How Natera designed an agentic AI system for lab ops with Amazon Bedrock AgentCore
Natera built SWARM, an agentic AI system on Amazon Bedrock AgentCore that captures institutional lab-operations knowledge and deploys it at scale. This post walks through the full architecture, from knowledge base construction to deployment and evaluation, delivering about 130x faster first response and 96.2% automated coverage across support tickets.
First seen by Oriel on .
AI summary
This AWS case study describes how Natera built SWARM, an agentic AI system on Amazon Bedrock AgentCore, achieving roughly 130x faster first responses and 96.2 percent automated support-ticket coverage.
- SWARM combines swarm-agent for autonomous Jira-ticket diagnosis with swarm-hive for interactive engineering support, sharing a library of 33 independently testable skills.
- The architecture routes tickets through Amazon EventBridge and Amazon SQS, uses DynamoDB for deduplication and run metadata, and stores reports in Amazon S3.
- Key security measures include keyless GitLab CI deployments through OIDC and AWS STS, least-privilege IAM, private VPC networking, rotating secrets, and audit logging.
- A custom document pipeline uses Amazon Titan Text Embeddings, Amazon Bedrock Knowledge Bases, 1,000-token chunks, 150-token overlap, and batches of 100 documents.
- Across 524 tickets, first response time fell from 17.4 hours to roughly 8 minutes, while High-confidence diagnoses reached 74 percent accuracy across 154 evaluated cases.
#agentic ai #amazon bedrock #lab operations #retrieval augmented generation
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| AWS Machine Learning Blog | 787b1b13cc7c9546be4afedb6d427a9b0c7e6990 | https://aws.amazon.com/blogs/machine-learning/how-natera-designed-an-agentic-ai-system-for-lab-ops-with-amazon-bedrock-agentcore/ Read discussion |