HIPAA-aligned AI documentation layer for a multi-tenant EHR platform
A four-week proof of concept progressed to a production rollout plan, supported by an itemised production cost model in which AI and ML services represented the substantial majority of the workload.
Customer background
A US-based, multi-tenant electronic health record and practice-management platform serving independent practices and digital health organisations. The platform handles protected health information across many tenants, each with its own data boundary and compliance obligations.
Challenge
Clinicians using the platform were spending a significant proportion of each day on documentation. The vendor required an AI capability, but multi-tenancy and PHI handling ruled out a general-purpose assistant.
Solution
Rabbit Cloud designed an encounter-to-note pipeline on Amazon Bedrock with Amazon Transcribe and Comprehend Medical, tenant-scoped retrieval on Amazon RDS for PostgreSQL with pgvector, and per-tenant AWS KMS keys. The cost model was established first so that the feature could be priced before development.
Solution architecture
The engagement was delivered on the Clinical & Contact-Centre Voice Intelligence reference architecture, adapted to the client's data, security and operating requirements.
Reference architecture
Results and benefits
A four-week proof of concept progressed to a production rollout plan, supported by an itemised production cost model in which AI and ML services represented the substantial majority of the workload.
- Encounter-to-note pipeline delivered as a four-week proof of concept with a defined production rollout path
- Per-tenant data isolation and AWS KMS key management, supporting HIPAA obligations across the customer base
- Cost per encounter modelled before development, enabling the feature to be priced for the platform's customers
- AI and ML services accounting for the substantial majority of the workload, with a documented architecture for production scale
AWS services used
About Rabbit Cloud
Rabbit Cloud is an AWS partner delivering cloud migration, platform engineering, managed operations, data platforms and AI/ML solutions for enterprises and software companies. Engagements are delivered by senior engineering teams to the AWS Well-Architected Framework, with security, compliance and cost governance addressed from the initial architecture.
Client identity withheld under a confidentiality agreement; identifying detail has been generalised.
Explore related engagements
Twelve engagements across software, construction technology, healthcare, payments, banking, logistics and retail.