Home/Server-authoritative control before models run
Platform architecture

Server-authoritative control before models run

CODE S provides a server-authoritative control layer for approved AI workflows. Workspace, role, data class, sovereignty mode, model route and policies determine whether processing is permitted.

Request a pilot
Request control

A server-side policy decision before a model route is selected

The browser can neither set nor bypass approvals. Only an active, available route permitted for the workspace can be selected.

  • Workspace membership and role
  • Data class and sovereignty mode
  • Active approved model route
  • Limited agent permissions and tools
  • Redacted policy and security events
Example policy decisionExample
  1. Workspace verifiedWork order belongs to an approved workspacePassed
  2. Context evaluatedRole · data class · sovereignty modeLimited
  3. Policy decision madeModel route · tools · external processingRequirements met
  4. Route approvedOnly an active and permitted model routeSelected
Processing within the approved scopeRedacted policy event
Server-side

Example policy decision. The actual model route and permitted data processing depend on workspace policy, provider review and approval.

Control core

One shared server-side control layer

The existing core makes policy decisions for approved workflows. Providers, connectors and target systems are reviewed and activated separately.

Approved Model Routes

Local, European and separately reviewed external routes are managed centrally. Providers and models are activated only after technical, data protection and contractual approval.

Route management

Policy-based Routing

Model and tool selection occurs only within routes that are active, available and permitted for the workspace.

Decision logic

Data Classes and Sovereignty Policies

Public, internal, confidential and restricted. External models, web access, connectors and shell tools are limited by class and operating mode.

Data control

Agent Permissions and Work Context

Roles, knowledge sources, tools and local workspaces are limited server-side.

Permissions
Decision flow

From work order to permitted model route

This flow represents the verifiable control core. Specific providers, tools and integrations require separate approval.

  1. Identify

    Workspace and role

  2. Classify

    Data and risk

  3. Evaluate

    Policy and sovereignty mode

  4. Route

    Active model route

  5. Record

    Redacted policy event

© CODE S. Platform for sovereign AI.

Package prices are monthly AEGIS platform fees. Costs for models, cloud services, infrastructure, third-party services and, where applicable, implementation are charged separately.