Criterica Intelligence's launch formalizes the relationship between the Criterica institutional identity and the underlying technology that makes it work: Splitifi, the legal data science platform that builds and owns every production outcome model, and the court-records infrastructure those models are trained against. This is not a licensing or reseller relationship — the model fleet and the data infrastructure beneath Criterica Intelligence are Splitifi-built assets, developed and owned outright, not assembled from third-party components or rented compute against someone else's data.
The distinction matters for how institutional buyers should evaluate the platform. A vendor that licenses a foundation model and layers a legal-specific prompt or interface on top is exposed to that foundation model's limitations, pricing, and roadmap decisions, none of which the vendor controls. Criterica Intelligence's models are calibrated statistical models — not language-model inferences — built and validated in-house against a corpus of real, filed court and regulatory records spanning federal circuits, state courts, and international tribunals. Ownership of the full stack, from raw record ingestion through model validation, is what lets Criterica Intelligence make specific, checkable claims about how a given model was built and what population it was tested against.
Criterica is the institutional identity — the name and go-to-market presence under which this Splitifi-built platform engages law firms, litigation finance funds, insurance carriers, and enterprise legal departments. The relationship is disclosed rather than obscured because institutional buyers doing diligence on a data science vendor deserve a straight answer to the question of who actually built and owns the technology they are being asked to rely on.
This launch also sets the pattern for how Criterica Intelligence will describe its own infrastructure going forward: real filed records; calibrated statistical models, not general-purpose language-model outputs dressed up as predictions; and a corpus and model registry that can be inspected, questioned, and validated rather than taken on faith. Institutional buyers are encouraged to ask the same questions of any vendor in this space — who built the models, what were they trained on, and how was that training population validated.
Law firms, litigation finance funds, insurance carriers, and enterprise legal departments engaging with Criterica Intelligence for the first time can expect the same standard applied consistently across every product: a stated methodology, a traceable data provenance, and validation results reported honestly, including where a model has not yet cleared the bar for production use.