Say hello

Tell us about your next mission

A real live person will reach out to you within the next 48 hours.


Monographer, AI-Based Drug Information Retrieval

Transforming Routine Medical Queries with AI-Powered Precision

Monographer desktop interface showing a user interaction with the AI on the left and the referenced Product Monograph on the right

Methodologies used

Natural Language Processing, Text Extraction & Parsing, Machine Learning Models, Conversational AI, Transformer Models, Question Answering with Pre-Trained Models, Reference Management.

AI solution providing accurate drug related information in record time. Leveraging transformer and learning models, we rethought the process of searching within product monographs through the lens of conversational interfaces.


Develop an AI solution that could provide comprehensive and accurate drug information from product monographs. The objective was to enable healthcare practitioners and patients to quickly find reliable drug data, thus reducing the time spent on routine queries, and spend more time on diving deeper into a drug’s utility and value proposition in clinical practice.


Healthcare practitioners spend excessive time searching for accurate drug information within product monographs that are difficult to navigate. Routine queries were burdening Medical Affairs teams, leading to inefficiencies and manual effort in answering rote questions. Product Monographs, while a staple industry document; were uniquely and differently utilised based on therapeutic area, manufacturer, location of use.


We identified the critical need for speed and accuracy in accessing drug information.  Healthcare practitioners required a dependable resource that could provide precise data without delay, freeing them to focus on patients. There was also a significant challenge to ensure the AI did not "hallucinate" or provide incorrect information, in order to meet the highest regulatory standards.

Designed for unique implementation

Wherever the mission is to deliver seamless access to comprehensive drug information, embedding our solution could drastically reduce the time spent on routine drug information queries. An effortless UI enhances the overall user experience and facilitates the adoption of the AI solution; with technical rigour focused on speed, accuracy, and reliability ensures trust is never lost. The solution is primed for implementation into any digital channel patronised by healthcare practitioners, with customization capabilities that allow pre-engineered additional prompts & contextual factors to be instilled into each unique instance.

Detail of the Monographer desktop interface showing a user interaction with the AI; the AI replies to the prompt with an answer backed with references to the PDF pages on which it found the relevant information.

Time to answer retrieval<10s

Expected implementation time<4 Weeks

Concept to Prototype2 Weeks

Traversing the technical challenges

The technical challenges included building a comprehensive database that included all current and relevant product monographs, developing an engine that could accurately identify and retrieve relevant content, training the engine to understand and respond to natural language queries of clinical nature, and ensuring the AI solution saves and cites the sources accurately to maintain reliability and trust. Our solution utilizes NLP technologies for understanding and processing human language, sophisticated models for summarization and question answering. By integrating PDF parsing libraries, we efficiently extract and index text and metadata from product monographs. Containerization guarantees consistent performance across different environments, and index and search capabilities enable fast, accurate searches through the content. The backend infrastructure ensures robust data processing and storage, while the frontend provides an intuitive conversational user interface.

Methodologies used

Index & Search, Text Extraction, Conversational UI, Reference Management, Content Summarization, Text Parsing & Libraries, Transformer Models.

Creatively reframing the AI Hallucination problem

Delivering precise line-specific references directly from the product monographs solves a complex technical challenge by design.

Turning AI’s tendency to hallucinate into an opportunity.

We knew that our target audience would find references and citations a particularly valuable feature. To address this need, and the hallucination challenge; we implemented robust validation mechanisms and citation systems to ensure that the AI solution saved and cited sources accurately, maintaining reliability and trust. We also employed machine learning and deep learning techniques to train models capable of performing tasks such as summarization, question answering, and document parsing. Continuous refinement of the models enhanced accuracy and complex query handling capabilities.

Unlocking the future of access to medical information by taking on AI missions with real outcomes