Omniscient Design

Intelligence Catalog

Product Roadmap

Advanced automation deferred from MVP. The current release focuses on catalog browsing, client intake, recommendations, and brief export.

MVP (Current)

Complete
  • Load design intelligence JSON dataset
  • Catalog browse with industry, style, personality, complexity filters
  • Entry detail views with SMB adaptation notes
  • Client intake form (industry, personality, goals)
  • Top-3 recommendation engine with scoring
  • Prompt generation for Claude Design, Claude Code, Codex, Cursor
  • JSON/Markdown brief export
  • Local brief catalog (browser storage)

Phase 2 — Data Pipeline

Planned
  • Daily Firecrawl research jobs for Webby, Awwwards, Dribbble, FWA
  • Automated entry classification via LLM with human review queue
  • Admin UI for approving, editing, and publishing new entries
  • Versioned dataset with changelog and rollback
  • Screenshot capture and thumbnail generation pipeline

Phase 3 — Intelligence Layer

Planned
  • Vector RAG over design entries, adaptation notes, and build outcomes
  • Semantic search across catalog (beyond keyword filters)
  • Learn from completed client builds — which references converted best
  • Industry-specific recommendation tuning from Omniscient client history
  • Competitive site analysis integration from omniscient-recon skill

Phase 4 — Platform Integration

Planned
  • Sync saved briefs to Omniscient OS client records
  • One-click dispatch to omniscient-creative and omniscient-conductor workflows
  • Design DNA extractor integration for live reference URLs
  • Vercel preview deploy from generated Cursor/Codex prompts
  • Team collaboration — shared catalogs, comments, approval flows

Phase 5 — Local AI & Automation

Planned
  • Optional local model integration for offline prompt generation
  • Automated wireframe generation from selected entries
  • Component library mapping (shadcn/Tailwind modules per reusable module tag)
  • A/B variant brief generation for client review
  • Post-launch analytics feedback loop into catalog scoring

Extending the Dataset

Add new entries to data/design_intelligence_catalog.json. Each entry follows the schema in src/lib/types.ts. Update lastUpdated when publishing changes. The app hot-reloads in dev; production rebuilds pick up the new dataset automatically.