R2
The Studio — tokens in, diagrams out
The operator tool. Punch in client brand tokens (name, accent, panel base, glyph, font) and a rough diagram brief (archetype, title, items), and a 2×2 grid of variants renders live as you type. v1 ships with the two archetypes we proved (Sources → Hub, Partners → Hub) × two styles, plus an Okta control for taste-checking.
Action: Open the Studio. Does the form shape work? Are the right knobs exposed? Anything missing for it to be useful day-one? Try a different client brand (e.g. accent #0F766E, glyph "NW") and watch the variants update.
v1 scope: structured form, not prose. SVG export is v2 (requires the archetypes refactored to pure SVG — separate session). LLM-driven "describe it in plain English" mode layers on after the form UX is dialed.
R1
First Okta-style proof + Northwind rebrand demo
Two of your simpler Okta references recreated as token-driven HTML — "Any source of truth" (cream panel) and "Partner 1/2/3 → Okta" (blue gradient) — then re-rendered in a different client's brand (Northwind, deep emerald). The diagram markup is identical across brands; only the token file is swapped. That swap is the engine.
Action: Open the gallery. Does the fidelity hit your Okta references? Does the Northwind rebrand feel equally premium? If anything's off, name the diagram + the issue and I'll iterate.
Honest note: headless screenshot was being killed in my environment, so I couldn't auto-gate the visuals against your references the way I usually would. Your eyes are the verification step here — flag anything off and I'll fix it before the rest of the proof set goes through.