MVP · Northern Virginia / Loudoun County
What can actually scale — and what has to exist first?
This workbench separates observed evidence from scenario assumptions. Unknown required inputs stay unknown instead of being converted into a confident-looking number.
Core question
Given a target AI scale, what architecture is required, what limits deployment, and what can be sustained by the target date?
Scenario
Target
Domain support assumptions
Blank = UNKNOWN. Any number entered here is explicitly classified as ASSUMED, not observed.
Engine output
Capacity state
Final deployable capacity
—
Provisional known minimum
—
Binding constraint
—
Model state
NOT EVALUATED
Unknown required domains
Not evaluated
Architecture gaps
Final capacity is withheld whenever a required domain is unknown.
Primary-source registry
Evidence
Human + agent
Shared live state
A WebMCP-capable agent can inspect this page, change scenario inputs, run the same engine, retrieve evidence, and inspect the dependency graph. Agent changes appear in this interface immediately.
get-scenario-state
set-scenario-target
set-domain-assumption
clear-domain-assumption
evaluate-capacity
get-evidence-record
trace-project-dependencies
calculate-project-critical-path
Activity
Visible actions from the human interface and WebMCP tools.
Verified project dependency example
Firehouse 230 kV Delivery
The PJM/Dominion record verifies a large data-center load request and supporting grid architecture. It does not establish regional available capacity.
Dependency closure
Loading
Critical-path status
Total verified lead time
—
Timing state
—
Unknown lead-time nodes
Loading
Loading dependency evidence.