What I Built

I built TalesForm, an AI-powered web application that enables anyone to become a children's book author. Starting with a simple idea, the platform generates a complete, beautifully illustrated, and personalized storybook that can be read online or downloaded as a PDF.

The application orchestrates multiple AI models working together. Gemini acts as the Storyteller, generating the narrative and story structure, Imagen creates custom illustrations for every page, and the ElevenLabs API transforms the generated story into expressive, natural-sounding narration, allowing every book to be experienced as an interactive audiobook in addition to a traditional picture book.

Why It's Playful

The entire experience is designed to encourage creativity and imagination. Rather than presenting users with a conventional form, TalesForm guides them through a playful, multi-step journey where they can customize the main character, choose the setting, and add themes or special ideas they want the story to include.

With its warm visual design, whimsical interactions, and optional creative prompts, the experience feels more like inventing a story together than filling out an application.

The playfulness extends beyond the story itself. Every generated book concludes with a custom coloring activity featuring the story's characters, inviting children to continue interacting with the world they've just explored and become co-creators of their own adventure.

What's Surprisingly Smart About It

The intelligence behind TalesForm lies in its multi-agent AI orchestration, where specialized models collaborate to produce a cohesive and immersive storybook.

The process begins with Gemini, which does far more than write a story. It generates a structured blueprint containing the narrative, page-by-page breakdown, and detailed descriptions of every main character—including clothing, hairstyles, colors, facial features, and other defining visual attributes.

This structured blueprint is then used to orchestrate the illustration pipeline. Instead of sending generic prompts to Imagen, the application dynamically constructs a detailed prompt for every page, embedding the character blueprint alongside the page-specific scene description. This ensures that characters maintain consistent appearances throughout the entire book, solving one of the biggest challenges in AI-generated storytelling.

To further enhance the experience, the final story generated by Gemini is passed to the ElevenLabs API, which converts the text into expressive, human-like narration. This allows readers to not only view a personalized illustrated storybook but also listen to it as a narrated audiobook, making the experience more engaging and accessible for young children.

By orchestrating multiple specialized AI models—Gemini for storytelling, Imagen for illustration, and ElevenLabs for narration—TalesForm creates a polished, immersive storytelling experience that feels far more cohesive than simply generating text and images independently.

Built With

Share this project:

Updates