Democratizing enterprise intelligence: BGL’s journey with Claude Agent SDK and Amazon Bedrock AgentCore

This publish is cowritten with James Luo from BGL.

Knowledge evaluation is rising as a high-impact use case for AI brokers. In accordance with Anthropic’s 2026 State of AI Brokers Report, 60% of organizations rank knowledge evaluation and report era as their most impactful agentic AI purposes. 65% of enterprises cite it as a prime precedence. In follow, companies face two widespread challenges:

• Enterprise customers with out technical information depend on knowledge groups for queries, which is time-consuming and creates a bottleneck.
• Conventional text-to-SQL options don’t present constant and correct outcomes.

Like many different companies, BGL confronted comparable challenges with its knowledge evaluation and reporting use circumstances. BGL is a number one supplier of self-managed superannuation fund (SMSF) administration options that assist people handle the advanced compliance and reporting of their very own or a shopper’s retirement financial savings, serving over 12,700 companies throughout 15 nations. BGL’s resolution processes advanced compliance and monetary knowledge by way of over 400 analytics tables, every representing a selected enterprise area, similar to aggregated buyer suggestions, funding efficiency, compliance monitoring, and monetary reporting. BGL’s prospects and workers want to search out insights from the information. For instance, Which merchandise had essentially the most destructive suggestions final quarter? or Present me funding developments for high-net-worth accounts. Working with Amazon Internet Providers (AWS), BGL constructed an AI agent utilizing Claude Agent SDK hosted on Amazon Bedrock AgentCore. By utilizing the AI agent enterprise customers can retrieve analytic insights by way of pure language whereas aligning with the safety and compliance necessities of monetary providers, together with session isolation and identity-based entry controls.

On this weblog publish, we discover how BGL constructed its production-ready AI agent utilizing Claude Agent SDK and Amazon Bedrock AgentCore. We cowl three key facets of BGL’s implementation:

• Why constructing a sturdy knowledge basis is crucial for dependable AI agent-based text-to-SQL options
• How BGL designed its AI agent utilizing Claude Agent SDK for code execution, context administration, and domain-specific experience
• How BGL used AgentCore to supply the perfect stateful execution classes in manufacturing for a safer, scalable AI agent.

Establishing sturdy knowledge foundations for an AI agent-based text-to-SQL resolution

When engineering groups implement an AI agent for analytics use circumstances, a typical anti-pattern is to have the agent deal with every part together with understanding database schemas, reworking advanced datasets, checking out enterprise logic for analyses and deciphering outcomes. The AI agent is prone to produce inconsistent outcomes and fail by becoming a member of tables incorrectly, lacking edge circumstances, or producing incorrect aggregations.

BGL used its current mature large knowledge resolution powered by Amazon Athena and dbt Labs, to course of and remodel terabytes of uncooked knowledge throughout varied enterprise knowledge sources. The extract, remodel, and cargo (ETL) course of builds analytic tables and every desk solutions a selected class of enterprise questions. These tables are aggregated, denormalized datasets (with metrics and, summaries) that function a business-ready single supply of reality for enterprise intelligence (BI) instruments, AI brokers, and purposes. For particulars on construct a serverless knowledge transformation structure with Athena and dbt, see How BMW Group constructed a serverless terabyte-scale knowledge transformation structure with dbt and Amazon Athena.

The AI agent’s function is to deal with advanced knowledge transformation throughout the knowledge system by specializing in deciphering the consumer’s pure language questions, translating it, and producing SQL SELECT queries towards well-structured analytic tables. When wanted, the AI agent writes Python scripts to additional course of outcomes and generate visualizations. This separation of considerations considerably reduces the danger of hallucination and presents a number of key advantages:

• Consistency: The info system handles advanced enterprise logic in a extra deterministic means: joins, aggregations, and enterprise guidelines are validated by the information workforce forward of time. The AI agent’s job turns into simple: interpret questions and generate primary SELECT queries towards these tables.
• Efficiency: Analytic tables are pre-aggregated and optimized with correct indexes. The agent performs primary queries fairly than advanced joins throughout uncooked tables, leading to a quicker response time even for big datasets.
• Maintainability and governance: Enterprise logic resides within the knowledge system, not within the AI’s context window. This helps be certain that the AI agent depends on the identical single supply of reality as different shoppers, similar to BI instruments. If a enterprise rule adjustments, the information workforce updates the information transformation logic in dbt, and the AI agent robotically consumes the up to date analytic tables that replicate these adjustments.

“Many individuals assume the AI agent is so highly effective that they will skip constructing the information platform; they need the agent to do every part. However you may’t obtain constant and correct outcomes that means. Every layer ought to resolve complexity on the acceptable degree” 

– James Luo, BGL Head of Knowledge and AI

How BGL builds AI brokers utilizing Claude Agent SDK with Amazon Bedrock

BGL’s growth workforce has been utilizing Claude Code powered by Amazon Bedrock as its AI coding assistant. This integration makes use of momentary, session-based entry to mitigate credential publicity, and integrates with current id suppliers to align with monetary providers compliance necessities. For particulars of integration, see Steerage for Claude Code with Amazon Bedrock

Via its each day use of the Claude Code, BGL acknowledged that its core capabilities prolong past coding. BGL used its skill to motive by way of advanced issues, write and execute code, and work together with information and techniques autonomously. Claude Agent SDK packages the identical agentic capabilities right into a Python and TypeScript SDK, in order that builders can construct customized AI brokers on prime of Claude Code. For BGL, this meant they may construct an analytics AI agent with:

• Code execution: The agent writes and runs Python code to course of datasets returned from analytic tables and generate visualizations
• Computerized context administration: Lengthy-running classes don’t overwhelm token limits
• Sandboxed execution: Manufacturing-grade isolation and permission controls
• Modular reminiscence and information: A CLAUDE.md file for undertaking context and Agent Expertise for product line domain-specific experience

Why code execution issues for knowledge analytics

Analytics queries typically return hundreds of rows and typically past megabytes of information. Customary tool-use, operate calling, and Mannequin Context Protocol (MCP) patterns typically go retrieved knowledge immediately into the context window, which rapidly reaches mannequin context window limits. BGL applied a special method: the agent writes SQL to question Athena, then writes Python code to course of the CSV file outcomes immediately in its file system. This permits the agent to deal with massive end result units, carry out advanced aggregations, and generate charts with out reaching context window limits. You may study extra in regards to the code execution patterns in Code execution with MCP: Constructing extra environment friendly brokers.

Modular information structure

To deal with BGL’s numerous product traces and sophisticated area information, the implementation makes use of a modular method with two key configuration sorts that work collectively seamlessly.

CLAUDE.md (undertaking context)

The CLAUDE.md file gives the agent with international context—the undertaking construction, setting configuration (take a look at, manufacturing, and so forth), and critically, execute SQL queries. It defines which folders retailer intermediate outcomes and remaining outputs, ensuring information land in an outlined file path that customers can entry. The next diagram exhibits the construction of a CLAUDE.md file:

SKILL.md (Product area experience)

BGL organizes their agent area information by product traces utilizing the SKILL.md configuration information. Every talent acts as a specialised knowledge analyst for a selected product. For instance, the BGL CAS 360 product has a talent known as CAS360 Knowledge Analyst agent, which handles firm and belief administration with ASIC compliance alignment; whereas BGL’s Easy Fund 360 product has a talent known as Easy Fund 360 Knowledge Analyst agent, which is supplied with SMSF administration and compliance-related area abilities. A SKILL.md file defines three issues:

• When to set off: What kinds of questions ought to activate this talent
• Which tables to make use of or map: References to the related analytic tables within the knowledge folder (as proven within the previous determine)
• Easy methods to deal with advanced eventualities: Step-by-step steering for multi-table queries or particular enterprise questions if required

By utilizing SKILL.md information, the agent can dynamically uncover and cargo the fitting talent to realize domain-specific experience for corresponding duties.

• Unified context: When a talent is triggered, Claude Agent SDK dynamically merges its specialised directions with the worldwide CLAUDE.md file right into a single immediate. This permits the agent to concurrently apply project-wide requirements (for instance, at all times save to disk) whereas utilizing domain-specific information (similar to mapping consumer inquiries to a bunch of tables).
• Progressive discovery: Not all abilities have to be loaded into the context window directly. The agent first reads the question to find out which talent must be triggered. It masses the talent physique and references to know which analytic desk’s metadata is required. It then additional explores corresponding knowledge folders. This retains context utilization environment friendly whereas offering complete protection.
• Iterative refinement: If the AI agent is unable to deal with some enterprise information due to an absence of latest area information, the workforce will collect suggestions from customers, establish the gaps, and add new information to current abilities utilizing a human-in-the-loop course of so abilities are up to date and refined iteratively.

As proven within the previous determine, agent abilities are organized per product line. Every product folder accommodates a SKILL.md definition file and a references listing with extra area information and help supplies that the agent masses on demand.

For particulars about Anthropic Agent Expertise, see the Anthropic weblog publish, brokers for the true world with Agent Expertise

Excessive-level resolution structure

To ship a safer and scalable text-to-SQL expertise, BGL makes use of Amazon Bedrock AgentCore to host Claude Agent SDK whereas protecting knowledge transformation within the current large knowledge resolution.

The previous determine illustrates a high-level structure and workflow. The analytic tables are pre-built each day utilizing Athena and dbt, and function the single supply of reality. A typical consumer interplay flows by way of the next levels:

1. Consumer request: A consumer asks a enterprise query utilizing Slack (for instance, Which merchandise had essentially the most destructive suggestions final quarter?).
2. Schema discovery and SQL era: The agent identifies related tables utilizing abilities and writes SQL queries.
3. SQL safety validation: To assist stop unintended knowledge modification, a safety layer permits solely SELECT queries and blocks DELETE, UPDATE, and DROP operations.
4. Question execution: Athena executes the question and shops outcomes into Amazon Easy Storage Service (Amazon S3).
5. Outcome Obtain: The agent downloads the ensuing CSV file to the file system on AgentCore, utterly bypassing the context window to keep away from token limits.
6. Evaluation and visualization: The agent writes Python code to research the CSV file and generate visualizations or refined datasets relying on the enterprise query.
7. Response supply: Last insights and visualizations are formatted and returned to the consumer in Slack.

Why use Amazon Bedrock AgentCore to host Claude Agent SDK

Deploying an AI agent that executes arbitrary Python code requires important infrastructure concerns. As an illustration, you want isolation to assist be certain that there’s no cross-session entry to knowledge or credentials. Amazon Bedrock AgentCore gives fully-managed, stateful execution classes, every session has its personal remoted microVM with a separate CPU, reminiscence, and file system. When a session ends, the microVM terminates totally and sanitizes reminiscence, serving to to make sure no remnants persist for future classes. BGL discovered this service particularly priceless:

• Stateful execution session: AgentCore maintains session state for as much as 8 hours. Customers can have ongoing conversations with the agent, referring again to earlier queries with out shedding context.
• Framework flexibility: It’s framework-agnostic. It helps deployment of AI brokers similar to Strands Brokers SDK, Claude Agent SDK, LangGraph, and CrewAI with a number of traces of code.
• Aligned with safety greatest practices: It gives session isolation, VPC help, AWS Identification and Entry Administration (IAM) or OAuth primarily based id to facilitate ruled, compliance-aligned agent operations at scale.
• System integration: This can be a forward-looking consideration.

“There’s Gateway, Reminiscence, Browser instruments, an entire ecosystem constructed round it. I do know AWS is investing on this path, so every part we construct now can combine with these providers sooner or later.”

– James Luo, BGL Head of Knowledge and AI. 

BGL is already planning to combine AgentCore Reminiscence for storing consumer preferences and question patterns.

Outcomes and influence

For BGL’s greater than 200 workers, this represents a big shift in how they extract enterprise intelligence. Product managers can now validate hypotheses immediately with out ready for the information workforce. Compliance groups can spot threat developments with out studying SQL. Buyer success managers can pull account-specific analytics in real-time throughout shopper calls. This democratization of information entry helps remodel analytics from a bottleneck right into a aggressive benefit, enabling quicker decision-making throughout the group whereas releasing the information workforce to concentrate on strategic initiatives fairly than one-time question requests.

Conclusion and key takeaways

BGL’s journey demonstrates how combining a robust knowledge basis with agentic AI can democratize enterprise intelligence. By utilizing Amazon Bedrock AgentCore and the Claude Agent SDK, BGL constructed a safer and scalable AI agent that empowers workers to faucet into their knowledge to reply enterprise questions. Listed here are some key takeaways:

• Spend money on a robust knowledge basis: Accuracy begins with a robust knowledge basis. By utilizing the information system and knowledge pipeline to deal with advanced enterprise logic (joins and aggregations), the agent can concentrate on primary, dependable logic.
• Manage information by area: Use Agent Expertise to encapsulate domain-specific experience (for instance, Tax Legislation or Funding Efficiency). This retains the context window clear and manageable. Moreover, set up a suggestions loop: constantly monitor consumer queries to establish gaps and iteratively replace these abilities.
• Use code execution for knowledge processing: Keep away from utilizing an agent to course of massive datasets utilizing a big language mannequin (LLM) context. As a substitute, instruct the agent to put in writing and execute code to filter, combination, and visualize knowledge.
• Select stateful, session-based infrastructure to host the agent: Conversational analytics requires persistent context. Amazon Bedrock AgentCore simplifies this by offering built-in state persistence (as much as 8-hour classes), assuaging the necessity to construct customized state dealing with layers on prime of stateless compute.

In the event you’re able to construct comparable capabilities on your group, get began by exploring the Claude Agent SDK and a brief demo of Deploying Claude Agent SDK on Amazon Bedrock AgentCore Runtime. If in case you have an identical use case or want help designing your structure, attain out to your AWS account workforce.

References:

Concerning the authors

Dustin Liu is a options architect at AWS, targeted on supporting monetary providers and insurance coverage (FSI) startups and SaaS corporations. He has a various background spanning knowledge engineering, knowledge science, and machine studying, and he’s keen about leveraging AI/ML to drive innovation and enterprise transformation.

Melanie Li, PhD, is a Senior Generative AI Specialist Options Architect at AWS primarily based in Sydney, Australia, the place her focus is on working with prospects to construct options leveraging state-of-the-art AI and machine studying instruments. She has been actively concerned in a number of Generative AI initiatives throughout APJ, harnessing the ability of Giant Language Fashions (LLMs). Previous to becoming a member of AWS, Dr. Li held knowledge science roles within the monetary and retail industries.

Frank Tan is a Senior Options Architect at AWS with a particular curiosity in Utilized AI. Coming from a product growth background, he’s pushed to bridge expertise and enterprise success.

James Luo is Head of Knowledge & AI at BGL Company Options, a world-leading supplier of compliance software program for accountants and monetary professionals. Since becoming a member of BGL in 2008, James has progressed from developer to architect to his present management function, spearheading the Knowledge Platform and Roni AI Agent initiatives. In 2015, he shaped BGL’s BigData workforce, implementing the primary deep studying mannequin within the SMSF business (2017), which now processes 200+ million transactions yearly. He has spoken at Massive Knowledge & AI World and AWS Summit, and BGL’s AI work has been featured in a number of AWS case research.

Dr. James Bland is a Expertise Chief with 30+ years driving AI transformation at scale. He holds a PhD in Laptop Science with a machine studying focus and leads strategic AI initiatives at AWS, enabling enterprises to undertake AI-powered growth lifecycles and agentic capabilities. Dr. Bland spearheaded the AI-SDLC initiative, authored complete guides on Generative AI within the SDLC, and helps enterprises architect production-scale AI options that basically remodel how organizations function in an AI-first world.