Breaking down healthcare’s walls with agentic AI

As healthcare leaders gather for this year’s World Health Summit under the theme “Taking Responsibility for Health in a Fragmenting World,” we must confront an uncomfortable truth: the challenge isn’t that we lack data—it’s that we’ve built walls where we need bridges. When 97 percent of healthcare data goes unused, and the average elderly patient must navigate seven different physicians across four practices, we’re witnessing the limitations of traditional healthcare design. To address these challenges, innovative solutions are being developed and implemented across the globe.

The path forward lies in autonomous, intelligent systems that can transform this fragmented landscape into a seamless continuum of care. This is where agentic AI emerges as a cornerstone for healthcare’s next evolution—not just as another layer of technology, but as an intelligent force that can both derive insights and autonomously initiate the actions needed to enhance human lives. It’s about moving from passive data collection to proactive care delivery that understands, anticipates, and responds to the full spectrum of human health needs. In this post, we will explore how agentic AI solutions are transforming healthcare delivery and creating more resilient health systems.

Transforming healthcare with AWS agentic AI solutions

Building on this vision of proactive, intelligent healthcare, Amazon Web Services (AWS) has developed a comprehensive suite of agentic AI solutions that are reshaping how healthcare organizations approach patient care, operational efficiency, and system resilience. AWS meets customers wherever they are on their agentic AI journey, offering everything from ready-to-deploy agents to tools for building sophisticated custom solutions. Unlike traditional AI that simply responds to prompts, these agentic AI solutions can reason, plan, and take autonomous actions to accomplish complex healthcare goals with minimal human oversight.

Amazon Bedrock is a fully managed service providing healthcare organizations access to industry-leading foundation models with built-in capabilities designed for healthcare applications. Bedrock’s advanced memory retention ensures continuous patient context across fragmented care settings, while its multi-agent collaboration enables specialized medical tasks with integrated guardrails for healthcare security and compliance. Amazon Bedrock AgentCore provides enterprise-grade tools to deploy and operate healthcare agents securely at scale, with robust controls for patient data privacy and regulatory compliance.

AWS Transform offers purpose-built solutions for healthcare system modernization, deploying specialized AI agents that can navigate the intricacies of diverse healthcare workflows across different regions and healthcare models. These agents leverage 19 years of migration experience to automate complex tasks like assessments, code analysis, and transformation planning—dramatically reducing project timelines while maintaining healthcare compliance standards.

Amazon Q Business provides a generative AI assistant customized for healthcare workflows, integrating with common healthcare enterprise software and automating medical administrative processes through a natural language interface—helping to standardize care quality while respecting local medical practices and regulations. Real-world implementations have shown significant improvements in operational efficiency. For instance, Availity, the largest real-time health information network in the US, achieved data insights 2X faster and reduced their release management reviews by 75 percent using Amazon Q agents.

By bridging the gap between data collection and intelligent action, AWS is enabling a future where healthcare systems don’t just react to illness, but proactively nurture health, and a future where every data point, every interaction, and every decision is part of a larger, intelligent system designed to serve the whole person. This vision is supported by robust infrastructure, including Amazon S3 Vectors which reduce vector storage costs by 90 percent while maintaining sub-second query performance. This enables healthcare agents to maintain comprehensive context from every patient interaction and clinical insight.

These innovations and their impact on global healthcare will be showcased at the upcoming World Health Summit 2025.

Join us at World Health Summit 2025

The presence of AWS at the World Health Summit 2025 in Berlin this October 12-14 will showcase how these innovations bridge gaps and foster collaboration in our interconnected—yet often fragmented—healthcare landscape. We invite you to join our roundtable “Future-Proofing Healthcare: Agentic AI for Resilient Health Systems,” where we’ll explore how agentic AI is accelerating medical research and innovation.

Also, don’t miss Dr Myriam Fernández´s presentation on “Innovating for Health Equity: How AI Can Strengthen Frontline Systems” on Tuesday, October 14 from 14:00–15:30 CET and the AWS-led session “From Data to Discovery – AI’s Transformative Role in Health Research” on Monday, October 13 from 14:00-15:30 CET.

Real-world impact: Agentic AI case studies

To illustrate the transformative potential of agentic AI in healthcare, we will present two case studies that demonstrate how these technologies are being applied in diverse healthcare systems.

Case study: ALMA – Transforming healthcare in Catalonia, Spain

A compelling example of agentic AI’s transformative potential can be seen in Catalonia’s public health service (CatSalut). Faced with the challenge of keeping 20,000 healthcare professionals current with rapidly evolving medical guidelines, CatSalut implemented ALMA (Advanced Learning Medical Assistant), an innovative agentic AI solution built by BinPar on AWS.

Implementation architecture and results

The implementation architecture of ALMA combines Amazon Bedrock’s core AI capabilities with a custom Retrieval Augmented Generation (RAG) implementation for clinical knowledge management. The system features secure SSO integration with the health system, and provides immediate access to evidence based clinical guidelines. The results have been remarkable: 65 percent have integrated ALMA into their routine work, 98 percent user satisfaction rating, and 98 percent accuracy on the Official Medical Residency exam. Most significantly, the solution is successfully scaling across the primary care network and is poised for expansion into additional services. By delivering contextually relevant clinical data to physicians at the point of care, ALMA empowers healthcare providers to make optimal clinical decisions.

Security and compliance

Security and compliance remain paramount in our interconnected yet fragmented healthcare landscape. ALMA’s European GDPR (General Data Protection Regulation)-compliant architecture, end-to-end encryption, and comprehensive audit trails ensure that sensitive medical information remains protected while maintaining full transparency of AI interactions across diverse regulatory environments.

Case study: NoHarm – Preventing adverse events in Brazil’s public health system

NoHarm is a nonprofit AI initiative developed to support Brazil’s Public Health System (SUS) by preventing adverse drug events and improving patient safety. Already deployed in over 200 hospitals, NoHarm has directly impacted more than 3 million lives across Brazil, with its AI analyzing over 5 million prescriptions per month and its Named Entity Recognition (NER) system processing 250,000 clinical notes per day.

Implementation architecture and results

Built on AWS, NoHarm’s architecture integrates hospital electronic medical records with large language models (LLMs) for clinical summarization, alongside a custom Portuguese-trained Named Entity Recognition for patient safety (NER) system to identify risks in prescriptions and notes.

The system analyzes prescriptions in real time, flagging potential risks such as drug-drug interactions, duplications, and incorrect dosages. The system has already delivered significant results: Hospitals using NoHarm report up to 8x faster prescription analysis, enabling the same clinical team to evaluate up to 800 patients per day compared to 100 before NoHarm, and reduced workload for clinical pharmacists—freeing up professionals to focus on high-complexity cases. In addition, the system generates an average of USD $500,000 in savings per year for every 200 hospital beds.

Technical innovations and learning capabilities

NoHarm addresses challenges unique to Brazilian healthcare: fragmented data, local terminology, and limited access to specialized staff. Its innovations include:

• A domain-specific NER model trained on Portuguese clinical corpora, enabling accurate detection of adverse event risks directly in medical text.
• LLM-based summarization of patient data and clinical notes to support faster, safer decision-making at the point of care.
• A feedback-driven learning loop where pharmacist input continuously refines model accuracy, ensuring alignment with clinical practice.
• An open-source anonymization system developed, ensuring compliance with Brazil’s LGPD (General Data Protection Law)—the equivalent of HIPAA in the US—while enabling safe model training and deployment.

Security and compliance

As a health AI deployed in public hospitals, compliance is central. NoHarm follows Brazil’s LGPD data protection law with end-to-end encryption, anonymization of patient data, and secure access controls. Audit logs track every AI-driven recommendation, maintaining accountability and trust. By prioritizing open-source development, NoHarm fosters transparency and reproducibility while adhering to international privacy standards.

These case studies highlight the practical applications and benefits of agentic AI in healthcare, and showcase how these technologies will continue to shape the future of health systems.

Building resilient health systems

Agentic AI is transforming the healthcare industry. It is revolutionizing resource optimization through intelligent scheduling and predictive maintenance for medical equipment, which is crucial in resource-constrained settings. It also enhances clinical decision support with real-time access to guidelines and evidence-based treatment recommendations, helping to standardize care quality across fragmented systems. Healthcare access improves through remote care support and multilingual capabilities, breaking down language and geographic barriers. Quality standardization benefits from consistent application of clinical protocols and reduced care variability, essential in creating resilient health systems that can deliver consistent care despite global fragmentations.

To learn more about how AWS can help your organization leverage agentic AI for healthcare transformation, visit our healthcare solutions page or connect with our specialists. Together, we can build the future of healthcare.

Further reading

• Webpage: AWS agentic AI solutions
• Webpage: AI for Good at AWS
• Complimentary whitepaper: Expanding Horizons: Scalable Technology Solutions for National Healthcare
• Blog: How agentic AI systems can solve the three most pressing problems in healthcare today