EGPAF: glAIser - AI Applied to Global Health
Executive Summary
The Elizabeth Glaser Pediatric AIDS Foundation (EGPAF) is a global organization dedicated to the prevention and eradication of pediatric HIV, operating in multiple countries with large-scale public health programs. The foundation works with massive volumes of health data, distributed across different national information systems, which made consolidating and analyzing this data a critical challenge.
The main problem faced by EGPAF was the fragmentation of health data: disconnected systems, report generation times measured in months, and the absence of an intelligent layer that translated raw data into actionable insights for field decision-making.
Squadra Digital, in partnership with Microsoft, led the development of glAIser — an intelligent AI-based assistant that interprets natural language commands, cross-references large volumes of health data, and generates reports, visualizations, and action recommendations. The solution, initially conceived at Microsoft Hack4Good, was evolved by Squadra into a production platform with global scale, covering the 17 countries served by EGPAF.
With Genius in action, the time between data collection and report generation dropped from months to seconds, strengthening the data culture and directly contributing to saving lives through early diagnoses and focused interventions.
The Challenge
EGPAF operates in several Sub-Saharan African countries, each with its own health information systems. The need to consolidate, analyze, and act on this data quickly and reliably was essential to the organization's mission. The main challenges identified were:
Data fragmentation across countries: Health information systems varied among the countries of operation, lacking interoperability. Critical data remained isolated in silos, hindering comparative analyses and a consolidated view.
Excessive time for report generation: The cycle between data collection and the production of actionable reports took months, compromising agility in responding to outbreaks, allocating resources, and monitoring programs.
Lack of data intelligence: There was no AI or advanced analytics layer to transform raw data into practical and accessible insights for different user profiles — from strategic managers to field professionals.
Technical barrier to information access: Users with different technical levels needed to consult data, but existing tools required specialized knowledge, limiting the democratization of access to information.
The Solution
Squadra Digital led the development of glAIser in a progressive approach, moving from a hackathon prototype to a global production platform.
Genius in action: Conversational AI for public health
Phase 1 — Hackathon and Prototyping
The project originated at Microsoft Hack4Good, where Squadra Digital acted as a strategic partner, mobilizing expertise in data, AI, and technology to formulate the glAIser proposal.
AI-driven scope definition: Consolidation of health systems from multiple countries, natural language queries, and generation of actionable visualizations were defined as the prototype's pillars.
Functional MVP: Squadra took the project beyond the minimum MVP, ensuring it became the foundation for a global-scale initiative.
Phase 2 — Foundational MVP
After the hackathon, Squadra took the lead in scaling up, transforming the prototype into a robust and operational platform.
Data Lakehouse Architecture: Data centralization in a data lakehouse with ETL pipelines for extraction, transformation, and loading of data from EGPAF's countries of operation.
Governance and Security: Orchestration and security via Azure, with defined access levels for strategic, operational, and clinical use.
The construction of glAIser involved the entire data and AI stack, with Squadra acting as the technical and strategic integrator.
Natural Language AI Agents: Conversational interface based on Azure OpenAI Service with RAG (Retrieval-Augmented Generation) for command interpretation and insight generation.
Data Pipelines: Azure Data Factory and Azure Databricks for large-scale data processing and transformation.
Scalability and Interoperability: Platform architected to allow integration with partner government systems, expanding the solution's reach.
User Experience: Focus on usability to ensure professionals with different technical levels could operate the platform autonomously.
Phase 3 — Technical Evolution and Global Scale
Squadra ensured the solution evolved from a hackathon concept to a globally mission-driven production context.
Advanced AI Architecture: The solution evolved into an advanced architecture based on multiple intelligent agents, capable of interpreting complex queries, distributing tasks among specialized agents, and increasing the reliability of analyses performed.
Enhanced Analytical Capacity: The platform expanded from 38 indicators to support over 142 complex combinations of clinical data, significantly increasing its analytical and insight generation capabilities.
Global Coverage: The solution now serves all countries where EGPAF operates, integrated into the foundation's unified data ecosystem, ensuring standardized analyses on a global scale.
Accuracy and Reliability: The new architecture was developed to provide more accurate and reliable analyses, strengthening data-driven decision-making in a critical context for public health.
Results
glAIser, led by Squadra, transformed how EGPAF conducts its public health operations. The solution connected fragmented data from multiple countries into a unified platform with artificial intelligence, generating measurable impact across all organizational fronts.
Ultra-Fast Response Speed: The time between data collection and report generation dropped from months to 15-20 seconds, democratizing access to information and strengthening a data-driven culture.
Broad Geographic Coverage: The platform covers 100% of the foundation's operational countries in a standardized manner through the Glaser360 metadata framework, directly contributing to saving lives by enabling early diagnoses, more efficient monitoring, and focused interventions in critical regions, bringing the world closer to eradicating pediatric HIV.
Cutting-Edge Architecture (Multi-Agent State Machine): The major technical breakthrough was the migration from a simple rule-based model to a structure of Multiple Intelligent Agents Orchestrated via LangGraph.
Clinical Data Security: The AI features a real-time self-healing loop. If a generated query has any syntax errors, the system itself identifies the error, corrects it, and recalculates the route before displaying the final result, eliminating hallucinations and ensuring the security of clinical data. Furthermore, it achieved 100% accuracy in automated tests against the Golden Dataset (a set of real, highly complex data and scenarios provided by the client's data specialists).
The project stands out globally due to its technical quality, Squadra's commitment, and professionalism in delivering a robust solution, prepared to support critical decisions and generate real impact in global public health:
"This is truly impressive work... Seeing the solution in its current state had a pretty profound impact on me."
— Jon Drew (Account Team, Microsoft)
"A heartfelt thank you to the Squadra team for your outstanding delivery, unwavering commitment, and consistently high-quality work. This has been a remarkable collaboration from start to finish."
— Hugo Marques (SMB & Partner Lead, Microsoft Elevate)
"Thanks Cristiano for the dedication and professionalism through this process."
— Charlie Maere (Global Digital Health Director, EGPAF)
Gallery
Want results like these?
Talk to our experts and find out how AI can transform your business.
