Incident management · Production

BMW's generative AI solution for cloud incident root cause analysis using Amazon Bedrock Agents

The problem

BMW's root cause analysis for cloud incidents was cumbersome and time-consuming, requiring engineers to manually check many interdependent systems, iteratively form and reassess hypotheses, and track down culprits across geographically dispersed teams and components.

Workflow diagram · grounded in source
1
Incident described to agent
trigger
“an on-call engineer gives a description of the issue at hand to the Amazon Bedrock agent”
2
Architecture Tool maps system
ai_action
“The Architecture Tool uses C4 diagrams to provide a comprehensive view of the system's architecture. These diagrams, enhanced through Structurizr, give the agent a hierarchical understanding of component relationships, dependencies, and …”
3
Logs, metrics, and infrastructure evidence gathered
ai_action
“it scans the logs, metrics, and control plane activities of only those components that are involved in remotely unlocking car doors”
4
Hypotheses ranked and presented
output
“the agent infers and defines several hypotheses and presents these to the user, ordered by their likelihood”
5
Engineer reviews and directs
human_review
“The engineer can then resolve the issue, or give pointers to the agent to direct the investigation further”
Reported outcome

The Amazon Bedrock ReAct agent correctly identifies root causes in 85% of cases, reduced diagnosis time from hours to minutes, and lowered the barrier for junior engineers to diagnose issues effectively.

Reported metrics
RCA agent accuracy85%
Diagnosis timesignificantly lower diagnosis times
Incident resolution process durationminutes, compared to the hours
Root cause identification rate in test cases85%
Reported stack
Amazon Bedrock AgentsAWS LambdaAmazon CloudWatchAWS CloudTrailCloudWatch Logs InsightsC4 diagramsStructurizr
Source
https://aws.amazon.com/blogs/machine-learning/innovating-at-speed-bmws-generative-ai-solution-for-cloud-incident-analysis?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The Amazon Bedrock ReAct agent correctly identifies root causes in 85% of cases, reduced diagnosis time from hours to minutes, and lowered the barrier for junior engineers to diagnose issues effectively.

What tools did this team use?

Amazon Bedrock Agents, AWS Lambda, Amazon CloudWatch, AWS CloudTrail, CloudWatch Logs Insights, C4 diagrams, Structurizr.

What results were reported?

RCA agent accuracy: 85%; Diagnosis time: significantly lower diagnosis times; Incident resolution process duration: minutes, compared to the hours; Root cause identification rate in test cases: 85% (source-reported, not independently verified).

How is this incident management AI workflow structured?

Incident described to agent → Architecture Tool maps system → Logs, metrics, and infrastructure evidence gathered → Hypotheses ranked and presented → Engineer reviews and directs.

Related incident management cases