Orion Health Saves 50 Staff Hours a Day with Generative AI Chatbot

Headquartered in New Zealand, Orion Health delivers healthcare software solutions to public health systems across the globe including UK, Canada, and Saudi Arabia. Its products are used by clinicians and citizens to manage health records, telehealth services, and more. As the company grew, internal teams struggled to access information scattered across six separate knowledge silos—ranging from technical documentation to past support cases.

Employees spent at least 15 minutes per silo each day searching for answers—with some losing over an hour daily. This fragmented process slowed resolution times, impacted SLAs, and reduced staff efficiency. “Instead of spending time supporting customers, staff were spending time looking for information,” says James Stainton, vice president of global operations at Orion Health.

Orion Health recognized the need to simplify and accelerate knowledge access to improve support response times and scale operational efficiency. The team set out to design a secure, intelligent solution to help employees retrieve accurate, context-rich insights from over half a million internal records.

After evaluating several approaches to improve knowledge access, Orion Health selected AWS for its modular architecture, support for generative AI workloads, and access to technical expertise. The team collaborated with the AWS Prototyping and Cloud Engineering (PACE) team to develop Oribot, an internal chatbot powered by Amazon Bedrock using retrieval-augmented generation (RAG).

The solution integrates a React front-end framework with AWS Lambda for orchestration, Amazon DynamoDB to manage session data, and a vector database to perform semantic search across more than half a million internal records. The entire system is hosted within Orion Health’s Amazon Virtual Private Cloud (Amazon VPC), enabling control over data access, compliance, and model configuration.

The AWS team worked alongside Orion Health to refine the architecture and implement secure data access policies. This collaboration supported Orion’s development process, with AWS providing input on prompt turning, retrieval workflows, and overall system integration. As a result, a functional prototype was launched within two months. “We couldn’t have delivered Oribot in just two months without AWS,” says Stainton. “Their guidance helped us move quickly and with confidence, without wasting time on trial and error.”

Deploying Oribot on AWS has helped Orion Health eliminate internal search delays and improve the speed and accuracy of support case resolution. Powered by generative AI, Oribot not only surfaces relevant information but also connects it across silos—helping staff discover known issues, previous resolutions, and related insights across product documentation and support history.  Employees can now retrieve information from more than 500,000 internal records in under a minute—enabling faster customer responses and freeing up time for higher-value tasks. “AWS helped us move fast with the right tools and support,” says Stainton.

The automation is expected to help Orion Health’s support team reclaim approximately 50 staff hours each day by reducing manual search effort—contributing to stronger SLA performance.  

Oribot has strengthened Orion Health’s internal governance by making it easier to locate misplaced sensitive data, eliminate duplication, and clean up outdated contents. Additionally, Orion Health found Amazon Bedrock to be roughly ten times more cost-effective at scale compared to other commercial chatbot platforms.

Building on these outcomes, Orion is exploring new use cases such as extracting product-level insights from aggregated support data and streamlining bid responses by dynamically referencing past proposals. “Once we proved the potential savings and impact on customer experience, we couldn't wait to get it out,” says Stainton. “We’re already seeing the benefits with internal pilot use and plan to extend them to customers as well.”