Summary
Treat the entire comparevi control plane as an autonomous capital deployment fabric.
Why
The system is no longer just a queue runner or a multi-lane coding helper. It now allocates funded execution capacity across live, background, hosted, remote, and shadow lanes. That means we need one governing model that connects:
- lane utilization
- queue continuity
- spend attribution
- invoice turns and funding windows
- consumer-proving evidence
- release-candidate promotion
Governing objective
Maximize validated throughput per funded dollar while keeping consumer-proving health green and eliminating dark spend / dark idle time.
Required properties
- no dark spend
- every live/background/hosted/shadow turn attributable to lane, issue, repo, provider, model, and funding window
- no dark idle
- every non-working lane classified as waiting-hosted, waiting-merge, policy-paused, blocked, prewarm, or operator-steering
- no silent funding drift
- invoice turns and overlapping funding windows must reconcile without double-counting outcomes
- no queue starvation
- merge queue and lane pool should be managed as capital-allocation surfaces, not passive status boards
Primary metrics
- sustained logical lane allocation
- spend attribution coverage
- idle classification coverage
- validated throughput per funded dollar
- merge-queue occupancy / refill latency
- heuristic-vs-invoice calibration drift
Initial success floors
- at least
50% sustained logical lane allocation
100% spend attribution coverage for tracked turns100% idle classification coverage for managed logical lanes- nonzero merge-queue inventory unless the actionable issue queue is genuinely near empty
Notes
This is a program umbrella. Existing issues for lane allocation, spend telemetry, invoice turns, calibration, model selection, queue continuity, and template verification should roll up under this operating model rather than drifting independently.
Summary
Treat the entire comparevi control plane as an autonomous capital deployment fabric.
Why
The system is no longer just a queue runner or a multi-lane coding helper. It now allocates funded execution capacity across live, background, hosted, remote, and shadow lanes. That means we need one governing model that connects:
Governing objective
Maximize validated throughput per funded dollar while keeping consumer-proving health green and eliminating dark spend / dark idle time.
Required properties
Primary metrics
Initial success floors
50%sustained logical lane allocation100%spend attribution coverage for tracked turns100%idle classification coverage for managed logical lanesNotes
This is a program umbrella. Existing issues for lane allocation, spend telemetry, invoice turns, calibration, model selection, queue continuity, and template verification should roll up under this operating model rather than drifting independently.