BGL Democratizes Data Analytics for 200 Employees with Claude Agent SDK

by

The Era When Non-Developers Can Analyze Data: Real-World Claude Agent SDK Use Case

  • Australian financial company BGL built a text-to-SQL AI agent for all employees using Claude Agent SDK
  • Secured security and scalability with Amazon Bedrock AgentCore, enabling 200 employees to analyze data without SQL
  • Core architecture: Data-driven separation + code execution pattern + modular knowledge structure

What Happened?

Australian financial software company BGL built a company-wide BI (business intelligence) platform using Claude Agent SDK and Amazon Bedrock AgentCore.[AWS ML Blog]

Simply put, even employees who don’t know SQL can say “this month’s sales” in natural language. If they ask “show me the trend,” the AI automatically generates queries and draws charts.

BGL was already using Claude Code daily, but realized it wasn’t just a simple coding tool—it had the ability to reason about complex problems, execute code, and interact autonomously with systems.[AWS ML Blog]

Why Is It Important?

Personally, this case is interesting because it shows a practical answer to “How do you deploy AI agents in production environments?”

Most text-to-SQL demos work brilliantly, but problems arise when applied to actual work. Table join mistakes, edge case omissions, incorrect aggregations. To solve this, BGL separated the database and AI roles.

They created well-refined analytics tables with existing Athena + dbt, and the AI agent focuses only on generating SELECT queries. Honestly, this is the key. If you leave everything to AI, hallucinations increase.

Another notable point is the code execution pattern. Analytics queries return thousands of rows, sometimes several MB of data. Putting all of this in the context window would explode. BGL allows the AI to directly execute Python to process CSVs from the file system.

What Will Happen in the Future?

BGL is planning AgentCore Memory integration. The goal is to store user preferences and query patterns to generate more personalized responses.

The direction this example shows is clear. In 2026, enterprise AI is evolving from “fancy chatbots” to “agents that actually work.” The combination of Claude Agent SDK + Amazon Bedrock AgentCore is one such blueprint.

Frequently Asked Questions

Q: What exactly is Claude Agent SDK?

A: It’s an AI agent development tool made by Anthropic. Instead of the Claude model simply responding, it enables autonomous code execution, file manipulation, and system interaction. Through this, BGL handles text-to-SQL and Python data processing in a single agent.

Q: Why is Amazon Bedrock AgentCore needed?

A: Security isolation is essential for AI agents to execute arbitrary Python code. AgentCore provides a stateful execution environment that blocks access to data or credentials between sessions. It reduces concerns about infrastructure needed for production deployment.

Q: Is it actually effective?

A: BGL’s 200 employees now perform analysis on their own without help from the data team. Product managers validate hypotheses, compliance teams identify risk trends, and customer success teams perform real-time analysis during customer calls.

If this article was helpful, please subscribe to AI Digester.

Reference Resources

Leave a Comment