Connected Compute

Route governed work to customer-owned machines or LumenFlow-managed runners with one policy, lease, evidence, routing, and metering model.

BYOC and managed runners under one governance model#

Connected Compute is the workspace primitive for running governed agent jobs on either BYOC sources you operate or LumenFlow-managed sources. A source can be a laptop, CI runner, Docker host, Kubernetes worker, or the first managed pool registered as lumenflow_managed.

Both paths use the same control-plane contract:

LayerWhat is shared
ClaimPOST /api/v1/runtime/claim assigns eligible runtime work
Leasedispatch_lease records bounded execution and retry state
EvidenceReceipts, approvals, and replay links stay workspace-scoped
RoutingWorkspace policy chooses byoc_only, byoc_first, managed_first, or managed_only
BudgetPer-WU and per-tier limits refuse before silent fallback

What is live#

  • connected-runtime enrollment and health posture
  • the runtime claim path and assignment lifecycle
  • compute source read models for operator dashboards
  • workspace routing policy UI and per-job "assigned to X, why" detail
  • the first managed-worker registration path for lumenflow_managed
  • a runner-side compute-class classification (the default is cpu-small), recorded on each assignment run

What is still planned#

Managed compute billing is landing in slices. The next pieces are compute_usage_events ingestion from managed assignment completion, credit wallet debits, Stripe meter emission, and customer checkout for managed compute credits.

Until those pieces are complete, do not describe managed compute as a fully billable marketplace. Describe it as the governed routing and execution substrate with metering and checkout in progress.

Private-network boundary#

Connected Compute is not a raw tunnel. Private-network access composes with the federation runtime and its admission gates. A BYOC source can execute approved work, but it does not silently become a private-network proxy.