Quality assurance · Production

QyrusAI and Amazon Bedrock power shift-left testing with 80% reduction in defect leakage

The problem

Traditional testing methods occurring late in the development cycle often result in delays, increased costs, and compromised quality, making it difficult for businesses to maintain rigorous standards while accelerating development.

Workflow diagram · grounded in source
1
Upload requirements document
trigger
“A user uploads a sample requirements document.”
2
Generate high-level test cases
ai_action
“TestGenerator, powered by Meta's Llama 3.1, processes the document and generates a list of high-level test cases.”
3
Claude refines test coverage
validation
“These test cases are refined by Anthropic's Claude 3.5 Sonnet to enforce coverage of key business rules.”
4
Generate UI tests from designs
ai_action
“VisionNova and UXtract use design documents from tools like Figma to generate step-by-step UI tests, validating key user journeys.”
5
API Builder virtualizes APIs
output
“API Builder virtualizes APIs, allowing frontend developers to begin testing the UI with mock responses before the backend is ready.”
Reported outcome

Early adopters of QyrusAI saw an 80% reduction in defect leakage, a 20% reduction in UAT effort, and a 36% faster time to market.

Reported metrics
Defect leakage reduction80%
UAT effort reduction20%
Time to market improvement36% faster
Reported stack
QyrusAIAmazon BedrockMeta's Llama 70BAnthropic's Claude 3.5 SonnetCohere's English EmbedPineconeAmazon ECSAmazon S3Amazon EFSAWS LambdaClaude 3 OpusClaude 3 SonnetMeta's Llama 3.1Application Load BalancerFigma
Source
https://aws.amazon.com/blogs/machine-learning/the-future-of-quality-assurance-shift-left-testing-with-qyrusai-and-amazon-bedrock?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Early adopters of QyrusAI saw an 80% reduction in defect leakage, a 20% reduction in UAT effort, and a 36% faster time to market.

What tools did this team use?

QyrusAI, Amazon Bedrock, Meta's Llama 70B, Anthropic's Claude 3.5 Sonnet, Cohere's English Embed, Pinecone, Amazon ECS, Amazon S3, Amazon EFS, AWS Lambda.

What results were reported?

Defect leakage reduction: 80%; UAT effort reduction: 20%; Time to market improvement: 36% faster (source-reported, not independently verified).

How is this quality assurance AI workflow structured?

Upload requirements document → Generate high-level test cases → Claude refines test coverage → Generate UI tests from designs → API Builder virtualizes APIs.

Related quality assurance cases