Gopa | Devpost

Gopa App

Inspiration

GOPA was born from a shared vision to bridge the gap between ancient heritage and the digital future. We both felt that while there is an abundance of high-quality animation for children today, mythological stories often feel "distant" or too complex for the youngest generation. We wanted to give back to our culture by creating a way for immigrant parents to plant seeds of trust and values early. Our goal was to make these stories feel as "magical" as a modern Pixar movie, but with a deeply personal twist: letting the child literally step into the story as Krishna’s friend.

What it does

GOPA is a multimodal AI storytelling engine that transforms universal Vedic values into personalized 3D adventures for children aged 3–5.

Values First: Parents select a core lesson, such as "Friendship" or "Kindness."

Personalized Persona: By uploading a photo, our system uses AI to "re-imagine" the child as a 3D-animated character.

Immersive Storytelling: The app generates a 4-scene story where the child and Little Krishna interact in the vibrant, safe world of Vrindavan, focusing purely on "Bal Leela" (childhood mischief and joy).

How we built it

We built GOPA using a serverless, multi-agent architecture on Amazon Bedrock, coordinated through AWS Amplify:

The Brain (Amazon Nova Pro): This agent sifts through mythological themes to write simple, age-appropriate scripts that focus on kindness rather than complex philosophy.

The Artist (Amazon Nova Canvas): This is the heart of our personalization engine. We utilized Image-to-Image workflows to stylize the child's photo into a 3D character that matches the aesthetic of the world.

The Frontend: Developed with React and hosted on AWS Amplify, ensuring a secure and responsive portal for parents to manage their family’s stories.

Challenges we ran into

Our biggest hurdle was The "Cuteness" Calibration. When working with AI, achieving a specific "toddler" aesthetic for a character as iconic as Krishna requires extreme precision.Prompt Tuning: Our early tests produced characters that looked too old or lacked the "storybook" feel.Iterative Testing: We spent days testing and re-testing prompts to tune facial proportions and lighting. We didn't stop until the characters looked exactly like 3-to-5-year-olds and the calves in the background looked genuinely "mesmerized" by the scene.Consistency: We had to find the "Golden Prompt" that kept the characters looking the same across different scenes, which we calculated as :

{Visual Success} = {Nova's Logic} X {Dozens of Prompt Re-iterations}

Accomplishments that we're proud of

We are incredibly proud of the Persona Stylization feature. Successfully turning a real-world photo into a consistent, 3D-animated "friend" for Krishna felt like true magic. We are also proud of our strict Safety-First Prompting—creating an AI environment that is 100% safe for toddlers by filtering out the adult themes of mythology and focusing entirely on joy and mischief.

What we learned

Building GOPA taught us that AI is most powerful when it has strong human guardrails. We learned the art of Multimodal Orchestration—how to make a text-model "reasoning" agent talk to an image-model "creative" agent to produce a cohesive result. More importantly, we learned that technology can be a profound tool for cultural preservation when guided by the right intent.

What's next for Gopa

We want to expand GOPA beyond static images by integrating Amazon Nova Reel for high-fidelity video generation. We also plan to add Multilingual Support so that the stories can be narrated in regional languages like Hindi, Gujarati, or Tamil, helping kids stay even more connected to their roots. Finally, we hope to build a "Bedtime Library" where parents can save and collect all the personalized adventures their children have shared with Krishna.