Australian startup Heidi just processed 81 million medical consultations globally after ditching their database setup. Here’s why infrastructure shift matters for healthcare AI.
What’s happening: Melbourne-based AI healthcare startup Heidi has scaled its clinical documentation platform to process 81 million medical consultations in just 18 months.
Why this matters: Healthcare professionals spend up to 40% of their working hours on documentation tasks rather than direct patient care. As demand for AI-driven healthcare solutions accelerates globally, the technical infrastructure underneath must keep pace.
When Heidi’s engineering team hit a wall in late 2023, it wasn’t a product problem. The AI medical scribe worked. Clinicians loved it. But the database couldn’t keep up.
Heidi had built its initial platform on Amazon DocumentDB. It handled early growth fine. Then the company scaled. Usage exploded. Meetings with health systems multiplied. Suddenly, the infrastructure that worked yesterday couldn’t support today’s demands.
“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.”
The problem wasn’t just about numbers. It was about what healthcare needs. A hospital emergency department can’t wait for a database to come back online. A GP practice can’t tell patients “maybe next Monday we’ll have access to their records.” In healthcare, infrastructure isn’t a backend concern. It’s a patient safety issue.
Why rigid databases don’t work for AI
Here’s where database choice becomes strategic. Traditional relational databases, the kind that power banks and airline reservations, work brilliantly when data fits neatly into rows and columns. A customer record. A transaction. A clear, predictable structure.
Medical data doesn’t cooperate. A patient’s health record includes referrals, clinical notes, forms from three different hospitals, lab results in different formats, and context that shifts between specialties. When you add AI into the mix, the problem multiplies.
Heidi needed to consolidate medical data from multiple sources into one consistent format. It also needed flexibility. As the startup added new AI features, clinical coding, patient history generation, ward round summaries, the data model had to evolve without rebuilding the entire system. And it had to do all this while meeting stringent healthcare security and compliance requirements across 190 countries.
This ruled out rigid relational databases immediately. Heidi needed a database built for AI. The company turned to MongoDB Atlas.
MongoDB’s document model changed everything. Instead of forcing data into tables with fixed columns, it works with JSON documents that can vary in structure. Medical data from referral systems, emergency departments, and specialist clinics could all flow into the same database in their natural format, then be transformed for AI workflows.
“MongoDB’s document model has been a game-changer for our developers,” explained Ocha Cakramurti, Senior Software Engineer at Heidi. “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.”
What came next surprised even the engineering team. “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,” Cakramurti added.
The shift wasn’t just about scale. It was about speed. Developers could now prototype new features faster. Operations teams could deploy updates without downtime. The infrastructure actually helped Heidi move quicker, not slower.
The tools reshaping modern healthcare
One of Heidi’s most sophisticated use cases shows what the platform can now do: clinical coding.
Hospitals use standardised alphanumeric codes, the International Classification of Diseases, to classify patient conditions and procedures. In Australia, the code K35.8 represents acute appendicitis. These codes drive billing, reimbursement, and health system planning. But generating them manually takes hours. A clinician sees a patient, writes a note, then someone has to read it, interpret the clinical details, and assign the correct code.
Heidi automated this. Using retrieval-augmented generation (RAG), the AI scribe reads a patient’s clinical notes, understands the medical context, and assigns the correct classification code automatically. MongoDB Atlas Vector Search made this possible. Rather than building separate vector databases for AI search and keeping a separate operational database, everything runs in MongoDB. Medical documents get converted to vector embeddings. The system searches semantically across them. Relevant information flows directly into the AI model. One architecture. No data duplication. No synchronisation headaches.
“Heidi is a great example of how choosing a flexible, multi-cloud data platform with embedded AI capabilities can empower developers to quickly move from prototype to production, and then scale,” said Simon Eid, Senior Vice President, APAC, MongoDB.
Another tool in Heidi’s suite is Ask Heidi, an AI assistant that helps with non-clinical tasks. It collects patient histories, creates ward round lists, and conducts clinical audits. According to Heidi, it can save up to 50% of clinicians’ time on these tasks, time that flows back into patient care.
The most advanced feature, Heidi Vector Scribe, uses semantic search to convert massive volumes of medical documents into vector embeddings. Clinicians can ask complex medical questions, and the system intelligently retrieves relevant external knowledge, connecting transcribed medical terms directly to corresponding clinical guidelines and evidence.
Building at healthcare speed
For Heidi’s leadership, the infrastructure choice solved a business problem beyond just scaling: recruiting.
“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.
In a competitive market for software engineers, developers want to work on meaningful problems. They want to deploy without fear. They want infrastructure that doesn’t bog them down in operational complexity. Choosing MongoDB meant Heidi’s team could focus on AI innovation, not database administration.
The company has grown fast on this foundation. Heidi now supports clinicians across emergency departments, general practice, and specialist clinics. The platform operates in 110 languages across 116 countries. Major health networks use Heidi, from Modality Partnership in the UK to Beth Israel in Massachusetts. In Australia, health systems adopted Heidi rapidly enough that Telstra Health built its own clinical documentation tool, Telstra Scribe, on top of Heidi’s infrastructure.
That growth hasn’t slowed. In October 2025, Heidi raised $65 million in Series B funding led by Point72 Private Investments, valuing the company at $465 million. The round included existing investors Blackbird and Headline, bringing total funding to nearly $100 million. The company is now expanding to new markets in France, Spain, Ireland, Hong Kong, Germany, and Singapore.
What’s next? Heidi is exploring an ecosystem of AI tools for clinical work. The company describes Heidi Scribe as an “agentic AI layer”, not just automating specific tasks, but working more autonomously across clinical workflows.
For healthcare globally, that matters. Clinicians spend more than two hours every day on tasks other than patient care. They lose an average of $65,000 annually to administrative burden. As healthcare systems face staff shortages and rising demand, even small improvements in efficiency compound.
Heidi’s scale to 81 million consultations in 18 months shows what’s now possible when infrastructure keeps pace with ambition. The database decision was technical. But the outcome is human: clinicians spending less time typing, more time listening to their patients.
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.
