Wonderline

Inspiration

Wonderline was inspired by the time I spent learning and making art with my son.

While looking for activities we could do together, I noticed that most coloring pages were mass-produced and nearly identical. They asked every child to color the same characters, inside the same lines, toward the same finished result. They were convenient, but they rarely responded to a child's own ideas, interests, or imagination.

I began wondering: What if the starting point were not a template created for thousands of children, but a drawing created by one child?

That question became Wonderline—a personalized art-learning experience that begins with what a child has already imagined. Instead of replacing their creativity with AI-generated artwork, Wonderline uses AI to notice their lines, shapes, colors, and characters, then opens new directions for exploration.

The goal is simple: make high-quality, personalized art education more accessible while ensuring that the child remains the artist.

What it does

Wonderline transforms a child's original drawing into three personalized, printable art activities.

A parent or child uploads a drawing, and Wonderline:

1. Observes its subject and visible qualities without judging or grading it.
2. Identifies artistic possibilities in its lines, shapes, colors, patterns, and symbols.
3. Connects it with three different artistic languages inspired by artists such as Paul Klee, Henri Matisse, Yayoi Kusama, Joan Miró, Wassily Kandinsky, and Jean-Michel Basquiat.
4. Creates three printable coloring and activity pages based on the child's own artwork.
5. Leaves generous open space for the child to continue drawing, experimenting, and telling their own story.

Each result is an invitation rather than a finished answer. Wonderline does not ask children to copy a famous artist, nor does it “improve” their drawing for them. It introduces new ways of seeing and making while preserving the recognizable marks and ideas of the original artwork.

How we built it

Wonderline is a full-stack web experience built with Next.js, React, TypeScript, and vinext.

The creative pipeline uses multiple specialized AI stages:

• An Artwork Interpreter analyzes the uploaded drawing and returns a structured description of its subject, visible traits, and possible areas to explore.
• An Art Language Matcher selects three distinct creative directions from an extensible catalog of artistic languages.
• A Creative Activity Planner turns each direction into an age-friendly prompt, story seed, and set of gentle creative hints.
• An Image Composer uses the original drawing as visual input to generate a personalized, printable activity page.

We used the OpenAI Responses API with vision and Structured Outputs to make each stage predictable and composable. We then used GPT Image to create portrait-format printable pages while preserving recognizable handmade elements from the child's drawing.

The interface guides families through a simple flow: upload a drawing, let Wonderline discover new creative paths, preview the three results, and download one or all of them. A private gallery is stored locally in the browser, so no account is required for the prototype.

Challenges we ran into

The hardest challenge was finding the right role for AI.

It is easy for an image model to produce something polished, complete, and visually impressive. But a finished AI illustration can unintentionally take ownership away from the child. Wonderline needed to do the opposite: provide just enough inspiration to encourage the next creative step.

We spent time refining the prompts so that the generated pages would:

• Preserve the child's recognizable ideas and handmade qualities.
• Avoid correcting, grading, or beautifying the original drawing.
• Introduce an artistic language without copying a specific artwork.
• Leave meaningful blank space instead of filling the entire page.
• Use short, open-ended invitations rather than rigid instructions.

Another challenge was turning an unpredictable creative process into a reliable product flow. We addressed this by separating interpretation, matching, activity planning, and image generation into specialized stages with structured schemas.

Generating three personalized images can also take time or fail because of API availability. To keep the hackathon demo usable, we designed clear progress states and added graceful fallback examples when generation is unavailable.

Accomplishments that we're proud of

We are especially proud that Wonderline treats a child's drawing as the beginning of the creative process—not as rough material for AI to replace.

The project successfully turns one small doodle into three genuinely different paths of exploration. A single creature might become a geometric map inspired by Paul Klee, an adventure in cut-paper shapes inspired by Henri Matisse, or a world of imaginary symbols inspired by Joan Miró.

We are also proud of creating an experience that feels warm, playful, and approachable for both children and parents. Families do not need to understand art history or prompt engineering. They only need a drawing, a printer, and curiosity.

Most importantly, the child still makes the final artwork.

What we learned

We learned that personalization in education should mean more than inserting a child's name into a standard worksheet. Meaningful personalization begins by paying attention to what the child has already created and responding to it with care.

We also learned that AI is most valuable in creative education when it acts as a guide rather than a substitute. It can connect a child's spontaneous marks with broader artistic ideas, but it should leave room for uncertainty, play, and personal expression.

From a technical perspective, we learned how much more reliable a multimodal workflow becomes when it is divided into focused agents with structured outputs. We also gained a deeper understanding of image prompting, especially the importance of describing not only what the model should create, but what it must intentionally leave unfinished.

What's next for Wonderline

Next, we want to expand Wonderline from a hackathon prototype into a richer creative companion for families and educators.

Future directions include:

• Supporting more artistic languages, cultural traditions, and contemporary artists.
• Letting parents and teachers adjust activities by age, materials, and learning goals.
• Creating multi-page creative journeys that evolve with a child's artwork over time.
• Adding painting, collage, sculpture, storytelling, and mixed-media activities.
• Building classroom tools that help educators create personalized activities for multiple students.
• Offering multilingual prompts for families around the world.
• Improving privacy controls and creating a secure portfolio where children can revisit their creative growth.

Our long-term vision is for every child's drawing to become a doorway: not toward one correct result, but toward many possible ways of seeing, making, and imagining.

Built With

• next.js
• openai
• react
• typescript