About the Company
The client is a real estate technology company that processes large volumes of documents as part of its product. It had already adopted machine learning on AWS and wanted to push its document processing further with the latest generation of AI.
The Challenge
While the company already used machine learning to process documents, setting up each new document type remained slow and manual. The team wanted a solution that could learn and adapt dynamically from real user input, rather than depending solely on individually fine-tuned models, and that improved both the speed and accuracy of processing. It also needed to fit cleanly into the AWS environment the company already trusted, with data sovereignty and regional processing treated as first-order requirements.
The Solution
NileForge extended the company's existing AWS environment with a generative AI solution built on Amazon Bedrock, AWS's managed service for building with leading foundation models, including Anthropic's Claude. Rather than replacing what worked, NileForge designed a hybrid approach that combines the company's established machine learning with generative AI, so each is applied where it performs best.
The solution runs on Amazon EKS with Amazon EC2, alongside the company's existing Amazon SageMaker models, with AWS CloudFormation automating infrastructure across the stack. Processing is triggered asynchronously by customer actions, so it scales with real usage, and the design kept data sovereignty, regional processing, and seamless integration with the existing AWS environment central throughout.
The Results
- Document setup that once was slow and manual became substantially faster
- More accurate processing for the company's customers
- A hybrid design that pairs traditional machine learning with generative AI
- A solution proven on its first use case, then adopted as the basis for the wider application
- Data sovereignty and regional processing preserved by design
Apply generative AI where it counts.
NileForge helps teams move generative AI from idea to production on AWS, grounded in their own data and built to last. If you have a use case in mind, we should talk. Talk to our team
