Human approval
Consequential actions can require a human decision before they run.
Relsor’s trust model begins with constrained authority and inspectable execution — human approvals, permission boundaries, tenant-scoped workflows, and evidence you can review.
Principles
Trust begins with constrained authority, human control, and execution you can inspect.
Consequential actions can require a human decision before they run.
Agents act inside explicit permissions — not open-ended authority.
Supported product workflows keep company work inside the authenticated tenant.
Jobs return inspectable artifacts, citations, and status in Mission Control.
Operators and owners see role-aware product surfaces matched to their authority.
External actions depend on connected providers, permissions, and maturity of each path.
Controls
Status labels show how each control applies across supported Relsor workflows.
Approval-required actions wait for a human decision. Rejection blocks execution.
Allowlisted actions and explicit permissions constrain what agents can invoke.
Authenticated tenant claims scope execution and stored work to the organization.
Handoffs pass bounded context and references — not credentials.
Security-relevant mutations write audit events; jobs and artifacts remain inspectable.
Owner, admin, member, and viewer roles shape what each person can see and decide.
Connected providers and granted permissions determine which external actions are available.
Outbound and consequential paths stay on governed tool routes with approval where required.
Delegated work, waiting approvals, and evidence surface in Mission Control.
Multi-step runs can expose handoff references and lineage across agents.
Credit and budget gates can stop spend when limits are configured.
Identity, deployment, and procurement requirements are reviewed with your team during enterprise planning.
Assurance
Organizations with specific security, architecture, privacy, or procurement requirements can review them with Relsor during enterprise planning.
Resources
Explore Relsor’s security model, enterprise adoption approach, and integration maturity.
How governed execution, permissions, approvals, and tenant boundaries work in product.
Controlled rollout, integration planning, and governance for larger organizations.
Honest maturity labels for connected systems — Available through Requires provider approval.
FAQ
Product-focused answers about authority, integrations, evidence, and enterprise review.
Actions run through allowlisted tool paths. When an action requires approval, a human must decide before execution continues. Rejection blocks the action.
A connection makes a provider available under the permissions granted for that connection. It does not automatically authorize every agent or every action.
No. Available actions depend on permissions, assignment, connection state, and the maturity of each integration path.
Supported product workflows use authenticated tenant claims to scope execution and stored work. Cross-tenant access is denied in the product model.
Administrators can inspect jobs, approvals, artifacts, citations where produced, handoff references, and orchestration lineage through Mission Control and related surfaces.
When approval is required, supported workflows pause until a human approves or rejects. Rejected work does not proceed to execution.
Handoffs carry bounded context and references needed for the next step — not credentials or secret material.
Identity, integration scope, data boundaries, logging, approval policy, budget policy, legal, and deployment topics are reviewed during enterprise planning. Contact Sales to start that conversation.
Share the controls and review topics that matter to your organization. We will walk through what is available today and what to plan during adoption.