How it works
Follow a commitment from its original source through evidence, review and authorised action.
ILLUSTRATIVE OBLIGATION GRAPH
FROM PROMISE TO PROOFA promise becomes
something you can act on.
The Obligation Graph links who owes what to whom, by when, and under which terms. It connects those commitments to real-world evidence and the approvals needed to respond.
THE PROMISESupply agreementTerms · owner · due date
THE REALITYDelivery evidenceWhat arrived · when
THE IMPLICATIONConnected exposureTerms matched to outcomes
THE RESPONSEAuthorised actionReviewer · decision · receipt
CORESources, accepted facts and decision history stay connected at every step.
SUPPLIER DELIVERY / ILLUSTRATIVE RECORD
First, establish the promise.
A supplier is due to deliver on 11 September. The commitment stays connected to its contract clause, owner and agreed delivery schedule.
- Commitment
- Deliver by 11 September
- Source
- Supply agreement · delivery schedule
- Control
- Extracted terms require review
A proposal becomes authoritative only through the required approval.
The result: a source-backed response, with the supporting terms, evidence and approval history kept together.
IN PLAIN LANGUAGE
A clearer picture.
Is ARVIONIQS another AI chatbot?
ARVIONIQS brings together Record for company memory and Verify for AI coding-agent oversight. AI proposes; source evidence, controlled fact states and human review determine what becomes authoritative.
Does it replace our existing systems?
The aim is to connect the evidence and obligations held across existing systems. The appropriate integrations and any workflow consolidation depend on your use case and deployment scope.
What happens when sources disagree?
Both accounts remain linked to their sources. The conflict is routed to an authorised reviewer rather than silently resolved by the model.
Can AI act without a person approving every step?
The governance model allows actions within explicitly granted authority. Decisions outside that scope, or those requiring human approval under policy, must be escalated.
How do we explore a first workflow?
Start with a representative workflow using synthetic or approved non-classified data. Agree the reviewers, evidence requirements and measurable pilot goals. Confirm readiness and security boundaries before introducing real company data.
TAKE THE NEXT STEP