Put it against a real system
Knowledge-system work turns sources, permissions, provenance, contradiction, and revocation into an operating architecture.
See the practice← Research agenda
02 · Knowledge
Canonical knowledge
Research question
Time from new information arriving to knowledge humans and AI both trust.
Why the problem is difficult
The mechanism and its failure modes.
Retrieval is not the hard part. A result can be easy to find and still be stale, contradicted, unauthorised, or impossible to defend.
The delay lives in custody: who supplied a claim, what supersedes it, who may use it, how it is revoked, and what the audit trail must preserve.
Current state
What Anant is doing now.
This is closest to Anant’s existing document and knowledge work. Tekt is the open engine; Konotree is a separate company built on it. Anant has not yet published a custody specification or a measured canonicalisation benchmark.
Proposed method, output, and evaluation
Proposed method: test provenance, contradiction, permission, and revocation events against a shared record and measure time to an auditable, trusted state. Proposed output: a custody model and repeatable test corpus for human and AI consumers. No benchmark is being claimed yet. Evaluation: test whether each record remains traceable, authorised, current, and revocable under the defined event sequence.
From question to output
Two ways this work becomes useful.
Produce an inspectable result
The result should apply beyond Anant’s commercial scope. Fund the measurement, specification, implementation, and public artifact—not a consulting subsidy.
Discuss funding