Protocol · Braid + Cortex
Lineage, not just output.
Braid is the provenance record. Cortex is the intelligence layer built from accepted work. Together they are what turns a one-off computation into infrastructure.
Braid
Three strands, one record
Every accepted result is intended to carry the job strand, the work strand and the validation strand woven into a single inspectable lineage.
- 01
Audit a result
Trace an accepted output back to its specification, worker evidence and validation record.
- 02
Attribute behaviour
See how a pseudonymous participant has performed across many jobs over time.
- 03
Reproduce later
Recorded inputs, hashes and versions make re-execution possible after the fact.
- 04
Score over time
When reality resolves a question, the original recorded claim is still there to score.
Cortex
Accepted work becomes context
Cortex assembles validated results into permissioned, traceable context that authorized applications can consume — so the network gets more useful as it does more work.
- 01
Structured accepted intelligence
Only work that survived validation is eligible for absorption.
- 02
Permissioned
Access is governed by job policy and application permissions, not open by default.
- 03
Traceable
Context carries its lineage; a consumer can see where a claim came from.
- 04
Reusable
Verified work should not need to be paid for and performed again from scratch.
What Cortex is not
- ·A centralized super-intelligence.
- ·A single model that knows everything.
- ·An unrestricted pool of everyone's private data.
- ·A guarantee that absorbed context is correct.
- ·Operational on a public network today.
Next