AWS re:Invent 2025 - Fidelity: AWS Health & Support Data Pipelines to GenAI Actions (SPS318)

Transforming Operational Challenges into Strategic Advantage with AWS Health, Support Data, and GenAI

Operational Challenges in the Cloud Era

• Organizations face growing complexities and large volumes of AWS accounts, resources, and events
• Reactive monitoring is no longer sufficient - proactive actions are needed to identify potential issues before they impact customers
• Study shows organizations with comprehensive data monitoring detect incidents 3.5x faster than those without

Harnessing AWS Health and Support Data

• AWS Health provides insights into service disruptions, scheduled changes, and account-specific notifications
• AWS Support cases contain valuable information on problems, solutions, workarounds, and lessons learned
• Key goals:
  1. Ingest AWS Health events at enterprise scale
  2. Ingest AWS Support case data
  3. Ingest and route data at scale
  4. Leverage data to drive proactive, data-driven operations

Fidelity's Cloud Event Notification Transport Service (SENSE)

• Fidelity's journey from a simple pull-based system to an event-driven architecture
• Challenges faced as Fidelity rapidly scaled its cloud footprint to 2,000 accounts and 5 million resources
• Key features of SENSE:
  • Leverages AWS Health's delegated administrative account and Event Bridge integration
  • Enriches raw events with additional context and metadata
  • Provides personalized notification preferences for users
  • Integrates with Fidelity's internal incident management and communication systems
  • Resilient architecture with regional backups and failover capabilities

Applying GenAI to Enhance Operational Intelligence

• Fidelity's "Machine Augmented Key Insights" (MAKI) framework:
  • Ingests and augments AWS Support case data with reference information
  • Analyzes event-level and aggregate-level patterns using different GenAI models
  • Provides summaries, recommendations, and remediation plans
• Use cases:
  1. Identifying trends and correlating events across health and support data
  2. Providing guidance and recommendations for deprecating events (e.g., RDS certificate expiration)
  3. Automating remediation and enhancing operational resilience

Towards a Holistic, Data-Driven Operational Platform

• Integrating additional data sources (observability, change management, cost) to enrich the operational view
• Exploring the potential of aggregating signals from multiple cloud providers and on-premises environments
• Leveraging public data sources (news, social media, weather, financial markets) to anticipate and proactively respond to potential issues

Key Takeaways

• Comprehensive data monitoring and proactive actions are crucial for managing the complexity of cloud environments
• Combining AWS Health and Support data provides valuable operational intelligence
• Fidelity's SENSE platform demonstrates how to ingest, enrich, and route this data at enterprise scale
• Applying GenAI techniques can further enhance operational intelligence and automate remediation
• Continuously expanding the data sources and integrating them into a holistic, data-driven operational platform is the ultimate goal