A leading regional bank with business lines in wealth management, commercial banking, and investment banking, holding over $20 billion in assets and a yearly revenue exceeding $1 billion, was looking for a comprehensive Gen AI platform. The goal was to enable their business units to deliver GenAI / AI-ML solutions for 20+ high-value use cases. Security and self-service capabilities were critical decision-making criteria in selecting an enterprise-grade platform.

After evaluating GenAI platform options, the customer decided to build their own best-of-breed approach using a proven platform such as AWS Bedrock, which needed to meet both GenAI use cases and classic AI/ML ones.

Hitachi Digital Services developed a scalable and purpose-built AI platform anchored on AWS Bedrock tailored to the customer’s unique business needs. We focused on delivering a robust foundation that included capabilities such as data ingestion pipelines, model development and deployment workflows, and model operations. We implemented key platform features such as prompt management, user feedback capture, and automated evaluation jobs that continuously assess model and RAG performance using real user feedback.  

Guardrails from Bedrock were applied to ensure responsible AI, and a metrics dashboard was developed to track evaluation results and cost metrics. To ensure the platform met strict enterprise standards, we also implemented a governance layer, including a customer-specific Trust Layer designed to uphold security and compliance practices. While the immediate focus was on Generative AI use cases, we deliberately designed the platform to be flexible and scalable enough to support classic AI and machine learning use cases in the future. The solution was grounded in real-world use cases such as risk assessment and fraud detection, KYC and customer onboarding, and regulatory compliance. 

The core components of the solution included: 

• AWS Bedrock 

• LLM Models provisioned through Bedrock 

• AWS Bedrock Knowledgebases 

• AWS Bedrock Guardrail 

• EKS 

• Lambda 

• AWS OpenSearch 

• AWS Neptune 

• Bedrock APIs (Boto3) 

• AWS SageMaker 

Hitachi Digital Services’ successful execution of the production delivered a GenAI platform that could host GenAI-developed applications. The MVP accelerated the customer’s GenAI adoption by providing a secure, feedback-driven platform to experiment and scale use cases efficiently.  

It enabled faster time-to-value with reusable components like the AI Platform portal for effective business user engagement, RAG framework, and evaluation pipeline. Business units are able to rapidly experiment with GenAI prototype applications in a secure environment, and the integrated feedback and metrics loop helped improve model accuracy, optimize costs, and build trust in AI outcomes. 

Key Callouts: 

• 20+ use cases enabled by GenAI 

• Faster time-to-value achieved 

• Improved model accuracy