The Opportunity: Bringing a Connected Care Platform to Production
Evergreen envisioned a FHIR-first and agentic-AI-first telehealth platform integrating services spread across separate systems. Starting with conceptual specifications, Evergreen and KMS needed to bring that architecture into production, with few established patterns to follow and multiple third-party integrations to build and test.
1. Novel Architecture with Few Established Patterns
The platform required a FHIR-first data design and production AI agents, with few established patterns to follow. Evergreen and KMS needed to turn theoretical specifications into a comprehensive, production-grade system.
2. Deep Integration Across Third-Party Systems
The platform needed to integrate an EHR, identity verification, payments, and pharmacy management. Each connection had to be built and integration-tested to bring these disparate systems together.
3. Clinical Data Moving Across Multiple Systems
SOAP notes and other clinical information needed to move from Grove into Canvas and onward to the clinical portal. The integration required correct data mapping, appropriate storage, and consistency across the workflow.
4. Three Applications to Build and Validate
The launch covered web, iOS, and Android applications. Requirements, design, development, and testing needed to progress together, while accommodating additional scope and maintaining human oversight throughout delivery.
The Approach: Connecting Care Through AI-Native Engineering
Evergreen’s platform runs on AI agents, and the team used AI throughout the software development lifecycle to accelerate delivery. Claude, running through Amazon Bedrock, supported work from requirements to testing. KMS engineers led the delivery, with strong human-in-the-loop oversight and an emphasis on clean, thorough documentation.
Define: Turn Concepts into Build-Ready Requirements
Claude drafted user stories and ran requirement gap analysis. When Evergreen added scope, the team turned new requirements into build-ready stories and caught gaps before development. Claude also generated UI/UX concepts, giving Evergreen options to react to early instead of waiting on full design cycles.
Key deliverables included:
User stories drafted with Claude
Requirements reviewed for gaps
Early UI/UX concepts for feedback
Added scope translated into build-ready stories
Connect: Integrate Clinical and Commercial Workflows
KMS turned FHIR-first principles into working integrations. FHIR R4 supported the movement of SOAP notes and other clinical data from Grove into Canvas and onward to the clinical portal. Alongside this clinical workflow, the platform connected third-party services for identity verification, payment processing, and pharmacy management.
Key deliverables included:
FHIR R4 clinical data integration
Clinical data mapping across systems
Identity verification and payment integrations
Pharmacy management integration
Build: Deliver Web, Mobile, and Agent Capabilities
Engineers used Claude Code for code generation and code reviews across web, iOS, and Android. Claude also supported infrastructure provisioning. The core V1 launched with three deployable applications and two agents covering clinical protocols and patient engagement, delivered through a human-led, AI-accelerated model.
Key deliverables included:
Web, iOS, and Android applications
Two V1 agents supporting care workflows
AI-assisted code generation and reviews
Infrastructure provisioning supported by Claude
Validate: Keep Testing Aligned with Development
With three applications to validate, testing ran alongside development. Reusable AI skills generated test cases, prepared test data, and ran automated tests, alongside Claude-written unit tests. This approach helped testing keep pace with the build, while the delivery model retained strong human-in-the-loop oversight.
Key deliverables included:
AI-generated test cases and test data
Automated tests using reusable AI skills
Unit tests written with Claude
One of the most important parts of this project was making sure clinical information could move correctly from Grove into Canvas and then be presented in a usable way within the Provider Portal. That meant more than simply connecting systems. The data had to be mapped properly, stored in the right place, and remain consistent across the workflow. Building that FHIR-based integration gave Evergreen a much stronger foundation for reliable provider workflows and future automation.
Murugalingam Poornachandran
Sr. Delivery Director, KMS Technology
The Outcome: A Connected Care Platform, Ready for Further Innovation
Working with Evergreen’s engineering team, KMS delivered the core V1 platform from concept to production in 75 working days, combining FHIR-first interoperability, AI agents, and web, iOS, and Android applications through human-led, AI-accelerated engineering.
What Was Built
Three Production Applications: Core V1 in production in 75 working days, with web, iOS, and Android applications, four live third-party integrations, and two V1 agents covering clinical protocols and patient engagement.
Connected Care Platform: One connected system for care, with patients, providers, and fulfillment partners working from shared, interoperable data.
FHIR R4 Clinical Integration: FHIR R4 integration moving SOAP notes and other clinical data from Grove into Canvas and on to the Clinical Portal.
Clinical Data Foundation: Connected clinical data in production, enabling Evergreen to pursue real-time analysis of patient and clinical outcomes.
Interoperability-Ready Architecture: A foundation built on the CMS interoperability standard, allowing Evergreen to adapt as rules evolve.
Expanded Platform Capabilities: Subsequent expansion by Evergreen’s engineering team into support, its core transaction platform, and machine learning analytics.
Agent Orchestration: An agent catalog developed by Evergreen to avoid duplication, with super agents monitoring the system and humans observing, evaluating, and remediating.
Our platform is AI in action from both sides. It runs on AI agents that we’re building our technical moat around, and we used AI throughout our SDLC to accelerate delivery, while our experienced engineers applied their subject-matter expertise against well-defined requirements to build a superior product. KMS helped us achieve this innovation with a great dev team, a super strong focus on clean and excellent documentation, and a no-nonsense approach to getting the job done.
Greg Robinson
CTO, Evergreen Telehealth
What This Proves
Evergreen moved from conceptual specifications for a FHIR-first, agentic-AI-first telehealth platform to a connected care platform in production.
The engagement brought disparate clinical and commercial systems together, including an EHR, identity verification, payments, and pharmacy management. The resulting platform supports shared, interoperable data across patients, providers, and fulfillment partners, while providing a foundation for further innovation.
Evergreen’s engineering team subsequently expanded the platform into support, its core transaction platform, and machine learning analytics.