AI Platform-as-a-Service
Operate governed AI capabilities for multiple users, teams or customers from one controlled platform.
Interactive platform map
Architecture in context
The focused blueprint, its required foundation, and declared recommendations.
AI Platform-as-a-Service
Operate governed AI capabilities for multiple users, teams or customers from one controlled platform.
What this replaces
Organizations want shared AI but do not want every employee or customer managing separate provider accounts, keys and policies.
What your team gains
A multi-user AI service layer with centralized identity, model access, usage controls, shared knowledge and multiple delivery surfaces.
What is inside the blueprint
How it fits the platform
Users/clients -> Authentik -> Open WebUI/API -> LiteLLM -> models + knowledge + tools; Postgres stores tenant/application state.
Authentik, Open WebUI, LiteLLM, PostgreSQL, Traefik, cloudflared
- AICORTEX Core PlatformThe multi-user platform is an assembly of Core-managed services.
- Private AI WorkspaceThe tenant-facing experience is the Private AI Workspace.
- Universal AI GatewayPer-tenant budgets, keys and model policy come from the Universal AI Gateway.
- Zero-Trust Edge & IdentityTenant SSO and private ingress come from the Zero-Trust Edge.
From prerequisites to operation
- Core platform
- User/tenant model
- Commercial usage and support policy
- Provider/compute capacity
- Define groups/tenants
- Provision SSO and workspace access
- Create model keys/budgets
- Segment knowledge/tools
- Publish tenant-facing routes/APIs
- Roles/groups
- Model permissions
- Budgets
- Rate limits
- Knowledge access
- Branding
- API key policy
- Provisioning/deprovisioning
- Cost allocation
- Capacity
- Tenant isolation
- Support and audit
What this unlocks with other layers
Managed AI Business
A provider can sell governed AI access to teams and customers and track usage and cost centrally.