Identify the workload
Distinguish the application workload from a request for inference.
Modules · Dumpla
Prisman
Connect applications and AI workloads to node compute and inference services.
Prisman in your universe
Prisman serves compute workloads and model inference. Dumpla sells the memberships used in its experience; Prisman accounts for the compute and inference service.
Prisman in practice
A capability needs execution
An application container or model request uses the Prisman service appropriate to its workload.
Use the appropriate service
Access the node or inference service assigned to the activity.
Keep usage in scope
Follow the resources and terms of the service being used.
Prisman
Compute behind the workload
A node supplies execution capacity, a deployment stores its snapshot, and inference serves a different workload.
01
Node Compute
Nodes provide processing cores, RAM, and storage for workloads.
02
Deployments
A deployment is a stored snapshot ready for execution on an appropriate node.
03
Inference
Prisman supplies inference used by Dumpla experiences and API consumers.
04
MCP servers
Hosting an MCP server requires its own Prisman container rather than the Dumpla MCP client alone.
Modules
Build your next experience
Explore Modules and connect the capabilities your activities need.