What was the impact of the AI solution?

Vital supports over 3 million patients annually. The system processes nearly 1 billion tokens daily. The company reports 94% of frequently missed incidental findings are now surfaced. The new approach delivers about 23x cost savings versus prior alternatives. QA evaluations have consistently achieved over 99% accuracy.

Source

ProductionEvidence: Medium65/100

Vital Surfaces 94 Percent of Missed Incidental Findings Using Amazon Nova and Amazon Bedrock

Use case typeHealthcare workflow automationUpdated Jun 13, 2026

Vital, a US healthcare company, uses Amazon Bedrock and Amazon Nova to run generative AI workflows that improve patient communication and help surface high-risk findings across clinical records. The company migrated from a more complex multi-cloud setup to AWS to reduce cost, improve scalability, and support HIPAA-aligned healthcare workloads.

Organization: Vital
Industry: Healthcare
Location: United States
Published: May 2026

Reported outcomes

23x

costCost savings

94%quantified impact99%accuracy

Strategic outcomes

Risk & complianceImplemented HIPAA-aligned patient communicationNew product / capabilityStructured clinical notes into taxonomyNew product / capabilitySurfaced high-risk clinical findingsCost efficiencyDelivered lower-cost AI workflows

Catalog median for cost savings deployments: −40% across 177 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Quantified impact: 94%

AWS Solutions Case StudyMay 13, 2026Case studyInferred claimMedium evidence strength

The company reports 94% of frequently missed incidental findings are now surfaced.

Normalized claim

Cost: 23 x decrease

AWS Solutions Case StudyMay 13, 2026Case studyInferred claimMedium evidence strength

The new approach delivers about 23x cost savings versus prior alternatives.

Normalized claim

Accuracy: 99%

AWS Solutions Case StudyMay 13, 2026Case studyInferred claimMedium evidence strength

QA evaluations have consistently achieved over 99% accuracy.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Vital
Provider: AWS
Maturity: Production
Linked source: AWS Solutions Case Study

Its prior generative AI stack had rising inference costs, limited model flexibility, and multi-cloud operational complexity

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 4

1Clinical document processing
2Patient communication
3Workflow automation

Vital needed a scalable, cost-effective way to parse unstructured clinical notes at high volume.
Its prior generative AI stack had rising inference costs, limited model flexibility, and multi-cloud operational complexity.
The company needed a secure approach for HIPAA-sensitive patient communication and clinical data extraction.

Vital migrated its generative AI workflows to Amazon Bedrock on AWS.
Amazon Nova Micro ingests free-text clinical notes and classifies them into a structured taxonomy.
The system extracts follow-up instructions, medication names, and high-risk findings such as precancerous lesions.
Vital uses a layered QA framework with an information quality model, a physician-validated safety model, and a secondary judge model before outputs reach patients.

Technologies

Amazon Bedrock Amazon Nova Amazon Nova Micro

Vital supports over 3 million patients annually.
The system processes nearly 1 billion tokens daily.
The company reports 94% of frequently missed incidental findings are now surfaced.
The new approach delivers about 23x cost savings versus prior alternatives.
QA evaluations have consistently achieved over 99% accuracy.

Sources & evidence1

Evidence: Medium65/100Evidence strength