How AWS SMGS uses an AI-powered conversational assistant to transform business management with Amazon Bedrock AgentCore

Background

AWS leaders face the challenge of making time-sensitive, global decisions based on complex data scattered across multiple hierarchies and systems. Traditional business intelligence methods were a major bottleneck. Leaders spent excessive hours manually gathering and reconciling data from disparate dashboards before reviews. This data fragmentation meant insights were inconsistent and incomplete. Furthermore, complex dashboards required specialized knowledge, creating dependencies on reporting teams and preventing on-demand access, which delayed critical decisions and limited organizational agility.

Key Points

The NarrateAI solution is built on a two-layer architecture that separates intensive data processing from instant user interaction.

• Automated Narrative Generation Layer (Batch Processing): This is a pre-processing pipeline that runs regularly. It uses three stages:

  1. Data Extraction: Parameterized SQL templates pull structured data from Amazon Redshift, adapting to each user's specific role and permissions for secure, multi-level breakdowns.
  2. Data Transformation: AWS Lambda processes the raw data into a structured JSON format, organizing it with specific section-type logic and hierarchical mappings.
  3. Narrative Rendering: Jinja templates convert the structured JSON into human-readable narrative text for each persona (e.g., for the CEO vs. a field manager).

• Real-Time Interaction Layer: This is the conversational interface, accessible through Amazon Quick. It uses Amazon Bedrock AgentCore as its core orchestration engine. This removed the need to build custom infrastructure, providing serverless architecture, authentication, memory management, and model integration. This layer handles user queries in natural language, routing them to specialized AI agents for intelligent processing and delivering immediate, contextual answers derived from the pre-computed narratives and live data.

Significance

NarrateAI represents a strategic shift from passive data reporting to active, conversational intelligence. The separation of batch processing and real-time interaction is a key engineering pattern for production scalability: comprehensive data is prepared in advance, allowing the conversational layer to deliver fast, accurate responses without overloading source systems. By leveraging Amazon Bedrock AgentCore, AWS accelerated deployment from months to weeks while ensuring production-quality security and observability. The solution's core significance is its ability to remove barriers between leaders and their data, enabling faster, better-informed decisions across the entire SMGS organization by providing immediate, actionable insights through a simple conversational interface.