Achieved a Word Error Rate (WER) below 10 for complex medical dictations using Gemini models
Boosted medical language specialist productivity by 25% to 50%
Processed over 150,000 hours of voice recordings per month securely
DeliverHealth uses Gemini models for documentation and coding—freeing healthcare providers to focus on patients, not paperwork.
DeliverHealth uses Gemini models for documentation and coding—freeing healthcare providers to focus on patients, not paperwork.
Cutting out after-hours paperwork
with Gemini models
Doctors often have to spend time after their shifts catching up on paperwork, a backlog the industry calls ‘pajama time.’ Healthcare technology company DeliverHealth wants to solve this and let doctors focus on their patients and reduce time spent typing into a medical record system.
The company needed highly accurate AI to run its InstaNote (clinical documentation) and InstaCode (autonomous medical coding) products. It also needed a secure system to process 150,000 hours of medical audio a month, a difficult task made more complex as it involves multiple speakers, background noise, diverse accents, and heavy medical jargon. A single wrong word can ruin a patient’s chart. And any solution used had to be compliant with patient data protection laws.
So, DeliverHealth decided to expand its existing Google Cloud setup. By integrating Gemini models 2.5 Pro and 3.5 Flash, DeliverHealth optimized its ambient documentation and autonomous coding platforms.
We are providing a tool that allows clinicians to experience the patient-doctor conversation fully, which is key to the healthcare encounter. At the core, we are allowing doctors to be doctors.
Percy Bhathena
VP of Product and AI, DeliverHealth
The decision to use Google Cloud in this instance went beyond simply consuming an API. “We have a partnership with Google that has allowed us not just to be a consumer of a model that exists, but to be able to say, ‘How do we make this better for our use case?’,” says Percy Bhathena, VP of Product and AI at DeliverHealth.
Through this collaboration, DeliverHealth uses a single-cloud view to manage development pipelines, optimize costs, and maintain compliance. Google Cloud provides the elasticity needed to process massive data volumes securely while strictly adhering to necessary certifications.
Delivering higher accuracy
and autonomous workflows
Since switching to Gemini, DeliverHealth’s word error rate dropped below 10, meaning far less manual fixing. InstaNote accurately captures doctor-patient chats and turns them into organized notes. Then, InstaCode reviews those notes and produces billing codes with confidence scores. This setup lets medical language specialists focus only on the tricky cases—like figuring out the exact location of an injury—boosting their productivity by 25% to 50%.
DeliverHealth is now focused on using Google Cloud to automate complex insurance approvals and speed up its own software development so it can bring new features to market even faster and help doctors in what they do best—take care of patients.
My advice is that you should pick a partner who is going to lean in with you, and I can tell you from experience that the Google Cloud team is invested in our success.
Percy Bhathena
VP of Product and AI, DeliverHealth
DeliverHealth uses AI to reduce manual typing work for healthcare providers and enhance patient care through solutions such as ambient clinical documentation and autonomous coding tools.
Industry: Healthcare and Life Sciences
Location: India
Products: Gemini 2.5 Pro, Gemini 3.5 Flash
