Deployment
Choose the environment around the code.
Run on your own OpenShift environment in a private data centre or a customer-controlled cloud. A fully disconnected setup is an option where the workload requires it.
Learn more
ATHISE / SOVEREIGNTY
Your cluster.
Your rules.
Run ATHISE in your own single-tenant OpenShift cluster — in a public cloud, private data centre or fully air-gapped environment. Keep source code, data and model calls inside the same controlled boundary.
DESIGNED FOR CONTROL
Sovereignty, in practice.
Learn more
Models
Define which models a project can use.
OpenShift AI serves local open-weight models on your GPUs and provides the MCP layer. Isolated execution environments keep agent work inside the cluster; GPU capacity and concurrency inform sizing.
Learn more
People
Keep responsibility visible.
Engineers review outputs and decide whether a stage is ready to advance. Human oversight is part of the workflow, not an afterthought.
Learn more
ARCHITECTURE
The cluster boundary is the product boundary.
ATHISE provides the pipeline, approval gates and user interface. OpenShift AI provides model serving and the MCP layer. The deployment architecture brings these parts together inside your environment.
01
ATHISE
Pipelines · human gates · workspace
02
OpenShift AI
MCP servers · model serving on your GPUs
03
Approved registries
Packages and images from approved sources only
04
Code & artifacts
Source code · container images · Helm charts
Inside this boundary: your code, your data and every model call.
PRODUCT SCOPE
Clear about controls. Clear about scope.
Keep execution, dependencies and engineering decisions accountable. Define the deployment boundary, permitted sources and approval responsibilities for your workflow.
Human decisions
Approve, adjust or reject at defined gates. Agent reviewers contribute security, requirements and quality assessments; architecture approval and final sign-off stay with your engineers.
Approved dependencies
Packages and container images resolve only from approved registries. Source code stays in your repository; built images and Helm charts remain inside the deployment boundary.
Local execution
Pipeline, human gates, user interface, model serving and MCP run in your OpenShift environment. Single-tenant deployment supports public cloud, private data centres and fully air-gapped operation.
Traceable decisions
The product design connects agent actions, reviewer verdicts and human approvals into an audit trail. Confirm event coverage, audit-report export and additional governance capabilities for the release used in your deployment.
FAQ
A few things worth knowing.
Does source code leave our environment?
In a fully local deployment, models run on your infrastructure and the workflow stays inside your cluster. Fully air-gapped operation is supported. The actual data boundary is verified as part of the deployment design.
Is compliance automatically included?
No. Sovereign deployment and human gates provide architectural controls. Security review, legal assessment and any regulatory approval remain specific to your organization and use case.
Do we need infrastructure for a demo?
No infrastructure is required on your side for an initial product demonstration. A customer pilot is scoped separately around your environment and the workflow you want to evaluate.
LET’S MAKE IT CONCRETE
See your next workflow
in ATHISE.
Explore a complete pipeline in a product demonstration. Talk through your
requirements with the people building the platform.
- Product enquiries
- sales@cloudwerkstatt.com
- General enquiries
- office@cloudwerkstatt.com
- Product support
- support@cloudwerkstatt.com