rokcet-dev/DEMO_APPS.md

Rokcet × Buckets business-app model

Rokcet turns a novice's business idea into a polished, usable application. Buckets is the professional operational platform behind it. A Rokcet build should therefore present business outcomes—not infrastructure controls—while remaining a real Buckets-powered app rather than a static mock-up.

Product boundary

• Rokcet.dev is for novices. The builder asks plain-language questions, proposes sensible defaults, generates the app, previews it, and explains only the next decision the user needs to make.
• Buckets.id is for professionals. Advanced configuration, schemas, scopes, provider setup, agent inspection, API keys, Developer MCP, Customer MCP, observability, and operational administration stay in Buckets.
• Rokcet may deep-link to the relevant Buckets screen under an Advanced or Open in Buckets action, but it must not reproduce the Buckets control plane.

Canonical generated-app grammar

Every generated app begins with the same adaptable product grammar:

1. Discover — browsable people, services, experiences, or resources with strong media and search.
2. Chat — a persistent AI-led experience that can answer, recommend, retrieve gated context, and take authoritative actions through Buckets tools.
3. Activity — bookings, requests, purchases, approvals, messages, and clear status history.
4. Library — saved media, unlocked or gated content, documents, favourites, and reusable customer context.
5. Profile — identity, access, Buckets Credits (BC), preferences, consent, and support/handoff.

Labels may adapt to the business, but generated apps should retain this coherent mental model. The mobile experience is primary. On desktop, the third column must have a clear job: it may be a persistent concierge where that helps the customer journey, or a screen-specific helper that disappears inside a focused conversation. It must not become an unrelated global chat competing with the current task. The result may take visual cues from modern chat-first discovery products, but must use original branding and implementation.

Capabilities to demonstrate through outcomes

Use the app's normal customer journey to show the relevant Buckets capabilities:

• managed AI agents and structured business context;
• authoritative Customer MCP actions for availability, booking, updates, purchases, and other runtime operations;
• media libraries and delivery;
• identity, permissioned access, and gated content;
• bookings, schedules, availability, reminders, and channels;
• global Buckets Credits (BC), where relevant;
• persistent customer context and activity history;
• human handoff with a visible status and return path;
• auditable operations and useful failure/recovery states.

Do not expose raw IDs, JSON, schemas, MCP terminology, scope strings, tokens, infrastructure providers, or setup forms in the novice-facing experience.

Seeded showcase apps

Velora

A premium AI-model booking engine based directly on the attached Luxevibe product and visual system. The homepage is deliberately denser than Luxevibe: smaller image-first cards show many more rated models at once, while preserving card carousels, voice samples, availability, ratings, traits, pricing, full model profiles, gated media, reviews, direct booking, and model-specific private chat. Velora also includes scenario templates, a four-step booking engine with add-ons and credit validation, full private chat rooms with rich message interactions, a wallet/VIP flow, booking management, and a custom-model studio.

GoGame

A full cyber-gaming marketplace and Pro Buddy coaching template based directly on the supplied GoGame app. The original game store, detail hubs, DLC/equipment/skins, Pro Buddy roster, verified reviews, multi-step coaching booking, live tactical session room, gifts/score progression, NEXUS-7 concierge, Cyber Vault, wallet/VIP controls, layout modes, and custom Pro Buddy studio are retained. Rokcet preview provides deterministic interactions; connected deployments route managed conversations through Buckets Agents, coaching reservations through Buckets bookings, and live wallet balance through global Buckets Credits.

OKgo

A source-backed luxury travel experience and itinerary engine. The supplied stays catalogue, destination discovery, property and room details, 4K galleries, bespoke experiences, itinerary cart, single-checkout package flow, digital trip passes, cancellation and rescheduling, VIP profile, rewards, currency selection, and audio interactions are retained. Its concierge supports rich property, review, location, gallery, experience, itinerary timeline, budget, weather, dining, poll, and checkout-ready messages. Connected deployments route concierge replies through Buckets managed conversations, trip confirmation/update/cancellation through Buckets bookings, and the visible travel-credit balance through global Buckets Credits.

Fred Rentals

A source-backed AI car-rental discovery and booking product. The supplied warm Bedrock interface, concierge-first landing screen, fleet catalogue, rich product/review/comparison/quote/location/gallery/poll message types, vehicle details, checkout, active rentals, Digital Stone Keys, rental extensions, roadside assistance, currency controls, and membership rewards are retained. Connected deployments route concierge text through Buckets managed conversations, confirmed rentals and extensions through Buckets bookings, and customer identity through Buckets OAuth while preserving the source app's structured UI.

FOOTPI

A source-backed football talent intelligence and AI scouting product. The supplied scouting workspace, role-aware flows for scouts/agents/club owners, multi-variable criteria builder, shortlist, rich player cards, pinned talent profile, membership gating, radar/comparison charts, detailed player profiles, tactical analysis, and acquisition/contract intelligence are retained. Rokcet preview uses deterministic scouting responses against the supplied player dataset; connected deployments route scouting conversations through Buckets managed agents while preserving FOOTPI's structured intelligence interface.

These are five distinct, high-fidelity expressions of Rokcet's reusable Buckets-powered business-app model, each preserved as a dedicated source-backed showcase template rather than flattened into a generic theme.

Preview contract

• A seeded demo must render without a React, ReactDOM, or icon CDN, without a live Buckets dependency, and without waiting for the parent authentication cookie to settle.
• Rokcet submits the selected source files to a stateless preview endpoint. The server validates them, bundles the exact source with the installed React runtime, and returns one self-contained browser document.
• Seeded apps include semantic CSS, so the optional Tailwind browser helper used for arbitrary generated utility classes is not required for their layout.
• Source edits are always compiled directly; there is no static fallback that could conceal a broken user change.
• Preview states include useful loading, retry, timeout, empty, compile-error, and runtime-error treatment rather than a blank iframe.
• Interactive demo state is local and disposable, but actions and UI mirror the contracts a live Buckets integration would use.

Generation rules

• Start from the user's business outcome, not a technology selection screen.
• Prefer a complete vertical slice over many disconnected screens.
• Chat must do useful work; it is not decorative support chat.
• All action results need visible confirmation and a place in Activity.
• Use realistic media, hierarchy, spacing, responsive behavior, and accessible controls.
• Avoid generic dashboards, toy landing pages, and cloned brand layouts.
• Never fabricate live Buckets data. Seeded preview data must be clearly deterministic; connected builds use the actual Buckets authorization and runtime.

Design-source preservation — 2026-08-26

The current seeded catalogue deliberately keeps each supplied source application's own design language rather than applying one Rokcet theme across every demo.

Design principles applied across Velora, GoGame, OKgo, Fred Rentals, and FOOTPI:

• retain the supplied information architecture, visual hierarchy, interaction density, and branded personality;
• keep AI assistance operational and scoped to the current customer task rather than adding unrelated global chat;
• preserve rich discovery → detail → conversation/booking → activity flows;
• represent live availability, bookings, media/access, Buckets Credits, and customer identity as customer outcomes;
• provide deterministic preview behavior without fabricating Buckets traces, then switch the same interface to real Buckets routes in a deployment;
• keep graceful media and integration fallbacks so a temporary external failure never turns Preview into a blank app.

Velora preserves the Luxevibe-derived dark frosted-glass model marketplace; GoGame preserves the supplied cyan/violet cyber cockpit; OKgo preserves the supplied bright, refined OTA-style luxury travel interface; Fred Rentals preserves the supplied warm limestone-and-amber Bedrock rental interface with its concierge-first structure; FOOTPI preserves the supplied clean slate-and-emerald football intelligence workspace with its scouting-desk composition and rich profiles.