The Solution

In partnership with CloudDog, Catálise developed an intelligent chatbot based on Amazon Bedrock. The solution uses a Retrieval-Augmented Generation (RAG) architecture and the Claude 3.5 Sonnet language model, enabling the AI to analyze exclusively the documents stored in the company’s own database, ensuring privacy and accuracy.

In addition to the AI implementation, Catálise contracted CloudDog’s Managed Services. This ensures that the entire environment—including Amazon Aurora, DynamoDB, Lambda, and S3—is continuously monitored and optimized, allowing Catálise’s IT team to focus on business strategy while CloudDog ensures governance, security, and cloud infrastructure availability.

Solution Architecture

Main services used:

• AWS Lambda

• Amazon Bedrock

• Amazon S3

• Amazon DynamoDB

• Amazon API Gateway

• Amazon Aurora PostgreSQL

Results and Benefits

The solution went live in early 2025 and delivered immediate impacts on the asset manager’s operational efficiency:

• 80% reduction in the time required to analyze documents such as contracts and regulations.

• 60% reduction in the need for full manual document reading.

• Average savings of 25 hours per month per legal team member.

• Increased accuracy, with a 70% reduction in errors caused by misinterpretation of text.

• Full ERP integration, allowing the chatbot to operate directly within the company’s management system.

• Governance and specialized support through CloudDog’s Managed Services.

“Meeting security requirements was fundamental to choosing AWS, whose platform offered exactly what we were looking for.” — Pedro Deguchi, IT Manager at Catálise Investimentos.