Find the data. Fund the people who made it.

fiund is a licensing marketplace for real-world audio and video. AI teams find the data they need; the people who own it get paid, and keep their rights.

The name sits between find and fund, and that's the whole business in one word. AI teams find the data. The people who own it get funded.

Why we exist

Public data is exhausted and contaminated, and the deals that matter fall apart over rights, not quality. We built around that problem first: every asset carries a signed licence, explicit AI-training rights, and consent where people are identifiable — sourced from owners, never scraped.

The marketplace model

fiund is an AI training data marketplace built on licensing rather than acquisition. Two sides meet here. Buyers — AI labs, product teams, research groups — bring briefs: the modality, languages, conditions, and licence terms their use requires. Owners — podcasters, studios, mocap stages, contact centres, archives — bring material they've spent years making.

Between them sits the licensing layer. Rights are checked before buyer match, consent is papered where anyone is identifiable, and fiund handles buyer negotiations and licensing under one standing contributor agreement. Owners keep ownership; the same archive can earn again under a new licence. Rates are set per deal, against the actual material — never off a headline rate card.

Done properly, AI training data licensing is slower than scraping. That's the point. The speed of a scrape is exactly what a buyer's counsel can't defend and an owner never agreed to. The marketplace exists to make the defensible route the convenient one.

What we hold ourselves to

• Owners keep ownership. License, don't sell. Nothing here transfers the archive.
• Consent is explicit. Training rights in writing; voice and likeness consent papered separately wherever a person is identifiable.
• Provenance is documented. Every asset travels with its record — source, licence, consent, manifest.
• Deals are documented. Contributors see what was licensed, the buyer identity or category where confidentiality applies, the scope, and their payment.

The mission

A working market where the people who made the data are in the deal — named, consenting, and paid. The next generation of models will be trained on material someone chose to license. We're building the place where that choice gets made fairly, and where choosing it is easier than taking.