Retail / Convenience and fuel retail
Circle K: Cloud incident investigation and response summaries for store digital services
- Company
- Circle K
- Country
- Multiple countries
- Adoption stage
- In operation
- Source published
- In use from
- Date basis
- The date the source was published. It can differ from the date adoption started.
- How the source was checked
- Read the full source text
The work problem
When alerts fired in the cloud supporting point-of-sale, loyalty and electric vehicle charging services, engineers had to gather logs, service health and code across multiple tools. Circle K says triage took 45 to 60 minutes before remediation began, and investigation depth varied with the engineer's familiarity with the service.
Technology and data
A Lambda function receiving a Grafana alert starts an Amazon Bedrock investigation. The initial Flows architecture linked multiple agents; it was later consolidated into one agent on AgentCore Runtime that gathers AWS resource status, CloudWatch and OpenSearch logs, GitHub code and change history, and service documents. Gateway controls access to these sources. Findings reach Slack as a summary containing the probable cause, evidence, responsible team and recommended remediation. Engineers perform the actual fix.
Results
Circle K put the AgentCore version into production in April 2026. The team's internal incident-response metrics show mean triage time dropping from 45 to 60 minutes to under 15 minutes, a reduction of 67 to 75%. Circle K says consistent evidence and next steps help engineers escalate to other teams and evaluate false alarms regardless of who is on call.
Limits and open questions
The reduction concerns triage time, not total service recovery time. AI execution of restarts, scaling and deployment rollbacks remains planned.
Sources
- How Circle K accelerated DevOps incident triage to under 15 minutes with Amazon BedrockAWS / Circle K, Accessed
Compiled from public sources. These are not results from ATF Works customers.