Albin

Teaching AI to see art the way a critic does.

Industry: Arts, E-Commerce | Stack: Next.js, GPT-4 Vision, Supabase, Stripe, Vercel | Status: Live | Visit Albin

what Albin does#

Albin is an AI art analysis platform that uses GPT-4 Vision to evaluate artwork across composition, color theory, technique, style, and historical context. Artists upload or send any image. The AI returns structured analysis: what the piece does well, where it sits historically, and how it compares to established work in the same tradition.

The platform serves artists refining their portfolios, curators planning exhibitions, teachers building art curriculum, and students learning to see critically. It runs as a SaaS product with tiered subscriptions, from a free tier with 300 monthly queries up to a premium tier supporting 50,000 queries and 200 in-depth analyses per month.

the problem with AI in visual arts#

Most AI tools treat images as classification problems. Label the subject, tag the style, move on. That works fine for stock photo libraries. It doesn't work for an artist who wants to understand why a composition works, or a curator who needs to articulate the relationship between two pieces in an exhibition catalog.

Our client had looked at existing tools and found two camps: surface-level consumer apps (style transfer, image generation) and deeply technical pipelines that required an ML engineer to configure. Neither was usable by a curator sitting down to write catalog copy. The ask was for something in between: an accessible tool that could think about art with real analytical depth.

what we built#

GPT-4 Vision integration for compositional analysis#

Getting GPT-4 Vision to produce useful art criticism took more work than plugging into the multimodal API. Raw API calls return generic descriptions. We needed structured, art-literate output, which required substantial prompt engineering and a layered analysis architecture. The pipeline evaluates artwork across compositional structure, color harmony, brushwork and texture, emotional register, historical context, and technical execution.

Each dimension returns structured output, not a wall of text. The artist gets specific, actionable feedback organized by category. The curator gets language usable in catalog copy. The student gets a framework for understanding what they are actually looking at.

browser extension for contextual analysis#

Albin includes a browser extension that lets users right-click any image on the web and send it directly for analysis. A curator browsing an online gallery can evaluate a piece without switching tabs. A student researching a movement can analyze examples while reading about them. The analysis happens in context rather than in isolation, which changes how you use it.

SaaS architecture with tiered access#

The platform runs on Next.js with Supabase handling authentication and data persistence, Stripe managing subscription billing, and Vercel handling deployment. We built four tiers (Free, Plus at $7.99/month, Pro at $19.99/month, Premium at $39.99/month) based on usage patterns from early testing. Query and analysis limits scale with the tier. All paid plans include a 3-day free trial, prompt library access, and conversation sharing. Higher tiers add increased analysis depth, ad-free experience, and priority support.

prompt library and chat modes#

Beyond single-image analysis, Albin includes a curated prompt library and multiple chat modes. Artists can run comparative analyses between their work and reference pieces, explore stylistic evolution across a body of work, or use guided conversation modes to build exhibition narratives. The chat interface was designed around art discourse rather than adapted from a general chatbot.

key capabilities#

• Compositional analysis: structure, balance, color theory, and technique evaluated across multiple dimensions using GPT-4 Vision
• Historical context: placement within stylistic movements, identification of influences, comparable artists, and period characteristics
• Portfolio tools: structured feedback with trackable analysis history for artists iterating on a body of work
• Exhibition planning: curatorial support for articulating relationships between pieces and building exhibition narratives
• Browser extension: right-click analysis of any image on the web, directly in context
• Tiered access: 300 free queries per month for individual artists, up to 50,000 for institutional users

results#

Albin is live at albin.ai. Art professionals who tested it during development used it differently than we expected: less for formal critique, more for generating language they could adapt for catalog copy and grant applications. That was the validation. The tool was useful for real work, not just interesting as a demo.

Building AI for a specialized domain? See our Custom AI Solutions or book a free Automation Audit.