Normalized claim
Information search time: 95% decrease
reduced information search time by 95%
New Relic transforms productivity with generative AI on AWS (New Relic NOVA)
Use case typeKnowledge managementUpdated Feb 9, 2026
New Relic built New Relic NOVA, an enterprise AI assistant, to help engineers and employees find answers across fragmented internal documentation and systems. The solution uses Amazon Bedrock, Amazon Bedrock Knowledge Bases, Amazon Kendra, Amazon Q Index, Amazon S3, Amazon Nova Lite, Amazon Nova Pro, Strands Agents SDK, and MCP to support knowledge retrieval and transactional workflows. It integrates with Confluence, GitHub, Salesforce, Jira, Slack, and internal systems, and includes PII detection, masking, and an LLM-as-a-judge evaluation framework.
- Organization
- New Relic
- Industry
- Tech & Comms
- Location
- United States
- Published
- February 2026
Reported outcomes
Information search time: −95%
Time & speed
Daily queries processed: More than 1,000 queries/day
Catalog median for time & speed deployments: −50% across 295 reported metrics. Compare benchmarks →
Why do we believe this?Outcome claims, sources, and evidence checks
Normalized claim
Daily queries processed: 1,000 queries/day increase
processes over 1,000 daily queries
Normalized claim
Response accuracy: 80% increase
maintaining 80% accuracy in its responses
Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks
- Customer
- New Relic
- Provider
- AWS
- Maturity
- Production
- Linked source
- AWS Machine Learning Blog
Engineers spent significant time searching fragmented documentation and systems for knowledge and operational questions, sometimes taking more than a day
Customer identity supportedSource describes one deploymentMaturity supported
Primary read
Use case focus
Showing 3 of 3
- 1Knowledge management
- 2Workflow automation
- 3Agent orchestration
- Built New Relic NOVA as a multi-layer AI assistant with a central orchestrator that routes requests to specialized agents.
- Used Amazon Bedrock Knowledge Bases for Confluence, Amazon Kendra for GitHub Enterprise, Amazon Q Index for Slack history, and Strands Agents for Salesforce and Jira workflows.
- Added custom document enrichment, PII detection and masking, and an LLM-as-a-judge evaluation framework to improve retrieval quality and latency.
- Used Amazon Nova Lite and Amazon Nova Pro through Amazon Bedrock to balance response quality, latency, and cost efficiency.
- Reduced information search time by 95%.
- Processes over 1,000 daily queries.
- Maintained 80% accuracy in responses.
- Kept responses under 20 seconds.
- Automated complex operational workflows such as permission requests and rate-limit management.
Architecture
A central orchestration layer routes user intents to retrieval and action agents. The retrieval layer uses Amazon Bedrock Knowledge Bases, Amazon Kendra, and Amazon Q Index over multiple internal data sources, while the action layer uses Strands Agents and MCP for transactional workflows. Security controls include PII detection and masking, and performance is tuned with an LLM-as-a-judge evaluation framework.
Sources & evidence1
Evidence: Medium50/100Evidence strength
- Customer explicitly identified
- Deployment status explicitly supported
- Quantified outcome available
- Technical implementation details available
Type: Blog PostPublished: Feb 9, 2026Publisher: AWSEvidence: VendorConfidence: Medium
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Customer evidence
Provider evidence
AI patterns
Browse the catalog
Was this useful?
Community
Comments
No published comments yet.