Natera builds AI system that cuts lab equipment ticket response time 130x

Natera, a company specializing in cell-free DNA testing, built an agentic artificial intelligence system named SWARM to keep its lab equipment running. Equipment breakdowns often disrupt sample processing, and the company faced challenges when senior technicians left, taking years of diagnostic knowledge with them. Before the new system, the lab handled over 65 support tickets daily with an average first response time of 17.4 hours.

How SWARM Operates

The engineering team built SWARM using Amazon Bedrock AgentCore, a managed service designed to run artificial intelligence agents. The platform uses two main components: an autonomous agent that generates diagnostic reports before engineers open tickets, and an interactive tool that lets staff explore troubleshooting steps in real time. The architecture relies on an event layer to route tickets, a context layer that pulls from historical data and manuals, and an observability layer for tracking metrics.

Security and Results

Because the laboratory operates in a clinical environment, the system incorporates strict security measures, including a keyless deployment model using OpenID Connect and temporary credentials. The platform runs on services that comply with healthcare privacy regulations without exposing patient data. According to the company, SWARM achieved a response time roughly 130 times faster and automated 96.2 percent of support ticket coverage.

• System handles over 65 daily tickets across multiple equipment queues.
• Uses two runtimes sharing a single skill library for automated and interactive diagnostics.
• Operates under healthcare data regulations with audit logging and temporary security credentials.