CBRE and Amazon Web Services collaborate to execute natural language queries on structured data using Amazon Bedrock

Spread the love



CBRE, the world’s largest commercial real estate services and investment firm, is leveraging artificial intelligence (AI) to enhance their services. With a vast data environment and enterprise-grade technology, CBRE can deploy AI solutions throughout the commercial real estate lifecycle. To enable customers to make custom queries using natural language prompts, CBRE partnered with Amazon Bedrock to develop a solution. Amazon Bedrock offers high-performing foundation models from leading AI companies, simplifying the development process while ensuring privacy and security.

The project involved developing a custom query environment that allows clients to query existing data using natural language prompts in a user-friendly way. The system uses Lambda functions, OpenSearch Service, and Anthropic Claude 2 on Amazon Bedrock to search the client’s database and generate responses in plain English, along with the reasoning and SQL code. A web interface and API management layer were built to facilitate user testing and integration into CBRE’s existing business intelligence tools.

The solution provides a scalable way for clients to query data, with specific requirements for natural language queries, SQL generation, and response times. Performance tuning was applied to optimize input contexts, prompt engineering, and inference parameters. Monitoring was done through Amazon CloudWatch logs and metrics, and the solution was deployed using AWS CDK stacks for infrastructure provisioning.

The project showcases the potential of generative AI in data analytics, demonstrating how CBRE can harness AI technologies to enhance their business offerings and maintain their leadership in the real estate market. The team behind the project included professionals from CBRE and AWS, working together to develop a cutting-edge solution that supports client needs and showcases the power of AI in real estate services.

Article Source
https://aws.amazon.com/blogs/machine-learning/cbre-and-aws-perform-natural-language-queries-of-structured-data-using-amazon-bedrock/