Rackspace Careflow

Healthcare providers are under intense pressure to reduce clinician burnout, improve patient access, and modernize clinical workflows while staying within strict HIPAA and security requirements. Clinicians routinely spend 15+ hours per week on documentation and administrative tasks, and many organizations struggle to trial AI solutions safely, with the right guardrails and network isolation in place. Rackspace Careflow Agents is designed to close this gap by standing up a HIPAA-ready, agentic AI care coordination platform on AWS that automates encounter capture, documentation, and follow-up workflows.

Rackspace Careflow Agents is a Rackspace professional services engagement that deploys the Rackspace Careflow Agents: a modular, agentic AI platform that unifies multiple clinical experiences (Ambient Listening for doctor–patient visits, AI Nurse Follow-ups, and future Care Coordination) into a single React-based Careflow Portal. The platform uses Amazon Bedrock with Amazon Nova Sonic for real-time speech-to-speech encounters, captures multi-turn conversations, generates medical summaries, and produces structured EMR-ready fields. AWS Step Functions and Amazon EventBridge orchestrate follow-ups and campaigns, while Amazon Cognito and fine-grained RBAC secure access across doctor, nurse, and clinic personas.

Working in an agile, six-week forward-deployed engineering (FDE) model, Rackspace AWS-certified architects and engineers design and implement a production-ready Careflow environment in your AWS account. The solution is built on serverless and managed AWS services, including AWS Lambda, Amazon API Gateway (WebSocket and HTTP), Amazon DynamoDB, Amazon S3, Amazon CloudFront, AWS WAF, AWS AppConfig, Amazon CloudWatch (Logs, Metrics, RUM), and AWS X-Ray, with private connectivity to Amazon Bedrock using AWS PrivateLink for PHI-bearing traffic.

Typical engagement activities include, but are not limited to:

• Assessing current clinical documentation, follow-up, and care coordination workflows, including EMR/FHIR integration points and HIPAA/security requirements.
• Designing a target Careflow architecture on AWS leveraging Amazon Bedrock (Nova Sonic for real-time S2S), AWS Lambda, Amazon API Gateway WebSocket/HTTP, Amazon DynamoDB, Amazon S3, AWS AppConfig, Amazon CloudWatch, Amazon HealthLake (optional), and Amazon Connect (optional for outbound calling). 
• Implementing the Careflow Portal (React SPA) with modular portals for Ambient/Clinic, Doctor, and Nurse roles, backed by Amazon Cognito User Pools with group-based RBAC and feature-flag enablement via AWS AppConfig.
• Building the real-time audio path from browser to Amazon Bedrock Realtime: Web Audio → AudioWorklet → Amazon API Gateway WebSocket → AWS Lambda broker → Amazon Bedrock Nova Sonic, including reconnect and heartbeat strategies for clinical-grade reliability.
• onfiguring agentic workflows for nurse campaigns, appointment reminders, and care coordination using AWS Step Functions and Amazon EventBridge, with persistence in Amazon DynamoDB and artifacts in Amazon S3.
• Hardening the deployment for HIPAA-aligned posture using VPC endpoints (PrivateLink) to Bedrock runtime, KMS-backed encryption at rest for S3 and logs, HTTPS-only S3 policies, AWS WAF at the edge, and CloudWatch Logs Data Protection for masking sensitive data.
• Instrumenting end-to-end observability with Amazon CloudWatch EMF metrics, AWS X-Ray traces, and SPA to monitor latency, error rates, session stability, and model usage.

Customer outcomes from the accelerator typically include:

• A HIPAA-ready, Amazon Bedrock–powered Careflow deployment in your AWS account, with ambient clinical documentation and nurse follow-up workflows running in a pilot clinic or specialty.
• Real-time speech-to-speech encounter capture with Amazon Nova Sonic and structured EMR-ready outputs for downstream charting, coding, and care coordination.
• Documented architecture, security model, and operational runbooks aligned to AWS best practices.
• An enablement session and handover for clinical, IT, and DevOps teams, plus a roadmap for scaling from pilot to multi-clinic deployment.