Heidi’s AI Scribe Scales to 81M Clinical Consultations With MongoDB

Clinicians spend up to 40% of their working hours on documentation tasks. Heidi, a rapidly growing Australian-based global AI startup, is on a mission to protect and extend the human touch in healthcare. In just 18 months, Heidi has returned more than 18 million hours to frontline clinicians by streamlining critical administrative tasks.

Heidi’s flagship product, Heidi Scribe, is an AI Care Partner that expands clinical capacity by automating administrative work—documentation, form filling, and task management—so clinicians can focus on patients. Used across emergency departments, general practice, and specialist clinics in over 190 countries, Heidi supports more than 2.3 million consults each week globally.

MongoDB Atlas is the data layer at the core of Heidi’s offering. MongoDB's flexible document model and globally distributed architecture, coupled with the power of MongoDB Atlas Vector Search, enabled Heidi to scale to 81 million medical consultations globally in just 18 months, and keep on building and adding new, innovative AI use cases to its suite of products. 

Building a secure, accurate, and efficient AI-driven healthcare platform presented three  complex data challenges:

• Heidi handles a diverse set of medical data collected from multiple sources,  such as medical forms, referrals, clinicians’ notes, and more. To connect seamlessly with AI workflows, it was essential to consolidate that data into one consistent format and one single location. 

• Operating in a dynamic, fast-changing environment, Heidi needed to adapt to constantly evolving data types, shifting data modeling requirements, and the regular addition of new use cases to their application. 

• Processing medical data meant the company had to comply with stringent security and regulatory requirements. 

These requirements ruled out building with a traditional relational database. Relying on rigid rows and columns, a relational database would have been ill-suited to handling the dynamic, varied data essential for AI applications. Heidi needed a database that was built for the AI era.

The document model: A perfect fit for AI and healthcare

From day one, Heidi built Heidi Scribe on a JSON document database. The team knew it would deliver an intuitive data model for AI workloads that empowered developers to work with data rapidly and easily. The document database’s flexible schema also enabled the data model to evolve in lockstep as application needs changed. 

Heidi had built the initial capabilities on AWS's Amazon DocumentDB. However, the team ran into limitations as the business and requirements grew in size and complexity. 

“We couldn’t scale without downtime, which was a critical issue: we operate in the world of healthcare where clinicians need seamless access to resources 24/7,” said Oscar Lukersmith, Head of Data at Heidi. “Our initial database set-up couldn't accommodate the level of growth that our users needed, it didn’t support search and index building functionalities—which are key in AI use cases—and we were experiencing increased latency.”

To address these challenges and accelerate the growth of its AI-driven healthcare platform, Heidi turned to MongoDB. MongoDB Atlas stood out because it combined the power of the document model—which allows seamless scale, flexibility, and high performance—with built-in AI-ready features such as MongoDB Atlas Vector Search.

With MongoDB Atlas Vector Search, Heidi does not need another ‘bolt-on’ vector database to augment its existing platform. It can perform RAG (retrieval-augmented generation) and retrieve relevant, up-to-date information from external sources directly into MongoDB Atlas. This simplifies the architecture and eliminates the need to migrate, synchronise, or duplicate data. Furthermore, building hybrid search capabilities with MongoDB Atlas means that Heidi’s team is able to streamline vector, full-text, and semantic search under one single API. 

Building AI tools to double the world’s healthcare capacity

Heidi has since built a wide range of AI tools and features, all on MongoDB Atlas, empowering clinicians around the world to save time on a range of administrative tasks— time that they can instead use to prioritise patient care. 

One example of this is clinical coding. Built on MongoDB Atlas, clinic coding is a core part of Heidi’s AI Scribe. The feature uses RAG to automate the process of translating patient health records into standardised alphanumeric codes. It does this using a health classification system used by hospitals, health service planning, health insurance companies, and others. 

For example, in Australia, the code K35.8 is assigned by hospitals for an acute appendicitis episode based on a patient's clinical documentation. Working with MongoDB, Heidi is able to accurately and efficiently capture clinical codes across multiple medical systems. This enables accurate billing, reimbursement, and documentation for public and private patients. 

Ask Heidi is another AI tool in Heidi’s suite of services, powered by MongoDB Atlas. It assists clinicians by streamlining non-clinical tasks such as collecting patient histories, creating ward round lists, and conducting clinical audits. Ask Heidi can save up to 50% of a clinician’s time. 

"MongoDB’s document model has been a game-changer for our developers. Its flexibility enables us to quickly adapt our AI applications to new use cases—helping us scale to more than 2 million consultations per week, without downtime or bottlenecks,” explained Ocha Cakramurti, Senior Software Engineer at Heidi “Since migrating to MongoDB Atlas, we've been able to reduce latency on key APIs by nearly one-third, ensuring seamless experiences for medical professionals in critical environments.” 

Another critical AI use case powered by MongoDB Atlas Vector Search is Heidi Vector Scribe. Built on Heidi Scribe's capabilities, Heidi Vector Scribe introduces advanced AI to medical information retrieval. It converts a huge volume of medical documents into vector embeddings, enabling semantic search to intelligently connect transcribed medical terms—extracted and embedded via Langchain in Atlas—directly to corresponding external knowledge.

Beyond powering Heidi’s core AI tools and features, MongoDB is also empowering the company on its journey to retain and recruit the top tech talent who are critical to the success of Heidi. 

“For tech talent, exciting technology matters—the scalability and efficiency of MongoDB Atlas make it a cornerstone of our success, helping us attract developers who want to spend time transforming healthcare, rather than managing databases,” said Cakramurti. 

Thanks to MongoDB, Heidi is on its way to deliver on its ambitious goal of doubling the world’s healthcare capacity.  Next, the company is exploring the application of MongoDB, LLMs, and Heidi's unique AI tooling to build an ecosystem for clinical activities, with Heidi Scribe acting as an agentic AI layer.