The umbrella product above veta / egan / ultra. Manages a customer's value-hypothesis portfolio
across the VE Operating Loop (Discover → Validate → Pilot → Scale → Re-prime). New product, designed
from scratch 2026-05-26; doubles as the SPINE of the Anthropic CS Showcase microsite.
See CONTEXT.md for project framing; release plan at ~/development/ventures/anthropic-showcase/release-plan.md.
Full visual reference: colors, type, spacing, icons, buttons, forms, placeholders, atoms.
Page skeleton — nav, sections, footer, mobile. quanta adapts with sidebar+main shell.
More wireframe patterns for building layouts beyond the deck.
Sibling consultation. quanta consumes veta's financial-modeling layer; same design idioms.
Home — portfolio state today: phases in flight, gates due, validated learnings since last visit, opportunity portfolio sorted by NPV × confidence, persona switcher. Microsite entry view. Identity-defining page.
Full portfolio: opportunities × phase × score × cohort. The three discovery lenses (financial run-cost baseline / capability map / VSM) visible. The "money page" for engagement leads. Showcase: Explore journey-phase spine.
Single opportunity in detail — value hypothesis card, pre-committed pivot/persevere thresholds, evidence state, 4-doc PLAN sections, engagement constitution. Includes Engagement Diagnostic Framework view (held-out eval + tier sweep + 6-bucket tree). Showcase: Validate journey-phase spine.
Engine Dashboard (Risk/Cost/Service KPIs, stage distribution) + value-realization tracking (modeled → measured). Includes "next defense prime engagement preview" — the methodology compounding tile. Showcase: Scale journey-phase spine.
Catalog of harvested patterns — Claude Skills, MCP servers, pattern playbooks, engagement templates. Reuse metrics per asset. R1 ships with 1 lead worked-example tile (Agent-notice processing, BNY Mellon-pattern).
The Engagement Diagnostic Framework view — held-out eval set + prompt-complexity tier sweep + marginal-curve visualization + 6-bucket decision tree. Embedded in Validate phase; surfaces in Pilot phase too. Lives inside the Validate wireframe in R1; standalone in R2.
Cross-cutting affordance — any recommendation traces to Concept → IdeaFramework → Book/Author + customer-evidence + financial frame. The silver-thread mechanic, surfaced from tomo's INFORMS edges. Triggers from any phase.
Per operator GF #4: recommend a harness framework, test efficacy, tune toward optimal. Six canonical layers (tool design / state mgmt / retry-verify / action gating / failure-mode taxonomy / observability). R1 frames it in essay; R2 wireframes the interactive HRA configuration panel inside Diagnostic.
Per operator GF #5: 8-pattern catalog (agent-notices / insurance claims / contracts / clinical / AP-AR / FOIA / M&A / regulatory). R1 publishes the structure + agent-notice lead tile (BNY-pattern); R2 fleshes the other 7 with worked-example tiles + decision trees.
Per operator 2026-05-26: ultra hosts AI risks alongside traditional compliance/security/availability. R1 ships the two-swimlane layout + 5-7 representative classes in the ultra wireframe; R2 builds per-class drill-downs with detection-primitive deep-dives + interpretability outputs.