BLW Journey: Understanding the Challenge

As a new mom, I’ve been feeling overwhelmed and anxious about starting Baby-Led Weaning (BLW) with my toddler. Like many parents, I worry about whether my child is eating enough or getting the right nutrition. But where do I even begin?

There’s so much conflicting advice out there. On one hand, some suggest including babies in the adult menu by offering small pieces of meat and vegetables from our meals. On the other hand, others recommend preparing separate, baby-specific meals. I’m trying to find a middle ground that works for our family, but it’s not easy.

One of my biggest concerns is keeping track of the foods my baby has tried to monitor for potential allergies. I’m also nervous about preparing food safely — like cutting vegetables into the right size to avoid choking hazards. And let’s not forget the practical challenges: grocery shopping with a stroller is no small feat, so I want to make the most of what I already have at home and create healthy meals without wasting food.

I’ve heard other parents complain about how messy BLW can be, and honestly, the thought of dealing with years of messy eating is daunting. But I also know that BLW helps kids explore different textures, tastes, and food cultures while learning to enjoy meals with their family. .

This is why I’m building a product to help parents like me navigate BLW with confidence.

Problem Statement

Parents, especially first-time moms like me, face overwhelming challenges in starting Baby-Led Weaning (BLW), including meal preparation, safety concerns, tracking food introductions, and managing mess, creating a need for a solution that simplifies the process and reduces anxiety while promoting healthy eating habits for toddlers.

Research Methodology

As a product manager diving into the BLW space, I knew that understanding our users’ genuine struggles was crucial. Rather than relying on traditional surveys or interviews, I took a novel approach that combined digital ethnography with AI-powered analysis.

My journey began in Singapore’s parenting Facebook communities, where I observed authentic, unfiltered conversations about BLW challenges. These weren’t just casual observations — I was conducting what UX researchers call “digital ethnography,” immersing myself in these online spaces where parents freely shared their daily struggles, victories, and questions about feeding their little ones.

What made this approach particularly valuable was its authenticity. Parents weren’t responding to prompted questions; they were expressing real concerns in their own words, sharing experiences with other parents who understood their journey. The Singapore context added another layer of cultural specificity, particularly around how families balance traditional Asian feeding practices with modern BLW approaches.

To transform these rich but unstructured conversations into actionable insights, I embraced AI as an analytical tool. I systematically documented these discussions and used Claude AI to help identify patterns and themes that might have been missed through manual analysis alone. This combination of human observation and AI-powered pattern recognition revealed deeper insights about parent coordination challenges, resource constraints, and knowledge management needs.

The most striking findings centered around three key areas: the complexity of aligning feeding approaches between multiple caregivers (particularly in Singapore’s context where many families rely on infant care centers), the universal challenge of time-constrained meal preparation, and the overwhelming nature of filtering through nutritional information.

This hybrid research methodology — combining traditional UX research techniques with AI-assisted analysis — proved invaluable in shaping BabyBites’ feature set. It ensured that we weren’t just building what we thought parents needed, but what they were actively seeking in their daily BLW journey.

Solution: BabyBites — AI-Powered BLW Assistant

Leveraging modern AI development tools, I built a web-based application that transforms the BLW journey through intelligent recipe generation and customization. The application is powered by Anthropic’s Claude 3 API for natural language processing and multi-modal capabilities.

Technical Architecture:

• Frontend: Streamlit for rapid UI development

• Backend: Python/Flask with Claude 3 API integration

• Features: Multi-modal input processing (image-to-text conversion), age-based recipe customization

Core Functionality:

1. Ingredient Processing

• Text input for manual ingredient listing

• Photo upload with automatic ingredient recognition

• Real-time inventory management

2. Recipe Generation

• Age-appropriate recipe suggestions

• Safety parameters for food size and texture

• Nutritional value optimization

• Alternative ingredient recommendations

3. Personalization Engine

• Age-based customization

• Allergy tracking system

• Meal type filtering (breakfast, lunch, dinner)

• Cultural cuisine preferences

Development Approach: Built iteratively using Replit’s browser-based IDE, allowing for rapid prototyping and testing. Leveraged AI pair programming for code optimization and debugging, significantly reducing development time for a non-technical founder.

The solution demonstrates how AI tools can be effectively utilized to build practical applications that solve real-world problems, while maintaining focus on user needs and safety requirements.

Challenges

Technical Challenges & Implementation Learnings

1. Image Processing & File Management Challenge:

• Initial implementation faced file size bottlenecks

• Inconsistent image upload behavior

• Legacy code conflicts from feature iterations

Solution:

- Implemented file size validation
- Standardized image format requirements
- Refactored upload handling logic

Learning: Legacy code management is crucial when iterating with AI-assisted development. Clean code practices become even more important when rapidly prototyping features.

2. Feature Evolution & Architecture MVP Phase:

• Core ingredient-to-recipe conversion

• Basic search functionality

• Minimal user interface

Iterative Additions:

• Meal type categorization (breakfast/lunch/dinner)

• Age-based filtering system

• Smart recipe recommendations

• Recipe management (deletion/history)

Technical Debt Encountered:

• Feature interdependencies created complexity

• Code cleanup requirements increased with each iteration

• AI response filtering needed optimization

3. Content Filtering Challenges Current Limitation:

• Recipe type filtering (soup, gravy, stir-fried) showing inconsistent results

• AI response parsing requires additional logic layer

• Edge cases in recipe categorization need handling

Key Learning: While AI accelerates development, it requires careful architecture planning and robust error handling, especially for natural language processing tasks.

4. UX Flow Optimization Original Process:

Input → Recipe Generation

Enhanced Flow:

Input → Age/Meal Selection → Recipe Generation → Additional Recommendations → Recipe Management

Development Insight: Working with AI tools like Claude 3 and Replit demonstrated both the power and limitations of AI-assisted development. Success requires balancing rapid iteration with structured development practices.

This experience highlighted the importance of:

• Early UX flow documentation

• Systematic feature integration

• Regular code refactoring

• Understanding AI model limitations

Next Step

As a product manager and a parent, I’m excited to share that I’m currently developing new features to enhance this solution, including gamified lessons on nutrition, tools to manage food waste, and multi-language support to make it accessible to more families. This project has become my AI testing playground — I’m leveraging AI tools to create engaging videos, songs, and games that educate both toddlers and parents in a fun and interactive way.

One of the most fascinating aspects of this journey is exploring how far a product manager can push the boundaries of AI. From writing code and generating art to building databases and measuring metrics, I’m testing the limits of what’s possible without deep technical expertise or traditional design skills. It’s a bold experiment, but one that I believe will unlock new ways to innovate and solve real-world problems for parents like me.

Let’s see where this AI-powered journey takes us — and how it can transform the way we approach parenting challenges!

MVP Link: https://babybites.chochan.xyz/