ProCap Financial’s specialized AI lab, Silvia, outranked top frontier models on complex tax queries and opened its benchmark framework for public testing.
On August 10, 2026, ProCap Financial, Inc. (Nasdaq: BRR), the first publicly traded agentic finance firm, published benchmark research demonstrating that its finance-focused AI agent lab, Silvia, outperformed leading frontier and open-source models on tax queries. Founded in 2025, ProCap raised over USD$750 million prior to listing. Silvia commands significant scale, connecting to over USD$50 billion in assets across more than 20,000 active users.
The internal evaluation tested seven AI products across ten expert scenarios involving federal law and tax codes for California, Texas, North Carolina, New York, and Florida. Silvia scored a top-ranking 8.73 out of 10 for factual accuracy, ahead of second-place Claude Desktop (8.40).
Silvia’s performance relied on a specialized architecture retrieving answers from a proprietary primary tax law library, engineered by a four-person AI team. ProCap presented this as evidence that small, specialized teams using targeted architecture can outperform massive general-purpose frontier models on domain-specific tasks.
Chairman & CEO Anthony Pompliano noted, “There is an unresolved question among institutional investors about whether general-purpose model providers will capture most of the value in AI or whether domain-specific systems will take share… Silvia beat all of them on tax with a team of only four AI engineers working on this problem.”
To ensure transparency, ProCap open-sourced the benchmark evaluation framework, methodology, and questions, enabling third parties to reproduce findings, contest scores, or test competing models.
