A
argbe.tech - news
1min read

BGL builds an agent-native BI layer with Claude Agent SDK and Bedrock AgentCore

BGL describes how it assembled a production text-to-SQL agent for business users using Claude Agent SDK, running in stateful sessions on Amazon Bedrock AgentCore. The approach pairs a curated analytics layer with identity-scoped access controls.

BGL outlines a production AI analytics agent that answers business questions in natural language while generating governed SQL via Claude Agent SDK running on Amazon Bedrock AgentCore.

  • BGL supports self-managed superannuation fund (SMSF) administration and reports serving over 12,700 businesses across 15 countries.
  • Its analytics environment includes 400+ domain-focused tables (for topics like compliance tracking, investment performance, and financial reporting) designed to be queried directly by non-technical users.
  • The data foundation is built on an Athena-based big data stack with transformations managed using dbt, producing denormalized, business-ready datasets for reporting-style questions.
  • The agent is designed for production controls relevant to financial services, including session isolation and identity-based access so results align with who’s asking.
  • The post cites agent adoption signals from Anthropic’s 2026 State of AI Agents Report: 60% of organizations rate data analysis/report generation as their most impactful agent use case, and 65% of enterprises rank it as a top priority.