Cuore automates votes and public notices with generative AI at operational scale

Cuore, a company specializing in management and governance solutions, made a significant leap in operational efficiency by automating the Financial Statement analysis process and thegeneration of votes and meeting notices. The initiative eliminated critical manual steps, reduced operational costs and created a scalable technological foundation for business growth.

Before automation, Cuore faced a structural limitation in its workflow. Financial Statement analysis relied heavily on manually reading PDFs, human interpretation of audit opinions, and manual filling of information into the system. This model made it unfeasible to create dozens or hundreds of daily votes, generating operational bottlenecks, high time consumption for accounting teams and risks of inconsistencies in the interpretation of reservations, emphases and abstentions.

In addition, the absence of an automated process compromised standardization, traceability and integration between internal systems, directly impacting productivity and the experience of end customers.

MadeinWeb developed a solution based on Generative AI to automate the extraction and categorization of information contained in audited financial statements. The project implemented two main engines:

• MPDF (Financial Statement Processing Engine) → Responsible for processing and interpreting financial documents, extracting information such as approval status, reservations and emphasis from auditors.

• MPA (Meeting Processing Engine) → Responsible for the automated generation of notices and the creation of votes on the Cuore platform.

AWS Services Used:
The solution was built using several AWS services to ensure scalability, security and efficiency:

• Amazon Bedrock → Used to process and interpret financial statements with Generative AI, using the Claude Haiku 3.5 and Claude Sonnet models 3.5.

• AWS Lambda → Serverless functions to perform the processes of information extraction, categorization and storage in the database.

• Amazon API Gateway → Manages application requests and integrates the different components of the solution.

• Amazon DynamoDB → NoSQL database used to store analysis results and voting status.

• Amazon SQS → Message queue for asynchronous processing, ensuring scalability in the voting creation flow.

• Amazon S3 → Secure storage for processed financial statements and generated notices.

• AWS IAM → Access control and security for the services used.

Partner Support (Pre & Post Implementation):

• During the initial phase, MadeinWeb worked closely with the Cuore team to define the project requirements, validate the AI models and configure the necessary integrations.

• During implementation, several tests were carried out to optimize the accuracy of the AI and ensure the reliability of the extracted data.

• Post-implementation, MadeinWeb offered continuous support for adjustments and improvements to the solution, as well as training for the Cuore team.

With the implementation of the solution, Cuore was able to transform a manual and time-consuming process into a highly automated and efficient workflow.

Specific Metrics:

• 100% accuracy in the latest validation tests carried out by the Cuore team.

• Reduction in processing time from 5 minutes and 32 seconds to 2 minutes and 30 seconds per document, representing an efficiency gain greater than 55%.

• Operational cost savings, reducing the need for manual review of extracted information.

• Extended scalability, allowing Cuore to process hundreds of votes per day without human intervention.

• Improved corporate governance for Cuore's customers, ensuring that meetings are generated reliably and in a timely manner.