Mamabase - AI Parenting Companion

Mamabase is an emotionally intelligent AI parenting companion— our first feature, StoryCoach turns real-life moments into personalized stories that help kids grow.

Comment

📖 Project Story: Mamabase – AI Parenting Companion

💡 Inspiration

As a mom of two — a curious five-year-old and a seven-month-old — parenting is constantly on my mind. Over time, I noticed something powerful: stories help my older child process big emotions far more effectively than lectures or logic.

Whether it was jealousy over a new baby sister, frustration about losing a game, or resistance to eating vegetables, I’d turn to storybooks to explore those feelings together.

Eventually, I ran out of good library picks. So, as a tech-savvy parent, I tried ChatGPT. While functional, the stories were too generic and emotionally flat.

That’s when the idea for Mamabase was born — an emotionally intelligent AI parenting companion that could create custom stories based on real-life parenting moments.

🧠 What it does

Mamabase is designed to support parents emotionally, while offering children personalized stories that help them understand and navigate tough feelings.

The first feature I’ve launched is called StoryCoach. It allows a parent to:

  1. Share a parenting moment (e.g., "My daughter cries when I leave daycare")
  2. Receive a warm emotional reflection and parenting insight
  3. Get a custom, 500+ word children’s story tailored to the situation, tone, and style

🛠️ How I built it

  • Frontend: Built using Bolt.new, a no-code prototyping tool that enabled me to design and launch a web app without writing code
  • Backend: Integrated with Supabase for user authentication, storage, and profile management
  • AI Engine: Originally built using Relevance AI, but later migrated to OpenAI’s GPT-4 API to gain more control over story freshness, output structure, and emotional tone
  • Illustrations: Each story also triggers an AI-generated black-and-white illustration based on story content

⚠️ Challenges I ran into

  • No-code pain points: While Bolt is powerful, debugging errors like “Fix Metro configuration for TypeScript support” can be frustrating for someone who doesn’t code
  • AI story repetition: Early versions of the agent kept generating forest scenes with owls and dragons. I had to carefully rewrite prompts to enforce variety, tone, and relevance
  • Tool chaining: It took a lot of iteration to orchestrate structured outputs like story metadata, image generation, and storage in Supabase
  • Prompt tuning: It was hard to strike the right balance between warmth, variety, and emotional depth — especially when guiding AI to behave like a wise parenting companion

🏆 Accomplishments that I'm proud of

  • Built a fully working AI parenting companion prototype — solo
  • Designed and deployed a warm, usable experience on Bolt.new without writing code
  • Created a structured output format with story, title, teaser, emotions, and illustrations
  • Introduced a truly helpful tool for emotionally intelligent parenting

📚 What I learned

  • Prompt engineering is a superpower: Small tweaks in structure made a huge difference in output quality
  • No-code ≠ no effort: Tools like Bolt and Supabase are amazing, but still demand a builder’s mindset and persistence
  • Emotionally relevant AI is possible, but only with human-centered design and iteration

🚀 What’s next for Mamabase

I'm just getting started. Here’s what’s on the roadmap:

  • Audio narration of stories
  • Profiles for each child
  • Learning progress tracking
  • A growing collection of story styles and tones
  • Agent-powered coaching tools beyond stories

Mamabase isn’t just about storytelling. It’s about empowering parents with AI that reflects empathy, warmth, and real support.

Join me on this journey — the beta is live and I’d love your feedback!

Built With

  • bolt.new
  • gpt-4
  • openai
  • react
  • supabase
Share this project:

Updates