Operating Models
Every Workshape Catalyst deployment is a dedicated client instance. The main decision is who operates it.
Operated Dedicated Instance
Section titled “Operated Dedicated Instance”Use this when a team wants the chat outcome but does not want to run infrastructure.
The customer provides:
- the instance brief
- user and role expectations
- agent instructions and workflow examples
- custom tool requirements or tool code
- model provider and region constraints
- retention, deletion, and audit requirements
- branding and chat copy
- customer application integration details
The instance operator owns:
- Docker images and runtime services
- database, object storage, and backups
- reverse proxy and TLS
- migrations and deploy scripts
- health checks and incident response
- production env files and secret handling
This is often the right starting point. The customer still controls the workflow and sensitive-data assumptions, but does not need to become the platform operator.
Self-Operated Instance
Section titled “Self-Operated Instance”Use this when the customer needs direct operational control or wants to run the instance inside its own infrastructure boundary.
The customer owns both layers:
- client assembly code and release config
- tool implementations and integrations
- runtime infrastructure
- backups, monitoring, TLS, secrets, and restore drills
- provider contracts and region controls
The same platform package boundaries apply. Self-operation should not mean forking the platform.
Customer-Hosted Integrations
Section titled “Customer-Hosted Integrations”Some systems remain outside Workshape Catalyst even when the instance is operated for the customer.
Examples:
- a customer API called by a custom code tool
- an internal document system called by a custom code tool
- a customer endpoint that exchanges an existing app session for a short-lived chat session token
These are customer-hosted integrations. They are not part of the Workshape Catalyst runtime, but the instance must document how it authenticates to them, what data it sends, and what failures look like.
Choosing A Path
Section titled “Choosing A Path”Choose operated dedicated instance when speed and low operational burden matter most.
Choose self-operated instance when infrastructure location, network boundaries, internal policy, or direct operational ownership matter most.
In both cases, write the Instance Brief before implementation.