Nowa is a burgeoning digital health startup developing a fertility app that supports individuals and couples with reliable information and tailored recommendations. With Rakam, they designed their product from the ground up, integrating AI capabilities for extracting medical data, generating personalised reports, and distributing content validated by experts.

Client

Nowa is a mobile application designed to guide individuals and couples on their fertility journey. It delivers clear, personalised information grounded in scientific evidence to help them progress confidently.
The app offers educational content on fertility and reproductive health. While it provides analysis, interpretations, and certain diagnostic elements, it is not intended to replace medical supervision by a doctor.

Challenges

Nowa approached Rakam with the ambition to transform an idea into a functional product. The key challenges were:

• Extraction of medical data: structuring information from complex medical documents (test results, reports, etc.).

• Personalised reports: generating reports tailored to each stage of the fertility journey, taking into account both medical data and user responses to questionnaires.

• Research and article recommendations: enabling search within a library of articles to recommend the most useful content.

• Reliable and responsible AI: integrating an expert-validated knowledge base to anchor LLM-generated answers and prevent misinformation.

An additional challenge was that the app did not yet exist and the product roadmap was not clearly defined. Hence, Rakam co-built with Nowa, prioritising and developing functionalities step by step.

What We Built

With Nowa, Rakam established:

• A structured data extraction pipeline using an OCR and LLM benchmark to process and standardise medical documents.

• A PydanticAI-based agent framework to orchestrate workflows and ensure the consistency of user data processing.

• A personalised report generation engine, adapted to each stage of the fertility journey and the user's specific context.

• Article search with Sentence Transformers, enabling the retrieval and recommendation of the most relevant content.

• An expert-validated knowledge base, providing a foundation for model-generated responses to enhance accuracy and trust.

• A front-end React integration, aligned with the app design.

Impact / Results

With the app still in development and not yet launched to end-users, quantifiable results are forthcoming. However, we have already enabled Nowa to:

• Transform an initial idea into a functional prototype with a clear roadmap.

• Validate functionalities upfront with stakeholders and medical experts.

• Lay the groundwork for a scalable, safe product suited to a sensitive medical context.

Services Used

• RAG / Knowledge Base Anchoring (expert-validated)

• Custom Agents (PydanticAI workflows)

• Intelligent Search Engines (embeddings with Sentence Transformers)

• Conversational Agents (LLM personalisation)

• Benchmark OCR + LLM for data extraction

• React Front-end