Incident management · Production

Formula 1 uses Amazon Bedrock generative AI to reduce race-day issue resolution time by 86%

The problem

F1 IT engineers faced critical operational issues during live race events that could take up to 3 weeks to triage, test, and resolve, requiring coordination across development, operations, infrastructure, and networking teams. A recurring web API issue alone consumed 15 full engineer days across multiple events.

Workflow diagram · grounded in source
1
Engineer submits natural language query
trigger
“Users can ask the RCA chat-based assistant questions using natural language prompts”
2
ETL pipeline transforms log data
integration
“raw logs were centralized into an Amazon Simple Storage Service (Amazon S3) bucket. An Amazon EventBridge schedule checked this bucket hourly for new files and triggered log transformation extract, transform, and load (ETL) pipelines bui…”
3
Knowledge base ingestion via RAG
ai_action
“Amazon Bedrock Knowledge Bases, an end-to-end managed Retrieval Augmented Generation (RAG) workflow capability, allowing the chat assistant to query them efficiently”
4
Agent queries internal and external systems
ai_action
“Amazon Bedrock Agents facilitates interaction with internal systems such as databases and Amazon Elastic Compute Cloud (Amazon EC2) instances and external systems such as Jira and Datadog”
5
Claude 3 generates RCA response
ai_action
“Anthropic's Claude 3 Sonnet model was selected for informative and comprehensive answers and the ability to understand diversified questions”
6
Route issue to correct team
routing
“it also routes the issue to the correct team to resolve, allowing teams to focus on other high-priority tasks”
7
Escalate challenging issues
human_review
“particularly challenging issues are automatically escalated to the F1 engineering team for investigation, allowing engineers to better prioritize their tasks”
Reported outcome

The RCA assistant reduced end-to-end resolution time by as much as 86%, cut initial triage time from more than a day to less than 20 minutes, and enabled engineers to receive query responses within 5–10 seconds.

Reported metrics
End-to-end resolution time reduction86%
Initial triage timefrom more than a day to less than 20 minutes
Query response time5–10 seconds
engineer days to resolve recurring issue (before RCA tool)15 full engineer days
Show all 6 reported metrics
end-to-end resolution time reduction86%
initial triage timefrom more than a day to less than 20 minutes
query response time5–10 seconds
engineer days to resolve recurring issue (before RCA tool)15 full engineer days
time to resolve with RCA tool3 days
previous critical issue resolution timeup to 3 weeks
Reported stack
Amazon BedrockAmazon Bedrock AgentsAmazon Bedrock Knowledge BasesAmazon CloudWatchAWS GlueApache SparkAmazon S3Amazon EventBridgeAmazon EC2AWS FargateAmazon ECSClaude 3 SonnetStreamlitJiraDatadog
Source
https://aws.amazon.com/blogs/machine-learning/how-formula-1-uses-generative-ai-to-accelerate-race-day-issue-resolution?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The RCA assistant reduced end-to-end resolution time by as much as 86%, cut initial triage time from more than a day to less than 20 minutes, and enabled engineers to receive query responses within 5–10 seconds.

What tools did this team use?

Amazon Bedrock, Amazon Bedrock Agents, Amazon Bedrock Knowledge Bases, Amazon CloudWatch, AWS Glue, Apache Spark, Amazon S3, Amazon EventBridge, Amazon EC2, AWS Fargate.

What results were reported?

End-to-end resolution time reduction: 86%; Initial triage time: from more than a day to less than 20 minutes; Query response time: 5–10 seconds; engineer days to resolve recurring issue (before RCA tool): 15 full engineer days (source-reported, not independently verified).

How is this incident management AI workflow structured?

Engineer submits natural language query → ETL pipeline transforms log data → Knowledge base ingestion via RAG → Agent queries internal and external systems → Claude 3 generates RCA response → Route issue to correct team → Escalate challenging issues.

Related incident management cases