“DEMI transforms any written work into a permanent, verifiable intellectual record by generating a tamper-evident SHA-256 fingerprint, a structured Conceptual Core extract, and a conversational AI edition—all sharing a single UPL Registry Code—in one pipeline pass.”
What DEMI is DEMI (Digital Evidence and Manuscript Identification) is a Publishing Pipeline and Central Registry. In one pass through the pipeline, a work moves through six stages: ● Fingerprint — the source file is hashed (SHA-256) inside the author's own browser, using the Web Crypto API. The file itself is never uploaded anywhere. Only its fingerprint leaves the browser. ● Extract — a large language model (Claude) reads the work and forms its Conceptual Core: Propositions, Constraints, Mechanisms, and Boundaries — a compact, structured statement of what the work actually claims and where those claims stop applying. ● Verify — the payment made to register the work is checked server-side, independent of anything the browser reports about itself. ● Identify — a permanent UPL Registry Code is issued for the work. ● Register — a record is written to a single, central, publicly readable Registry. ● Bridge — a one-file Certificate is generated, and a package is produced that republishes the work as a conversational AI edition — a Custom GPT carrying the same identifier — so the ideas inside the work become something a reader can actually talk to and build on. The interface is portable by design: any publishing company can deploy its own copy of the DEMI front end by changing a single configuration line, writing into the same shared namespace under licence. One Registry, many publisher doors — the model already used for UplitAU's own storefronts, and the same model any partner publisher would adopt.