Need-to-know should protect information without stopping people, partners or AI agents from doing the work they are authorised to do. And it should not require more access roles than employees.
01 / THE CHALLENGE
Access projects can become hard to operate when every application, partner and exception produces another permanent role. Users wait for permissions, administrators manage exceptions and teams may find ways around controls to get work done.
Broad VPN access creates the opposite problem: too much network reach. AI agents add new identities and permissions, while cloud-routed access can introduce latency and costs that matter in operational environments.
The question: Can we enforce need-to-know with a manageable access model that supports the work instead of getting in its way?
02 / A DIFFERENT APPROACH
Start with who or what needs to complete which task. Keep roles and entitlements understandable, then use identity, resource, context and policy to handle legitimate variations where supported. Define ownership and exceptions without turning each case into another permanent role.
Appgate uses identity, entitlements and contextual policies to authorise access to specific applications, services and APIs. Its direct-routed architecture avoids mandatory traffic backhaul through a central third-party cloud proxy. This can improve access performance and offer cost advantages in suitable architectures, including latency-sensitive industrial and remote maintenance scenarios.
Actual latency and savings depend on the customer's environment and must be validated.
Kiteworks uses identity, roles, permissions and policies to govern access to and exchange of confidential content across sharing, email, transfers and supported AI-agent requests to Kiteworks-governed data.
Both support the same need-to-know principle at different layers. Each can stand alone; no built-in Appgate and Kiteworks integration is assumed.
03 / WHAT CHANGES
04 / WHEN THIS MATTERS
Employee: Give a team access to the applications and documents needed for its work, with policies for sensitive data without creating a new permanent role for every occasional task.
External partner: Grant time-bound or otherwise appropriately scoped access to the specific system and information needed for an assignment, without opening the surrounding network or creating a maze of exceptions.
AI agent: Authorise a defined machine identity to use the services and content its task requires. Apply explicit policies instead of trusting it to ignore information it could technically reach.
Validate that the resulting workflow is secure, usable and performant, particularly for time-sensitive production systems.
05 / BEHIND THE IDEA
Appgate ZTNA uses identity, entitlements, roles and context-driven policies to restrict access to specific applications, services and workloads, including supported machine identities. Its direct-routed architecture avoids mandatory traffic backhaul through a third-party cloud proxy, which can reduce latency in hybrid and OT scenarios.
Potential cost advantage: direct routing may avoid cloud transit and egress charges and reduce overlapping access infrastructure. Savings depend on architecture, licensing and operations, so they must be assessed rather than assumed.
Kiteworks applies identity, role, permission and policy controls to sensitive file sharing, secure email and managed file transfer, with auditable access and exchange activity. Its Secure MCP Server can enforce existing access rights for supported AI-agent requests involving Kiteworks-governed content.
We help find a lean authorisation model that protects what matters without building more roles than users. Starting with a real workflow, we question which entitlements are truly necessary, which differences a policy can handle, and where controls make everyday work unnecessarily difficult.
The aim is simple: employees, partners and AI agents should be able to do their authorised work securely and efficiently. We bring ideas for practical roles and policies, examine usability, performance and costs, and involve Appgate or Kiteworks specialists where their capabilities fit. We do not begin with a large IAM redesign by default.
ONE OF MANY QUESTIONS
Can we enforce need-to-know without creating a new role for every employee, partner and AI agent?
Security that blocks legitimate work creates operational friction and encourages workarounds.
Start with a small, understandable entitlement model and add context-based policies where useful. Use Appgate to limit resource access and Kiteworks to govern sensitive data, then test the actual workflow, latency and administration effort.
Bring one employee, partner or AI-agent access workflow. In 20 minutes, let's explore how to enforce need-to-know without multiplying roles or slowing down legitimate work.
EXPLORE A LEANER ACCESS MODEL ↗