why it exists, who is building it, how it started, and how the work is kept honest. The case for why this is happening now, and why it lands on expert witnesses first, has its own page.
Why this exists
A society governed by law rests on three branches, and the judicial branch is the one asked to decide: on the record, in public, with reasons given. It is also the branch handed the least new capability in the last century, while the technical difficulty of what it must determine has grown without limit.
Expert evidence is where that gap is widest. A court is routinely asked to choose between two credentialed economists and has no instrument for doing it beyond the persuasiveness of the people in the room. What we are building is the beginning of a scored, timestamped, publicly checkable record of technical judgment: the first infrastructure that lets an expert claim be verified rather than deferred to.
Expert testimony is where that starts, not where it ends. Institutions built properly outlast everyone who built them, and the standard here is that this company should still matter when nobody remembers who started it.
Who is building it
Richard John Van Der Aa, founder. Today I am the whole team. The first hires will be engineering and a New York litigation-side commercial lead.
I grew up in Veldhoven, where the machines every advanced chip is printed with are built, and watched a good local employer become a company the world’s most powerful governments negotiate over.
I studied applied mathematics and later data science and artificial intelligence at Eindhoven University of Technology, and left the MSc in 2023, three quarters through and on track for cum laude, because the models arriving that year could already do most of what the coursework asked. Three years applying it commercially since, on top of four building alongside the degree.
How this company started
2024, a freelance job that became the thesis. I built a language-model pipeline that researched and wrote biographies of roughly 19,500 Dutch lawyers across 5,500 firms and published them as juristi.nl, now the largest SI-searchable lawyer and law-firm database in the Netherlands. By hand that is years of a research team; it took one person a few weeks, and law is the most text-heavy industry there is.
2025, six months of asking instead of building. I asked lawyers where the hours go and watched more than a hundred legal-technology product demonstrations on the LegalTechHub channel. Almost all of it automates drafting, research or review, work that was already cheap. Very little touches the analysis, because it is hard, and PhD-class reasoning only arrived in 2025.
Early 2026, the United States and the expert-witness market. In the largest legal market in the world, partners kept giving the same answer: expert work is the line item nobody can compress. A single expert report takes months and can cost hundreds of thousands of dollars. Even the first strategic read runs $25,000 to $45,000 and takes weeks, bought before anyone knows whether the matter can carry it.
October 2026, the first measured result. Predicted blind on real expert reports, Supreme Mind reproduced 78% of the expert’s key findings; the same model on its own, 28%. See the benchmark.
How we build
None of this matters if a partner cannot rely on the output, and in litigation the standard is absolute: one fabricated citation ends a vendor relationship permanently. Four rules keep the work honest.
Citations are checked or removed
Every supporting quote is checked verbatim against the public record. Anything that cannot be verified is removed. One fabricated citation ends a vendor relationship, so every output has to be checkable.
Predictions are scored against real outcomes
A model may already have read how a decided case ended, so we do not sell replays as a track record. Predictions are recorded on pending public dockets, timestamped before they resolve, and scored when they do. The first results are on the benchmark.
For internal preparation only
It is never filed or disclosed under Rule 26. Your retained expert still signs the report and takes the stand, which keeps this clear of the admissibility fight.
Modeled on expert classes
Each expert we simulate is a de-identified composite of how a class of expert witness testifies, built from the public record. It never claims to predict what a specific named individual will say in a specific deposition.