ArtCo

Your personal art concierge

Comment

ArtCo: Your Personal Art Concierge

About the challenge

Agents are getting smarter, but they're culturally blind. They can reason, plan, and execute — but ask them what to recommend, what to build, or what people actually want, and they're guessing. This hackathon is about fixing that.

The problem

Navigating the art market and discovering art that genuinely matches your taste is overwhelming, especially for newcomers. Existing search engines and platforms rely on keyword matches or generic categorization, failing to map deeper cultural preferences and visual connections. Furthermore, non-experts often struggle to bridge the gap between inspiration and acquisition, as art platforms rarely unite cultural research with clear, direct purchasing workflows.

The Solution

ArtCo is a user-centric, art concierge that combines broader cultural taste signals with artwork research across institutional sources (commercial sources in a future iteration). Implemented with a native Python orchestration layer, ArtCo leverages Qloo’s cultural intelligence API to map personal tastes and Google Gemini to interpret visual context, delivering deterministic artwork scoring, curated recommendation cards, educational guides, and streamlined commercial acquisition.

Key Features

Four Concierge Options

Explore art: Discover unexpected directions; optionally enable Use my saved taste to build on saved artworks, profile notes, likes/avoids, and recent ideas using commercial inventory explicitly available to acquire.

Find art: Search for a specific artist, artwork, style, subject, or medium across institutional and commercial collections.

Curate a collection: Assemble a coherent set of commercial works explicitly available to acquire. Learn about art: Generate a comprehensive, educational guide from a written topic (text-only guide featuring an overview, key ideas, looking prompts, and vocabulary).

Sequential & Visual Intelligence

  • Supports uploading up to 5 inspiration images (JPEG, PNG, WebP) compressed client-side. Sequentially executes a written search brief, followed by Gemini vision-derived visual term interpretation, and a dedicated visual search before ranking.
  • Incorporates Qloo /search and /v2/insights to resolve positive cultural references without injecting generic "things to avoid" or unrelated cross-domain entities.

Robust Infrastructure & Diagnostics

  • Real-time performance tracking.
  • Same-origin proxy recovery for museum image URLs, thumbnail error retries, and explicit HTTPS upgrading.

Design (App and UX)

  • Editorial Presentation: Built with Inter typography, supporting light/dark themes, responsive layout, keyboard accessibility, and a gallery-style UI rather than a standard chatbot interface. Structured Interface: Features distinct workspace routes (/, /concierge, /results, /saved, /taste) with /taste seamlessly managed inside the persistent user Dashboard.

  • Privacy-First Local Persistence: Client-side IndexedDB stores user taste references, avoid rules, saved works, and recent requests locally in browser storage without requiring an initial account creation step.

  • Request-Only Memory Handling: Visual inspiration files are held in page memory via one-time sessionStorage handoffs during request formulation and stripped immediately after consumption.

Tech Stack

  • Backend: FastAPI, Python, Vercel
  • Frontend: jinja2, css, javascript
  • Core AI/Data:
    • Qloo API: For all cultural taste and persona data.
    • Google Gemini API (gemini 3.6 flash): Powers the reasoning

Unfair advantage

ArtCo bridges the cultural gap that traditional AI agents face by embedding Qloo’s taste graph directly into its deterministic ranking model. Instead of relying purely on text descriptions, ArtCo combines multi-modal visual intelligence (Google Gemini) with real-world inventory validation, deterministic budget/acquisition constraint enforcement, and automatic image proxy recovery to ensure every returned recommendation is both visually relevant and commercially reliable.

Vision

ArtCo makes art accessible and enjoyable, especially for those who are new to the art scene. By combining what users already love with the reasoning of modern LLMs and cultural intelligence, ArtCo can be deployed as a stand-alone SaaS platform or integrated into existing gallery and marketplace websites to create an entirely new paradigm for exploring and acquiring art.

Next Steps

  • Incorporate art studio data into the app with a focus on North America (US and Canada).
  • Add expanded purchase features in the concierge using PayPal alongside Stripe, prioritizing artwork that is actively available to acquire.
    • Integrated Commercial Acquisition & Payments:
    • Direct checkout and seller redirection via Stripe Checkout (destination charges mapping artwork IDs to prices and seller connected accounts).
    • Dedicated commercial filters ensuring purchasable inventory sources (Artsy, Artlogic, Collect24)
    • Display integrated purchase and order history directly within the user dashboard. explicitly declare acquisition availability.
  • Deeper database integration to improve recommendation system

Built With

Share this project:

Updates

Submission history