Programmable IP

Programmable IP turns rights into usable infrastructure.

Programmable IP moves ownership, licensing, attribution, and usage rules from static paperwork into infrastructure for AI-native media and agent commerce.

By Jason Colapietro (Johnny Suede) · Founder & CEO of Suede Labs AI

A contract in a drawer cannot follow the file.

Static IP works as long as humans negotiate every transaction. A license sits in a PDF, a credit lives in liner notes, and a split table exists in a spreadsheet that three different parties have saved in different versions. That structure was already fragile. It becomes inoperable the moment AI agents start discovering, evaluating, and transacting creative work autonomously at machine speed. An agent has no way to read a PDF in a drawer. It queries endpoints, parses structured data, and either finds a machine-readable permission layer or it moves on to an asset that has one. Paper-bound IP does not fail slowly in the AI era. It is simply invisible.

The license, the split, and the permission have to move with the file.

Suede makes ownership executable by attaching rights context directly to the asset at the moment of creation: who made it, what is permitted, who gets paid, and at what split. When the file moves, the terms move with it. When an agent requests a license, the permission layer responds with a machine-readable answer. When a use triggers a royalty, the route is already defined and the settlement can execute without a human intermediary stepping in to broker the terms. This is not a description of a better filing system. It is a different model of what IP is: not a claim asserted after distribution, but a set of executable rules that travel inside the asset and settle automatically on every qualifying transaction.

Programmable IP means the license travels with the asset. The split is in the file. The royalty route is in the contract. The permission is machine-readable. When AI agents start transacting creative work autonomously, the only assets they can legally use are the ones with machine-readable permission layers already built in.

— Jason Colapietro, Founder & CEO, Suede Labs AI