License without opening your training set.

Turn unpriced exposure into a per-output cost you can forecast. Yodal's architecture is designed to prove licensing compliance without disclosing training data.

Today
  • Litigation exposure is unbounded and unpriceable.
  • Bilateral deals are lump sums unlinked to actual output.
  • Any audit mechanism exposes training data to the auditor.
  • Regulatory direction points toward per-use accounting.
The architecture
  • Per-output cost turns liability into a forecastable line item.
  • Payment is proportional to measured contribution.
  • Designed to run inside a TEE — training data never leaves the enclave.
  • Audit trail without disclosure: the enclave attests, not reveals.

The integration

Designed for sealed inputs and auditable outputs.

Your inference pipeline Prompt in, generation out — unchanged
↓
Yodal enclave Attribution runs inside a TEE. Sealed inputs, attested outputs. Training data never leaves.
↓
Contribution ledger Measured contributions, license terms, payment flow — all auditable, nothing disclosed.

How it works

01

Deploy the enclave

Yodal's attribution module is designed to run inside a trusted execution environment.

02

Route generations

Each output passes through the enclave. Measured contribution events are logged; training data is never exported.

03

License and attest

When a rightsholder's work likely contributed, the enclave records the event. If the publisher's terms are met, the license executes automatically.

We will never

Export training data

The enclave attests to contribution without revealing what was trained. Your data stays yours.

Set license prices

Publishers set terms. You see the rate before you opt in. Yodal measures and executes — nothing more.

Take a side

No models. No content. No consumer product. We're neutral by construction.

Talk to us

See what per-output licensing looks like for your stack. We'll walk you through the integration architecture and answer technical questions.