FunArts

Inspiration

FunArts began as a spark of inspiration.

As an artist with no technical background, I had endless creative ideas but no clear way to turn them into a working product. I originally wanted to build a fairer platform where artists and meaningful artworks could be discovered beyond the limits of traditional institutions and entertainment-driven social media.

GPT became my long-term thinking partner. Whenever a new idea appeared, it helped me analyse, question, and organise it, gradually transforming scattered moments of inspiration into a coherent product vision. I brought intuition, aesthetics, and imagination; GPT’s synthesis then inspired new directions in return.

FunArts was born through this continuous creative loop between human imagination and AI-supported reflection.

What it does

FunArts is a personalised art discovery platform that helps people explore art based on their individual taste, emotions, interests, and visual preferences.

Users begin with an interactive art personality test that creates an initial taste profile. They then receive personalised artwork recommendations and can explore real works, artists, museum collections, institutions, and locations.

FunArts aims to make high-quality human-created art easier to discover, understand, and connect with—without relying on AI-generated artworks.

How we built it

Codex and GPT accompanied FunArts throughout its entire journey, from an early idea to a working MVP now undergoing internal testing.

With Codex, I was able to:

• Build and connect the backend and Supabase database
• Integrate real museum collections through public APIs
• Design the artwork, artist, institution, tag, and user-preference data structures
• Develop a custom tagging and recommendation system
• Build the art personality test and personalised discovery flow
• Create internal content-management and artwork-tagging tools
• Improve the mobile experience and rapidly fix bugs through iteration

I provided the product vision, creative direction, user experience, and recommendation logic. Codex helped translate those ideas into functioning technical systems.

Challenges we ran into

The greatest challenge was building a technically complex product without a traditional programming background.

I had to understand databases, APIs, authentication, data relationships, recommendation logic, and security while simultaneously designing the product experience. Another challenge was translating something as intuitive and subjective as artistic taste into a structured system without oversimplifying it.

Working with real museum data also introduced inconsistent metadata, missing information, and complex classification problems. Each challenge required experimentation, discussion, and repeated iteration with GPT and Codex.

Accomplishments that we're proud of

The accomplishment I am most proud of is turning an idea that once existed only in my imagination into a real, testable product.

FunArts now includes:

• A working personalised onboarding and art personality test
• A recommendation system based on visual preferences and artwork tags
• A database containing real artworks, artists, and institutions
• Museum API integration
• Personalised artwork discovery and exploration
• Internal tagging and content-management tools
• A mobile-optimised MVP ready for testing and further iteration

For a solo artist with no previous technical background, building this complete system would not have been possible without Codex and GPT.

What we learned

I learned that AI does not have to replace human creativity. It can expand who is able to build, experiment, and bring creative ideas into the world.

GPT helped me transform intuition into structured thinking, while Codex enabled me to transform that thinking into a working product. Their outputs also challenged and inspired me, creating a continuous feedback loop between imagination, reflection, and implementation.

I also learned that artistic taste can be translated into useful product logic without reducing it to a simple genre or style. Taste is multidimensional, emotional, and constantly evolving—and the product must reflect that complexity.

What's next for FunArts

The next step is to improve the accuracy and transparency of the recommendation system, including a “Why this was recommended to you” feature.

I also plan to:

• Expand the database with more museum and cultural collections
• Refine the tagging system and user taste profiles
• Improve artwork search by mood, colour, and context
• Complete the journey from discovery to collecting or purchasing
• Work with artists and cultural institutions to distribute human-created art to new audiences
• Continue testing the product with real users and improve it based on their feedback

GPT and Codex are not merely tools used to build FunArts. They have been its most important creative partners.

Their support enabled me to independently create a product dedicated to celebrating purely human creativity. FunArts demonstrates that AI can become creativity’s most powerful collaborator—giving an artist without a technical background the ability to turn imagination into reality.

Built With

• art
• chatgpt
• codex
• creativity