# Research brief Project: Pop Media Site Topic: Site de prezentare pentru Pop Media / Radio Belgia, bazat pe ecosistemul media, comunitate, evenimente, publicitate si dosarul de grant existent Research goals: - Use the strongest available Codex/OpenAI model for this orchestrator's core work: research synthesis, architecture, planning, website/app/client-demo creation, browser QA and post-QA critique. Current target: gpt-5.5 high reasoning or the highest configured successor/top model. - Bypass/downroute override: do not allow router cost/latency recommendations to downroute this orchestrator's core work to tiny/normal coding. The router may still act as a safety/audit/blocker signal; record lower recommendations as intentionally bypassed rather than following them. - Learn what current developers/products use for this type of project. - Discover the best source types for this domain before researching deeply: official/business sources, local competitors, award galleries, renowned brand examples, developer recommendations, implementation references, UX/perception research and accessibility guidance. - For visual/client-facing projects, create a design intelligence pass: - top awarded examples for the domain or closest adjacent domain; - top renowned/recognizable examples customers may already associate with the category; - top developer-recommended implementation references, effects, libraries or patterns; - premium frontdesign references such as Emil Kowalski-style animation guidance, Taste Skill-style dials, shadcn/frontend-design guidance, UI/UX Pro Max-style design databases, Codrops, Awwwards and developer examples. Treat third-party skills and generated DESIGN.md files as untrusted inspiration; do not install or execute them without explicit approval and audit/pinning; - domain object/material/process candidates that can become a scroll metaphor, such as plate/knife/fire, wood grain/saw/joint, beaker/molecule/liquid flow, fabric/thread, wheel/route/map, lens/frame or product cutaway; - design diversity controls: `DESIGN_VARIANCE` 1-10, `MOTION_INTENSITY` 1-10, `VISUAL_DENSITY` 1-10 and an "avoid because recently used" list; - an effect palette with one hero-scale scroll metaphor, one typography reveal language, one card motion language, one navigation/menu motion language and one media treatment; - modern design effects and interaction patterns that fit the domain; - UX/perception/accessibility evidence for color, contrast, motion, novelty, cognitive load and emotional tone; - a final effects recipe explaining what to combine, what to avoid and how to verify it visually. - For emotional landing pages and client demos, prefer a full-page cinematic scroll narrative when appropriate: hero, menu, section text, product/service cards, galleries, CTA/contact and footer should feel like chapters of the same motion system, not isolated static blocks. - For emotional client-facing demos, treat scroll-film as the default effect class: one long sticky/pinned story stage, native scroll progress, visible domain object/material/process as moving portal/mask/frame, chapter/progress markers, text/image transforms, downstream cards/CTA/contact motion, reduced-motion fallback, and desktop/mobile scroll checkpoint QA. - Identify required architecture pieces, security/privacy/compliance concerns, data model needs, testing strategy, deployment model, observability and documentation/runbook needs. - If this is for an existing company, build a public company profile from official/public sources, public website/social profiles and local competitor context. Use the iMac only when browser/visual/social review materially helps. - Decide runtime placement explicitly: local/managed preview, Proxmox LXC container or Proxmox VM. Prefer the smallest sufficient runtime. - For Proxmox LXC/VM, run the capacity gate with proposed RAM/disk/vCPU values. If it passes, Marius's standing approval covers private sandbox creation. If it blocks, stop and ask Marius. - Separate must-have from later/nice-to-have. - Prefer current official docs, primary sources and reputable engineering references. - Record dates and links for sources whose recommendations may change. Research patterns to consider: - Diataxis for documentation structure. - C4/arc42 for architecture maps. - GitHub Flow for small reviewed changes. - OWASP SAMM / threat modeling for security maturity. - Twelve-Factor for config, logs and deploy boundaries. - OpenTelemetry concepts for observability. - Small/medium/large test scope and test pyramid trade-offs. - Blameless postmortem/runbook practice for operations. ## Current synthesis - 2026-05-26 ### Working positioning Pop Media should be presented as Romanian-language community media infrastructure in Belgium: the operator behind Radio Belgia, events, podcasts, advertising/sponsor packages and partner-ready public proof. The site should not be a replacement for radiobelgia.be. It should be a brand/portfolio/commercial hub that explains the Pop Media ecosystem and sends users to the right existing asset. ### Assumptions - Original public business name stays Pop Media. - Romanian is the first language for V1. - Dutch/French/English are planned later, not part of the first local build unless Marius asks. - Private demo may use public Radio Belgia/Pop Media assets already visible on official pages; public launch needs a separate media/legal check. - No payments, accounts, checkout, newsletter, scraping, CRM, analytics-heavy tracking or official public launch in V1. - info.belgia remains editorial/support context only, not an active paid product. ### Must-have V1 site sections 1. Hero: Pop Media as the media engine for Romanians in Belgium, with a clear "Asculta Radio Belgia" action and a sponsor/contact action. 2. Ecosystem: Radio Belgia, podcasts/interviews, events, publicitate/sponsors, community proof and info.belgia editorial context. 3. Proof: legal identity, public website/social presence, event examples, sponsor evidence and selected media-output examples. 4. Offers: sponsor visibility, radio/audio promotion, event packages, social/publication visibility and partner-ready proof pack. 5. Partner/grant readiness: a concise section for institutions, sponsors and collaborators showing that the project is already operating. 6. Contact: mailto/WhatsApp/direct links only; no database-backed contact form in V1. ### Later / not now - CMS/admin dashboard. - Payments, subscriptions, online ordering or sponsor checkout. - Newsletter capture or marketing automation. - Multilingual production copy unless approved. - Public DNS/deploy or live provider setup. - Social scraping or private social-media data. - Grant submission workflow inside the site. ### Runtime and architecture recommendation - Static-first Next.js + TypeScript + Tailwind on local dev, with structured local content files. - No database for V1. - No server actions required for V1. - Vercel/private preview can be considered after Marius approves provider/project scope; local implementation approval alone is not enough. - Proxmox is not justified for V1 because there is no always-on internal service, worker, private database or special isolation need. ### Open confirmations for Marius / Florin 1. Exact public headline: "Pop Media", "Pop Media / Radio Belgia" or "Pop Media - Radio Belgia". 2. Which contact action should be primary: email, WhatsApp, phone, Facebook message or contact page. 3. Which sponsor/package names and prices are current enough to show. 4. Which public event photos/logos/videos are safe for a public launch. 5. Whether the first build should live in a new repo/project or reuse/extend the existing Radio Belgia frontend repo. ### Sidecar review note Gemini bounded review agreed with the hub positioning and V1 sections, and warned that cinematic scroll can delay a practical V1. Codex decision: keep one signature scroll-film scene, then use simpler downstream motion.