stage: brand
product: prism-site
status: APPROVED — Glass Box selected at P6
skills_invoked:
- creative-brief
- marketing-strategy
owner: גל (creative cell), directions by מאיה (CD) input via שרה/דן
date: 2026-07-16
P6 decision, 2026-07-16: Glass Box selected. Atlas and Founder's Ledger remain rejected
alternatives for the taste record; do not blend them into the selected direction. Receipt:
decisions/p6-20260716.yaml.Directions only. No mockups, no DesignSync (dashboard.html/mobile.html/tokens.html), no
Cloudflare deploy until Ben picks one in P6. Each direction below maps to one of the three
positioning angles in research/research.md §5, and each deliberately diverges from every
existing PRISM product aesthetic (ben-portfolio = Astro minimal, 99–100 Lighthouse, neutral
type-driven; plakton = cork/organic/warm kids' world; aperture = Editorial Mediterranean warm
terracotta/blue). No shared aesthetic across the portfolio — the engine is shared, not the
style (memory eval rule).
Name feel: technical, unadorned — PRISM sans decoration, or a plain wordmark with a live
status indicator next to it. Tone: matter-of-fact, slightly terminal, zero marketing gloss —
copy reads like commit messages and decision-ledger entries, not sales copy. Visual world:
near-black background, monospace type for structural elements (nav, metadata, timestamps),
one accent color reserved only for "live" states (a link that resolves, a gate that's open). The
homepage literally surfaces real artifacts — a scrolling/paginated feed of real decision-ledger
entries and P6 approvals, git-diff-style visual motifs for "before/after" governance moments.
This is the anti-10Web move: instead of hiding the machinery behind a polished SaaS UI, the
machinery is the content — nothing about it should look like a landing-page template.
Name feel: PRISM as a cartographer's mark — a small glyph that reads like a map legend
symbol rather than a tech logo. Tone: editorial-explorer, confident but unhurried — the copy
frames each live product as a "territory" the factory has claimed and is actively running, not a
case study that's finished and dead. Visual world: warm paper/ink palette (aged cream, deep
navy or ink-black linework, one warm signal color for "currently live") with a literal map/atlas
motif — the portfolio section renders as a chart or index of claimed territories (ben-portfolio,
foodflow-home, etc.) each with coordinates-style metadata (status, last shipped, live link).
Typography leans serif-editorial for body copy against a technical mono for the data/status
layer — the tension between "old cartography" and "live telemetry" is the whole visual idea, and
it reads nothing like plakton's organic-cork warmth or aperture's Mediterranean palette.
Name feel: personal, first-person — treats the site like Ben's own public workbook rather
than a company brochure; PRISM appears as the engine credit, not the headline. Tone: plain,
confiding, un-hyped — closer to Pieter Levels/Marc Lou build-in-public journals than to a
startup homepage; short declarative sentences, dates on everything, no superlatives. **Visual
world:** a graph-paper/notebook surface (off-white with a faint grid), handwritten-style
annotations layered over clean set type for structure, a single ink-blue accent — the page
reads like an open notebook with sticky real numbers and real links pinned into it rather than
designed hero sections. No gradients, no glossy cards, no stock "AI" imagery — deliberately the
least "tech company" of the three, closest to a public engineering diary, and the furthest from
any prior PRISM product's visual language.
DesignSync).
§"Who it serves") is answered — Glass Box leans hardest toward clients/partners, Atlas toward
partners/talent, Founder's Ledger toward Ben's personal-face audience.