Classifai.

What it does

Classifai assigns institutional taxonomy labels to documents, records, and unstructured text — at scale, with explainable rationale on every classification. It is built for the moment an institution stops trusting generic out-of-the-box classifiers and starts needing labels that match its own policy language.

How it works

You bring the taxonomy — codes, definitions, the edge cases your team argues about in review. Classifai fits a hybrid pipeline (deterministic rules where the taxonomy is unambiguous, an LLM judge where it isn’t) and emits a label plus a structured rationale: which clause of the taxonomy triggered, what evidence in the source supported it, what confidence the system claims. Reviewers can sample, correct, and feed corrections back without rebuilding the pipeline.

Deployment & governance

Runs as an API your applications call, or as a batch job over a folder of documents. Deployable on-prem when data residency rules require it. Every classification is logged with prompt, model, version, and rationale — auditable on demand. Policy-checked through PromptOps when the two are deployed together.