Discover how the largest source of financing for social, economic, and institutional development for Latin America and the Caribbean implemented AI to streamline the information flow and improve productivity.
Problem
The client was having difficulties in accessing and finding specific information within their extensive knowledge base, particularly reports stored in SharePoint. The existing system made it difficult for bank employees to retrieve relevant information efficiently, raising a productivity issue.
Objective
To implement an AI-powered conversational interface in order to simplify and streamline access to information from the knowledge base in a scalable manner, reducing information search times.
Challenge
Creating a connector that could effectively read the SharePoint documents and deliver them to the AI tool. Additionally, it was important to ensure the solution had a robust prompt mechanism to understand the context and generate accurate responses.
Solution
To address these challenges, we leveraged GeneXus Enterprise AI to develop an AI assistant, which is connected to the bank's knowledge base and allows employees to ask questions related to its documents, providing immediate and reliable answers adapted to the bank's terminology.
Using a Retrieval Augmented Generation (RAG) approach, the assistant vectorizes blocks of documents from the bank's knowledge base. It then retrieves the most relevant documents based on user queries. Additionally, a metadata extraction step was implemented to further enhance the retrieval process.
Developing this AI solution with GeneXus Enterprise AI ensured:
Results
The platform's versatility allowed us to overcome challenges and propose a hybrid SaaS solution, in which the bank hosts the LLMs and databases, while we host the front-end and back-end of the solution. At the same time, the platform's flexibility to adapt the AI assistant's tone to the bank's terminology paved the way for an even more positive user adoption. Consequently, the development process has been extremely positive, encouraging the client to request an extension of the deployment to support a larger user base and to plan for the inclusion of all 5,000 employees.
The implementation of the AI Assistant will result in a significant acceleration of the information flow within the bank, as employees will be able to chat with the bank’s Sharepoint base using natural language and receive immediate responses instead of manually searching the documents.
The client was having difficulties in accessing and finding specific information within their extensive knowledge base, particularly reports stored in SharePoint. The existing system made it difficult for bank employees to retrieve relevant information efficiently, raising a productivity issue.
Objective
To implement an AI-powered conversational interface in order to simplify and streamline access to information from the knowledge base in a scalable manner, reducing information search times.
Challenge
Creating a connector that could effectively read the SharePoint documents and deliver them to the AI tool. Additionally, it was important to ensure the solution had a robust prompt mechanism to understand the context and generate accurate responses.
Solution
To address these challenges, we leveraged GeneXus Enterprise AI to develop an AI assistant, which is connected to the bank's knowledge base and allows employees to ask questions related to its documents, providing immediate and reliable answers adapted to the bank's terminology.
Using a Retrieval Augmented Generation (RAG) approach, the assistant vectorizes blocks of documents from the bank's knowledge base. It then retrieves the most relevant documents based on user queries. Additionally, a metadata extraction step was implemented to further enhance the retrieval process.
Developing this AI solution with GeneXus Enterprise AI ensured:
- A significant acceleration of implementation times.
- Data security and privacy in handling and processing.
- Efficiency and accuracy in processing vast amounts of documentation without compromising performance, allowing for seamless scalability and updates to the document base.
Results
The platform's versatility allowed us to overcome challenges and propose a hybrid SaaS solution, in which the bank hosts the LLMs and databases, while we host the front-end and back-end of the solution. At the same time, the platform's flexibility to adapt the AI assistant's tone to the bank's terminology paved the way for an even more positive user adoption. Consequently, the development process has been extremely positive, encouraging the client to request an extension of the deployment to support a larger user base and to plan for the inclusion of all 5,000 employees.
The implementation of the AI Assistant will result in a significant acceleration of the information flow within the bank, as employees will be able to chat with the bank’s Sharepoint base using natural language and receive immediate responses instead of manually searching the documents.