Millions Prescription Decisions: Vademecum

Introduction

Vademecum, a leading healthcare platform for drug information and clinical decision support, modernized its infrastructure by migrating to AWS and integrating generative AI with Amazon Bedrock in collaboration with AWS Partner Sufle. The transformation enabled the platform to process millions of daily prescription decisions with higher reliability, lower latency, and improved uptime, while introducing AI-powered semantic search for clinicians. Leveraging AWS’s scalable cloud services, Vademecum achieved faster performance, enhanced security and compliance, and a more intuitive user experience. Query response times dropped, throughput increased, and uptime reached enterprise-grade levels, allowing healthcare professionals to access critical drug information in seconds.

About the Customer

Vademecum Group is a healthcare technology provider with over 37 years of experience in drug information services. It delivers medication databases, drug interaction checks, and decision support tools used at the point of care, integrated into national e-prescription systems in Turkey and Qatar. Thousands of clinicians rely on Vademecum’s online and mobile applications daily to make safe prescribing decisions. Before its cloud transformation, the on-premises infrastructure—processing more than 10 million prescription-related queries per day—struggled to scale, maintain low latency, and rapidly update drug knowledge while meeting strict compliance and reliability requirements.

Challenge

Vademecum faced several challenges as it sought to scale and innovate:

• Scalability & Performance Limits: The legacy infrastructure struggled to handle surging query volumes (more than 10 millions per day) efficiently. Peak usage (e.g. during clinic hours) led to high server loads, threatening slower response times for clinicians. Consistently low latency is critical in healthcare, where immediate access to drug data can directly impact patient care decisions. Vademecum needed an elastic solution to scale on demand without performance degradation.

• High Availability & Reliability: As an integral part of e-prescription workflows, Vademecum’s services must be available 24/7. Outages or downtime could disrupt prescription processing and patient safety. The existing on-premises setup had single points of failure and required manual maintenance downtime. The company required an architecture with built-in redundancy (multi-AZ, fault tolerance) to achieve near-100% uptime.

• Generative AI Integration: Vademecum saw an opportunity to leverage AI to improve how clinicians and pharmacists search and retrieve drug information. Instead of forcing clinicians and pharmacists to perform keyword lookups or read lengthy monographs, a generative AI could allow natural language queries and provide concise answers. However, implementing such an AI assistant on legacy infrastructure posed challenges: it needed significant compute resources, a way to store and search embeddings (semantic vectors), and careful handling of sensitive medical data. Ensuring accuracy and safety of AI responses (e.g. for drug interaction checks) was a top concern, the AI had to be grounded in Vademecum’s trusted data.

In summary, Vademecum needed to modernize its infrastructure to meet growing demand and to enable new AI capabilities, all while maintaining the trust of users and regulators in a high-stakes healthcare environment.

Why AWS

Vademecum chose AWS as its cloud platform for several key reasons:

• Scalability and Global Infrastructure: AWS provides on-demand scalability and a global network of data centers. This allows Vademecum to seamlessly scale out capacity during peak prescription times and serve users in multiple countries with low latency. AWS’s multi-AZ architectures ensure high availability by design, so the platform could achieve reliable service even if an entire data center goes down. AWS’s breadth of managed services meant Vademecum could offload heavy lifting (server provisioning, clustering, etc.) and focus on its application.

• Healthcare Compliance and Security: AWS has a proven track record in supporting healthcare workloads. It offers 146+ HIPAA-eligible services and complies with 143 security standards and certifications including HIPAA, HITECH, GDPR and HITRUST. This gave Vademecum confidence that moving to AWS would not compromise data security. Fine-grained identity and access management (IAM), encryption (AWS KMS), network isolation (VPC), and monitoring (CloudTrail, Config) provided the tools to build a highly secure, compliant environment. AWS’s commitment to keeping customer data private and secure, with features like data sovereignty controls and AI service guardrails, aligned with Vademecum’s needs.

• Advanced Services (Generative AI and Analytics): AWS offers cutting-edge services that Vademecum could leverage to leap forward in innovation. In particular, Amazon Bedrock provides access to state-of-the-art foundation models via API, making it much easier to integrate generative AI into applications. Instead of building and hosting AI models from scratch, Vademecum could use Bedrock to tap into pre-trained large language models (LLMs) for tasks like Q&A and summarization. AWS also supports a Retrieval-Augmented Generation (RAG)approach through Bedrock Knowledge Bases, which automates the heavy lifting of combining vector search with model prompts. Additionally, AWS’s analytics and database services (from Amazon Aurora to Amazon OpenSearch) meant Vademecum had a rich toolbox to modernize data storage and retrieval with high performance. In short, AWS’s ecosystem offered “one-stop” access to the broadest set of generative AI capabilities and industry-leading models for healthcare use cases.

In sum, AWS and Sufle offered Vademecum an enterprise-grade cloud platform and professional services that meets critical healthcare requirements, scales effortlessly, and accelerates innovation, all essential for transforming Vademecum’s services and maintaining its reputation for reliability and accuracy.

Why Sufle

Vademecum partnered with Sufle, an AWS Advanced Consulting Partner, to plan and implement this transformation. Sufle was selected for its deep expertise in AWS cloud solutions and its experience with projects in regulated industries like healthcare. Key reasons for choosing Sufle included:

• AWS Expertise and Credentials: Sufle is an AWS Advanced Services Partner specializing in Cloud, DevOps/DevSecOps, Software Development and Compliance services. This meant Sufle had certified cloud architects and a track record of successful AWS deployments. Vademecum needed a partner who not only knew AWS services inside-out but could also tailor them to meet strict compliance and performance needs.

• Proven Migration Experience: Sufle has migrated hundreds of enterprise businesses to the cloud, demonstrating a robust methodology for large-scale cloud projects. Their team was well-versed in AWS Well-Architected best practices and could anticipate challenges in refactoring Vademecum’s legacy systems for AWS. This experience reassured Vademecum that the migration would be done with minimal disruption to ongoing operations.

• Compliance & Security Know-How: Sufle differentiates itself with knowledge of local and international compliance standards. Their architects design AWS infrastructures with security built in, from identity management to encryption and monitoring. Given the sensitive nature of prescription data, Sufle’s focus on trust and compliance was a perfect fit. They understood how to configure AWS services (like enabling audit logs, setting up HIPAA-eligible services, etc.) to satisfy healthcare regulators and Vademecum’s own data governance policies.

• Full-Stack Solution Capability: Beyond just the migration, Sufle could assist in the full solution implementation, including the generative AI component. Sufle’s team was up-to-date with AWS’s AI services (like Amazon Bedrock) and had internal machine learning expertise to help integrate the pgvector and Bedrock solution. Their ability to cover both the infrastructure modernization and the AI/ML engineering meant Vademecum had a one-stop shop to realize the project vision end-to-end.

During the transformation, Sufle not only executed the technical work but also mentored Vademecum’s engineers on AWS. “They taught us how AWS works… it would take weeks instead of days [without Sufle]”, noted one client. This knowledge transfer was invaluable: Vademecum’s team became proficient in operating the new cloud environment, ensuring long-term self-sufficiency. Overall, Sufle’s mix of AWS prowess, understanding of healthcare needs, and collaborative approach made them the ideal partner to lead Vademecum’s cloud and AI journey.

Solution

AWS Architecture Overview and Modernization

Vademecum’s platform was completely rebuilt on AWS using a microservices and serverless-first architecture. The solution runs in a secure AWS Virtual Private Cloud (VPC) spanning multiple Availability Zones for high availability. It combines containerized application services, managed databases, and AI integration via Amazon Bedrock.

At the heart of the modernized infrastructure is Amazon ECS on AWS Fargate for running Vademecum’s application containers. The monolithic legacy application was refactored into a set of microservices (for user interface, API, search, analytics, etc.), each packaged into Docker containers. Using AWS Fargate provides serverless container orchestration, so Vademecum does not manage any EC2 servers; containers are automatically run on-demand. With auto-scaling configured, the ECS cluster can scale out additional Fargate tasks within seconds when traffic spikes (for example, mid-day when prescription volumes are highest) and scale back down in off-peak hours, optimizing resource usage.

The platform’s primary data store moved to Amazon RDS for PostgreSQL (Aurora PostgreSQL). This managed database provides the transactional consistency and relational schema Vademecum needs for drug reference data. Aurora was deployed in a Multi-AZ setup, meaning it maintains a live standby in a second AZ to failover transparently if needed. Aurora’s high-performance and read scaling features ensure that even as the number of users and queries grows, the database can handle the load with low latency.

Amazon Simple Storage Service (S3) and Amazon CloudFront were introduced to serve static content and media much more efficiently. Vademecum’s drug information includes images (e.g. pill images), PDFs, and other static resources. These are now stored durably in S3 and delivered to end-users through CloudFront’s global content delivery network. This significantly reduces latency for users in different regions and offloads traffic from the application servers.

To streamline development and operations, Sufle helped implement a robust CI/CD pipeline. Source code repositories were integrated with a pipeline that automatically builds, tests, and deploys new container images to ECS. Deployments use blue-green and rolling update strategies to achieve zero-downtime releases. Infrastructure is defined as code, so environments can be reproduced and changes are version-controlled. This modernization of the software lifecycle enables Vademecum to release updates weekly or even daily, in contrast to the infrequent manual releases before.

In summary, the new AWS architecture modernized Vademecum’s platform into a scalable, secure, and highly available cloud application. The use of managed services means Vademecum benefits from AWS’s built-in automation and resilience. As one reference architecture notes, deploying a web application with containers on AWS in this manner ensures high availability, security, and scalability while leveraging fully managed services. Vademecum’s engineering team can now focus on product improvements rather than managing servers, and they can trust that the underlying AWS infrastructure will seamlessly handle growth and provide a stable experience to healthcare users.

Generative AI Implementation with Amazon Bedrock and pgvector

A highlight of the transformation was adding Generative AI capabilities to the Vademecum platform, turning its rich drug database into an intelligent assistant for healthcare professionals. Vademecum and Sufle implemented a retrieval-augmented generation (RAG) solution using Amazon Bedrock and the pgvector extension for PostgreSQL. This allows users to ask complex natural-language questions and receive accurate, contextually relevant answers sourced from Vademecum’s vetted medical content.

Vademecum’s drug reference content, including monographs, interaction descriptions, and guidelines, is vectorized and stored in Aurora PostgreSQL using the pgvector extension. Each document or text segment is transformed into a numerical embedding that represents its semantic meaning, then indexed for fast similarity search. With HNSW indexing, Aurora can perform vector queries up to 20× faster than traditional methods—critical for real-time usage.

When a clinician enters a natural-language query—for example, “Can I prescribe ibuprofen together with lisinopril?”—the application converts it into a vector using the same embedding model. This vector is then compared against stored embeddings in Aurora, which retrieves the most relevant, trusted information from Vademecum’s curated medical database. Aurora Optimized Reads further speed retrieval, ensuring quick responses even with millions of stored embeddings.

The retrieved context is passed to a foundation model hosted on Amazon Bedrock, such as Amazon Titan or Anthropic Claude, via a simple API. The model uses this context to generate a concise, natural-language answer—for instance, warning that ibuprofen may reduce lisinopril’s effectiveness and advising blood pressure monitoring. Responses are delivered in the UI with citations to original data sources, ensuring transparency and trust. The application also logs all queries and results, combining the precision of database search with the fluency of AI-generated language.

This generative AI capability has been transformative. Now, a user can simply ask a question or describe a clinical scenario, instead of manually searching through indexes or tables. By adding generative AI, healthcare providers can use natural language to search records and verify medication safety, rather than writing complex queries or combing through data. For Vademecum’s users, this feels like conversing with a smart medical advisor that can instantly reference a vast repository of drug knowledge.

Results and Benefits

By migrating to AWS and deploying generative AI, Vademecum realized significant improvements across technical performance, user experience, and business outcomes. The solution exceeded the goals set out at the project’s start. Key results and benefits include:

Quantitative Improvements: (measurable gains in performance and reliability)

• Dramatically Reduced Latency: End-to-end response times for drug queries dropped by over 70%. Before migration, the average query (especially complex interaction checks) could take 2–3 seconds on legacy systems. Now, with AWS’s optimized infrastructure, most complex and heavy queries return in under 500 milliseconds. The introduction of vector search with Aurora PostgreSQL further accelerates semantic queries, Aurora’s pgvector index delivers similarity matches up to 20× faster than conventional approaches. This ensures clinicians get answers almost instantly, even for AI-driven queries.

• Higher Throughput & Scalability: The platform can handle 2× to 3× more concurrent users and transactions than before. In load tests, Vademecum observed the system comfortably sustaining bursts of traffic without performance degradation, thanks to ECS auto-scaling. Where the previous setup risked slowing down at peak (~4000 queries/minute), the AWS setup processes the same load at <50% utilization. This headroom means Vademecum can support growing demand (new hospitals, pharmacies) with confidence.

• Faster Deployment Cycle: With CI/CD and infrastructure automation, the time to release new features or updates went from weeks to hours. Vademecum can deploy updates roughly 10× more frequently than before (e.g., moving from quarterly releases to bi-weekly sprints). Critical drug database updates that might have taken a full day to apply (with downtime) are now released in minutes with zero downtime.

Qualitative Benefits: (impact on user experience and operations)

• Enhanced Clinical Decision-Making: Clinicians using Vademecum now retrieve drug information and interaction checks more quickly and easily than ever. The generative AI search interface allows them to ask questions in natural language, a far more intuitive experience than keyword lookup. This saves precious time during patient consultations. Doctors report that they can get concise answers to complex queries in seconds, helping them make informed decisions on the spot. The outcome is faster decision-making and potentially improved patient outcomes, as critical warnings or recommendations surface immediately when needed.

• Better Access to Knowledge (Semantic Search): The platform’s search functionality is vastly improved. Users can find relevant information even if they don’t know exact keywords, because the semantic search understands intent. For example, a pharmacist could search “ACE inhibitor with NSAID precautions” and the system will intelligently retrieve the content about lisinopril and ibuprofen interactions. This was practically impossible with the old system. Vademecum’s vast repository of drug data is now unlocked via conversational AI, making it feel like a true digital assistant for healthcare professionals.

• Security & Compliance Confidence: The move to AWS has strengthened Vademecum’s security posture. All data is encrypted at rest and in transit, and fine-grained access controls are in place. Regular security assessments and AWS’s built-in protections (like GuardDuty alerts, automated patching of managed services) mean the environment is continuously protected and improved. Vademecum’s compliance team can more easily generate reports and audits now, thanks to AWS’s certifications and the detailed logging of every access. The healthcare institutions and government partners that integrate with Vademecum have expressed increased confidence knowing the platform runs on AWS, which is known for its stringent security standards in healthcare. In essence, Vademecum can demonstrate compliance and security with greater ease, leveraging AWS’s reputation and toolset.

In the words of Vademecum’s CTO, Güçlü Aydoğan this transformation “breathed new life into our platform”, enabling the company to fulfill its mission more effectively. Clinicians and pharmacists get timely, AI-driven insights and Vademecum as a business is stronger, more competitive, and poised for future innovation. The project demonstrates how adopting AWS cloud and generative AI can yield not just technical improvements, but tangible business benefits in the healthcare domain.

Lessons Learned and Next Steps

Vademecum’s AWS migration reinforced the value of starting with a strong foundation. By applying AWS Well-Architected principles from day one—multi-AZ resilience, infrastructure-as-code, and CI/CD pipelines—the team avoided rework and ensured performance, reliability, and cost control. Partnering with Sufle proved equally important: their AWS and DevOps expertise accelerated delivery while mentoring Vademecum’s engineers for long-term self-sufficiency. The project also highlighted that compliance and security in healthcare must be continuous, with guardrails, and audits integrated into daily operations for HIPAA/GDPR alignment.

Generative AI adoption emphasized the need for curated, high-quality data and prompt tuning to ensure accurate, safe outputs. Vademecum invested in embedding preparation, context design, and human-in-the-loop reviews by pharmacists to maintain trust. The move to AWS also drove a DevOps culture shift—breaking down silos, automating workflows, and adopting shared toolsets—which improved collaboration, release speed, and system stability. The team learned that technology transformation is most successful when paired with cultural change.

Vademecum’s transformation is an ongoing journey. The next steps above underscore a commitment to continuous improvement, leveraging the AWS platform not just as an infrastructure host but as a catalyst for ongoing innovation in healthcare. By partnering with AWS and Sufle, Vademecum has created a cloud-first, AI-enabled culture within the company. This paves the way for it to remain a leader in providing intelligent healthcare information services. In the words of AWS, “leading healthcare organizations trust AWS when scaling their generative AI because of our proven security commitment and industry expertise.” Vademecum’s experience exemplifies this, and the company is excited to continue its collaboration with AWS to push the boundaries of what’s possible, from improving day-to-day patient care to shaping the future of digital health with AI.