Rich Data Co deployed generative AI assistants on Amazon Bedrock to accelerate credit decisioning
Rich Data Co (RDC) is a Sydney-based SaaS provider specializing in AI-driven credit decisioning for business and commercial lending. RDC built two assistants to support data scientists and portfolio managers with model development, data querying, and portfolio analysis for financial institutions.
- Organization
- Rich Data Co
- Industry
- Finance
- Location
- Australia
- Published
- June 2026
Reported outcomes
+50%
quantified impactOther quantified impact
Strategic outcomes
Primary read
Use case focus
Showing 3 of 5
- 1Credit Decisioning
- 2Data Science Assistant
- 3Portfolio Analytics
- Improve credit assessments using richer data sources.
- Assist data scientists and portfolio managers with model development, querying, and portfolio analysis.
- Accelerate development while meeting regulated security requirements.
- Built two AI-driven assistants on Amazon Bedrock with Anthropic Claude.
- Used retrieval-augmented generation over Amazon OpenSearch Serverless for knowledge access.
- Implemented natural-language-to-SQL against Amazon Aurora MySQL for portfolio analysis with iterative query correction.
- Initial version deployed to production in three months.
- Development speed doubled.
- Pilot with leading Australian banks.
- Estimated increase in pre-assessed credit offers from below 50% to well above 50% for existing customers.
Architecture
RDC built two AI assistants on AWS: a Data Science Assistant using Anthropic Claude in Amazon Bedrock with retrieval-augmented generation over Amazon OpenSearch Serverless, and a Portfolio Assistant using natural-language-to-SQL over Amazon Aurora MySQL with iterative query correction and validation. The solution was developed with AWS Generative AI Innovation Center support and deployed to production in three months.
Implementation partners1
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Was this useful?