AWS SMGS launches NarrateAI on Bedrock AgentCore for conversational BI at scale

AWS SMGS (Sales, Marketing and Global Services) has deployed NarrateAI, an intelligent conversational assistant powered by Amazon Bedrock AgentCore, to transform how its leaders access business intelligence. The system tackles three core obstacles: time-intensive manual data preparation, fragmentation of insights across systems, and limited accessibility requiring intermediary reporting teams. NarrateAI uses a two-layer architecture: a batch processing pipeline that generates persona-based narratives overnight (data extraction via parameterized SQL from Amazon Redshift, transformation with AWS Lambda, and rendering using Jinja templates stored in S3 for row-level security), and a real-time conversational interface that allows leaders to query their data in natural language through Amazon Quick. This separation enables comprehensive pre-computation of narratives while delivering instant, contextually accurate responses during interactions.

By using Amazon Bedrock AgentCore, AWS avoided building custom orchestration infrastructure—benefiting from its serverless architecture, built-in authentication, memory management, and foundation model integrations. This accelerated deployment from months to weeks while maintaining production-grade observability via Amazon CloudWatch. NarrateAI’s specialized AI agents handle intelligent routing and validation, ensuring accurate responses across hierarchies from the CEO to field sales. The result: business leaders no longer spend hours reconciling dashboards; they can now ask “What are our top three underperforming regions in Q4?” and receive an immediate, data-backed narrative tailored to their role and permissions.