Who is allowed to act?
Roles, approval limits, signatory mandates and segregation of duties become part of the workflow.
Jarvis by OBX · Early beta
Give AI the rules of the organisation.
Jarvis is the governed execution layer OBX is building for work where authority, policy, approval and audit cannot be left to chance.
Jarvis / Early betaControl for AI-assisted work
Now in early beta
We are inviting a small group of organisations to test Jarvis on real approval, transformation and operating workflows.
Join the early betaThe control gap
AI can do the work. It cannot assume the authority.
Organisations still need to apply mandates, policies, budgets, approvals and accountability. Most AI tools do not know those rules.
Which rule applies?
Jarvis checks the relevant policy, required evidence and escalation path before work moves.
Can you prove what happened?
Leaders, risk teams and auditors get a reconstructable trail of actions, decisions and approvals.
The Jarvis mechanism
Control travels with the work.
Jarvis applies the organisation’s operating rules before, during and after an AI-assisted action.
Encode
Roles, mandates, policies, workflows and escalation paths become usable operating context.
Control
AI-assisted work is checked against authority, risk, workflow state and human approval requirements.
Record
Decisions, routing, approvals, exceptions, policy checks and agent actions remain traceable.
Improve
Leaders can see where work slows down, where risk appears and which controls need attention.
Early beta focus
Start with one workflow where delay or weak evidence is costly.
Approval workflows
Procurement, finance and compliance work with budget, mandate, supplier, policy and audit requirements.
Early beta use caseControl tower
Decision logs, dependencies, risks, benefits, actions and executive visibility across a transformation portfolio.
Early beta use caseBusiness control centre
Governance, cash and collections visibility, approvals, staff actions and funding-readiness routines.
Early beta use caseEarly beta
Bring us one workflow that needs better control.
We will configure it with your team, test it in practice and use what we learn to improve Jarvis.