AWS re:Invent 2025 - From Alert to Resolution: Supercharge AWS Ops with the Agentic AI SRE (AIM225)

AWS re:Invent 2025 - From Alert to Resolution: Supercharge AWS Ops with the Agentic AI SRE (AIM225)

Overview

• Presentation focused on how "Agentic AI SRE" can help solve problems in production environments by automating the process of investigating and resolving issues.
• Key goals are to reduce mean time to resolution (MTTR), decrease outages, and improve system availability and uptime.
• Agentic AI SRE, named "Hawkeye", can be deployed as a SaaS product or within a customer's VPC for security-conscious organizations.

Challenges Addressed

• Dynamic cloud environments with multiple layers, telemetry sources, and alerts lead to complex troubleshooting workflows.
• Manual processes of logging into various tools, analyzing dashboards, and coordinating teams are time-consuming and stressful.
• Finger-pointing and lack of clear root cause analysis often occur during incident response.

Agentic AI SRE Capabilities

• Automatically triggers investigations based on alerts from monitoring tools like DataDog, Prometheus, etc.
• Analyzes a wide range of data sources, including traces, logs, metrics, and configuration details, to determine root cause.
• Provides a detailed summary of the incident, including the description, timeline, evidence, root cause, and recommended corrective actions.
• Delivers the analysis and recommendations through various channels, such as Slack, Teams, ServiceNow, or directly within coding assistants like Cloud Code.

Business Impact

• Significant productivity gains for SRE teams by automating the investigation process.
• Reduced outages and improved system availability, leading to cost savings for customers.
• Examples of use cases:
  • Insurance company with complex on-premises and AWS footprint, focusing on reducing incident costs.
  • Shipping and logistics company managing data pipelines and nightly jobs, ensuring timely completion.
  • Bank in Australia using Agentic AI SRE for backup management and regulatory compliance.

Integration with Coding Assistants

• Agentic AI SRE integrates with coding assistants like Cloud Code, providing a self-documenting API (MCP server) for seamless interaction.
• SREs can use the coding assistant to onboard Agentic AI SRE, manage connections, initiate investigations, and automate remediation steps.
• Ability to summarize past investigations, identify patterns, and proactively address recurring issues.

Deployment and Pricing

• Self-service onboarding available through the AWS Marketplace, with a free trial and pay-as-you-go model.
• Enterprise customers can deploy Agentic AI SRE within their VPC, use their own language models, and integrate with custom observability and incident management tools.

Conclusion

• Agentic AI SRE, or "Hawkeye," aims to revolutionize incident response and problem-solving in production environments by automating the investigation process.
• The solution integrates with various monitoring and observability tools, providing a comprehensive and streamlined approach to root cause analysis and resolution.
• By empowering SRE teams with AI-driven insights and automation, Agentic AI SRE promises to deliver significant productivity gains, reduced outages, and improved system availability.