HILARY
An innovative, AI driven product to generate medical documentation.
About project
As a multidisciplinary team of professionals from the AI, medical, and UX/UI design industries, we initiated the development of a Polish open-source SLM called Hilary. The project focused on advancing the model’s language capabilities by training it on carefully curated medical documentation and domain-specific knowledge. Alongside the technological development, we created the branding for both the organization and the product itself, while also beginning early research and concept work for a future digital healthcare product powered by the model. We conducted both qualitative and quantitative research with medical sector professionals, including doctors, nurses, and researchers, to better understand real-world workflows, challenges, and opportunities for AI integration in healthcare. Due to organizational registration issues and the departure of key SLM engineers to other organizations, the project was ultimately concluded during the research and early development stage.
Client
SOWA (Supportive Optimization with AI)
Services
UX/UI Design
Year
2024
Industry
Medical
Problem
In the aging societies of the West, the medical industry is facing increasingly significant challenges. A growing number of patients and limited human resources are leading to the inefficiency of the healthcare system.
Vision
The solution to this problem is Hilary – a generative Small Language Model (SLM) specifically designed for the medical sector. This model is intended for producing high-quality medical texts in multiple languages, with particular focus on Polish and English.
The model’s availability and openness will enable healthcare-related companies and institutions to implement generative AI solutions within their own infrastructure – independently of commercial providers.
Hilary is expected to significantly contribute to increasing the efficiency of the medical sector – both in primary care (by reducing the time required to prepare documentation) and in hospital emergency departments (by rapidly comparing cases), where response time is critical.
What we did?
We began the project with extensive benchmark analysis and market research to better understand the rapidly evolving landscape of AI in healthcare, existing language model solutions, and the unmet needs within the medical sector. This early research phase helped us identify opportunities for a Polish open-source SLM tailored to real clinical environments and workflows.
Based on these findings, we developed the branding and visual identity for both the organization and the Hilary product, while simultaneously exploring the direction and structure of a future digital healthcare platform powered by the model.
As the project evolved, we conducted both quantitative and qualitative research with medical professionals, including doctors, nurses, and researchers. The feedback was highly enthusiastic, with many participants highlighting the potential of AI-assisted tools to significantly accelerate medical case analysis and support decision-making processes. Particularly in emergency departments, where time-sensitive decisions can directly impact patient survival.
A significant part of the research focused on ethical considerations surrounding AI adoption in healthcare. We explored concerns related to model errors, accountability and responsibility in clinical decision-making, as well as the psychological and practical boundaries of risk-taking while working under pressure in high-stakes medical environments.
