SMART WORLDJOURNALSubmit Manuscript
← All technology news

Artificial Intelligence

Natera Deploys Agentic AI System SWARM via Amazon Bedrock AgentCore to Revolutionize Lab Operations

Natera has introduced SWARM, an AI-driven system leveraging Amazon Bedrock AgentCore to dramatically accelerate lab instrument diagnostics and support. SWARM achieves a 130x reduction in response times and automates 96.2% of ticket resolutions by capturing vital institutional knowledge in clinical lab environments.

NEWS

Natera, renowned for advancing cell-free DNA testing, faced the critical challenge of managing diverse lab instruments whose maintenance requires deep specialist knowledge. Departures of senior technicians often resulted in lost expertise, prolonging equipment downtime with an average 17.4-hour first response time to support tickets.

To address this, Natera collaborated with AWS to develop SWARM (Systematic Workflow Analytics and Resolutions Manager), an agentic AI system built on Amazon Bedrock AgentCore. SWARM integrates institutional knowledge across equipment families, accelerating diagnostics and enabling near-instantaneous ticket response.

Amazon Bedrock AgentCore provides a secure, managed runtime environment with native integrations to tools like Jira, Confluence, DynamoDB, and AWS X-Ray, reducing infrastructure overhead. This setup supports autonomous agent operation at scale, allowing Natera to focus on diagnostic precision rather than backend maintenance.

SWARM employs two complementary AI agents sharing a skill library: swarm-agent autonomously generates structured diagnostic reports instantaneously upon ticket creation, while swarm-hive offers engineers real-time interaction including image uploads and alternative troubleshooting pathways. This preserves engineer involvement while vastly expanding accessible expertise.

The system's architecture enforces stringent security with managed identity and access controls, private networking, and audit logging compliant with clinical regulations like HIPAA. SWARM processes support tickets through event-driven pipelines, ensuring high availability and observability via comprehensive metrics and tracing.

Through continuous evaluation across over 500 tickets, SWARM achieved a remarkable 130-fold faster first response—approximately 8 minutes—and 96.2% automated coverage, substantially reducing lab instrument downtime. The structured skill library facilitates versioned improvements, while knowledge base enhancements, particularly in vendor documentation, significantly boost diagnostic accuracy.

However, SWARM's accuracy is data-dependent with current high-confidence diagnoses achieving around 74% accuracy, indicating room for improvement. Data governance restricts patient-identifiable information, focusing solely on equipment metadata, which may limit diagnostic context. Moreover, SWARM complements but does not replace human expertise, preserving critical oversight.

Natera's experience highlights essential design practices for agentic AI in regulated environments: decoupling batch and interactive processing, embedding traceability in skill logic, tuning document chunking for semantic search, and rigorous continuous monitoring. Their approach offers a replicable model for AI-driven operational knowledge systems in healthcare and beyond.

Source & further reading

How Natera designed an agentic AI system for lab ops with Amazon Bedrock AgentCore ↗

Source published 2026-11-12. Claims and limitations should be checked against the original announcement or research.