AI is bringing a revolution in how families plan meals. And we got to see it firsthand while building a state-of-the-art AI health assistant for Little Lunches. The app itself has become one of the fastest-growing personalized nutrition services in the world, with more than 500,000 parents using it every day. It offers age-appropriate recipes and feeding advice reviewed by certified dietitians, pediatricians and feeding therapists.
With the community growing at lightning speed, delivering meal plans and nutrition guidance that reflect each family's individual needs becomes more demanding. Meeting those expectations calls for a dynamic system that can adapt while maintaining the quality and consistency parents rely on.
Why Little Lunches needed more than a static ai health assistant
As Little Lunches expanded, parents increasingly looked for quick and reliable answers about their children's nutrition and everyday meal decisions. The platform offered expert-reviewed resources, but without an AI health assistant, it relied heavily on static content and manual support. This made it extremely difficult to respond to a growing number of users, especially across different languages, while still giving families advice that matched their individual situations.
The team also needed to protect its proprietary content. Without a structured data foundation, there was a greater risk of exposing original material and limited ability to build lasting, personalized ai health assistant support.
Jessica Facussé, the founder of Little Lunches, in a recent goLive episode with Rishabh Sood, founder and CEO at GoML, explained that feeding a family goes far beyond just following a recipe. Parents plan meals, shop for groceries, work around food dislikes and make countless small decisions every day. She wanted the app to reduce that daily effort by offering practical guidance whenever families needed it.
Meeting that goal meant building a multilingual AI Nutrition Assistant designed specifically for family nutrition, one that could deliver personalized support while keeping the platform's trusted expertise secure.
Why GoML was the right partner for Little Lunches
The GoML team built this platform for Little Lunches on Amazon Web Services (AWS), giving the team a foundation that could support growing datasets, personalized nutrition workflows and real-time recommendations as the app expanded. This allowed new capabilities to be added without repeatedly changing the underlying infrastructure.
To make the most of that environment, the team looked for a partner with practical experience in generative AI, strong AWS expertise and a track record of building nuanced AI systems in health and nutrition.
When Jessica partnered with GoML, the company had already delivered around 130 generative AI applications, many within the same industry. Jessica praised the collaboration with GoML, stating that the team was dynamically focused on finding practical solutions that kept the project moving.
What was built into the Little Lunches nutrition assistant
GoML built an MVP AI nutrition assistant focused on family and children's nutrition. The assistant uses Agentic AI blueprint as part of the AI Matic framework to answer parents' questions in natural language, offering responses that consider each family's circumstances. Alongside nutrition guidance, it recommends relevant recipes, articles and feeding guides by retrieving metadata rather than exposing proprietary content.
Parents interact with the ai health assistant assistant through a conversational interface designed to make asking nutrition questions feel simple and approachable. It supports both English and Spanish and adapts its responses based on details such as a child's age, allergies, and dietary preferences. The assistant can also suggest helpful next steps, such as converting a recipe to be dairy-free or adding a meal to a family's meal plan, making everyday meal planning easier to manage.
Security and privacy are built into every interaction. User prompts are securely processed through Claude on Amazon Bedrock, allowing the assistant to answer questions within the scope of nutrition and feeding. It uses the context shared during a conversation without retaining or exposing full proprietary content, helping protect Little Lunches' intellectual property while still delivering relevant guidance.
The assistant retrieves relevant information while keeping proprietary content protected. Recipes, articles, and feeding guides are indexed as vector embeddings in a PostgreSQL database using pgvector for semantic search. When a user submits a query, the system returns only metadata such as the title, summary, tags, and content ID. Any related resources are presented alongside the response without revealing the original source material.
Interaction data is also handled with privacy in mind. Usage logs are stripped of personally identifiable information before they are stored in an Amazon S3 data lake. The records capture details such as user prompts, assistant responses, language, dietary tags, and requested nutrition topics. This information helps the team study how the assistant is being used and refine its performance over time without compromising user privacy.
Results
- 80% scaled expert nutrition guidance through an AI nutrition assistant
- ≥90% successful recipe/article match rate
- Sub-4 second response latency for assistant interactions
"Our experience has been extremely positive with GoML. The team has been very responsive, and they've been supportive throughout the entire process. It's been a pleasure working with the team to develop this into a platform that can support millions of families."
~Jessica Facussé (Co-founder), Little Lunches
Watch the complete conversation between Jessica Facussé and Rishabh Sood (Founder and CEO, GoML) in our full GoLive episode to learn more.
The picture of success for Little Lunches
Parents have now been able to use Little Lunches's services from nearly 100 different countries. Jessica's final message to founders was to start with an everyday problem rather than start with an ai health assistant and to use intelligence to reduce friction rather than use it just to have it.
Keep an eye out on the GoML blog to stay up-to-date with the latest in AI and ML engineering.
