What challenges did Audi face before AI implementation?

Manual cloud security, cost optimization, IaC retrieval, and architecture guidance work was slow and error-prone. Audi needed a robust, scalable, low-latency solution that could integrate with existing systems and comply with internal security and policy requirements.

What was the impact of the AI solution?

Manual IaC building block retrieval that took around 10 minutes was reduced substantially. Security findings remediation time and cost-report generation time were reduced. Audi reported up to 76 minutes per day saved across a typical mix of tasks.

Source

ExpandedEvidence: Low35/100

Audi & Storm Reply: GenAI multi-agent devbot for cloud security, cost optimization, and IaC on Amazon Bedrock

Use case typeCode assistantUpdated Jun 13, 2026

AUDI AG, in partnership with Storm Reply, built a production-ready generative AI devbot for its Cloud Foundation Services team. The solution supports internal cloud management by helping with security findings remediation, cost optimization analysis, validated infrastructure-as-code building block retrieval, and AWS architecture guidance. It uses a serverless multi-agent architecture with Amazon Bedrock, Bedrock Knowledge Bases, Bedrock Guardrails, AWS Security Hub, AWS Cost Explorer, Amazon OpenSearch Service, AWS Lambda, Amazon S3, and AWS Well-Architected guidance.

Organization: Audi
Industry: Automotive
Location: Germany
Published: May 2025

Reported outcomes

Daily time saved: Up to 76 minutes per day

Time & speed

Manual IaC retrieval time: Approximately 10 minutes

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

Normalized claim

Well-Architected documentation lookup time reduction: 99% decrease

AWS BlogMay 13, 2025Blog postExplicit claimLow evidence strength

resulting in a direct time saving of nearly 99%.

Normalized claim

Daily time saved: 76 minutes per day increase

AWS BlogMay 13, 2025Blog postExplicit claimLow evidence strength

they can save up to 76 minutes per day.

Normalized claim

Manual IaC retrieval time: 10 minutes decrease

AWS BlogMay 13, 2025Blog postExplicit claimLow evidence strength

manually searching for pre-validated IaC can take around 10 minutes.

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

Customer: Audi
Provider: AWS
Maturity: Unknown
Linked source: AWS Blog

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 4

1Agentic AI
2Internal Developer Productivity
3Cloud Operations

Built a production-ready serverless multi-agent architecture with specialized agents for Security Findings, Cost Optimization, validated IaC Building Blocks, and architecture guidance.
Used Amazon Bedrock Knowledge Bases with Amazon OpenSearch Service vector search for RAG-grounded responses, Amazon Bedrock Guardrails for policy compliance, and AWS Lambda-based orchestration to route requests to the right agent.
Integrated AWS Security Hub and AWS Cost Explorer as data sources, stored validated code blocks and conversational memory in Amazon S3 and OpenSearch, and added multilingual support plus reranking for retrieval quality.

Technologies

Amazon Bedrock Amazon Bedrock Knowledge Bases Amazon Bedrock Guardrails AWS Security Hub AWS Cost Explorer Amazon OpenSearch Service AWS Lambda Amazon S3 AWS Well-Architected Framework AWS Control Tower

Manual IaC building block retrieval that took around 10 minutes was reduced substantially.
Security findings remediation time and cost-report generation time were reduced.
Audi reported up to 76 minutes per day saved across a typical mix of tasks.

Architecture

A production-ready serverless multi-agent architecture routes user requests through a custom AWS Lambda orchestrator to specialized agents. Security and cost agents query AWS Security Hub and AWS Cost Explorer. The architecture agent uses Amazon Bedrock Knowledge Bases with an Amazon S3 source adapter and Amazon OpenSearch Service vector store for retrieval-augmented generation grounded in AWS documentation. The solution includes Amazon Bedrock Guardrails, multilingual streaming responses, conversational memory in OpenSearch, validated IaC retrieval, and scalable modular agent expansion.

Implementation partners1

Storm Reply

Sources & evidence1

Evidence: Low35/100Evidence strength