support

Pragma's Gen AI Solutions for an Insurance Company

About the client

Our client, a leader in the insurance sector in Central America, offers a wide range of services, including policies, social security, insurance, and trusts. It sought to optimize the interaction with its customers and collaborators through innovative technology solutions such as Generative Artificial Intelligence (Gen AI).

Imágenes-Caso-Gen-AI_Thumbnail-pagina

The challenge we solved with this Gen AI Assistant

Pragma's Generative AI assistant enables queries to databases.

The insurer needed to improve efficiency and accuracy in the management of its insurance portfolio. The challenge was to develop a solution that:

 

  • Provide updated, detailed, and accurate information on the different types of insurance, their coverages, and exceptions.

  • Perform customized risk analysis to recommend the most appropriate coverages to customers.

  • Help the sales team keep up to date on regulations and best practices.
    Transform large volumes of data into easy-to-understand insights and provide intuitive, real-time access to company information.

Our Outcome

97.8%

Model response accuracy.

1.58 

Seconds of response time.

96.6%

Employees preferred a formal interaction style from the three chatbot personalities offered

The virtual assistant has transformed the user experience. Thanks to this project, we reduced the time it takes to get accurate insurance, and price recommendations.

The Gen AI assistant we developed became the sales team's primary support, helping them stay updated and offering reliable real-time information.  In addition, integrating new functionalities such as risk analysis and customized estimates improved the sales team's ability to advise clients in a personalized way.

The architecture developed for this solution also ensures privacy, security, and governance of the data processed.

An insurance sales agent uses the generative AI assistant.

The project started as a proof of concept, evolved into a Minimum Viable Product, and finally into a deployed solution that is constantly improving. Currently, It has significantly improved the operational efficiency of the sales team with eight new functionalities:

  1. Data query tool: Collaborators can now directly access relational databases to obtain insights without the need for advanced technical knowledge. This allows the generation of reports and visualizations that facilitate the understanding of data, democratizing access to information and improving data-driven decision-making.
  2. Price estimation Wizard: Calculates customized estimations by taking different variables into account. This speeds up the quoting process and improves accuracy by eliminating the risk of human error.
  3. Document Reader: AI processes PDF or text documents, extracting key information and responding to specific queries. 
  4. Drafting Assistant: This assistant enables the generation of content for social networks and facilitates the drafting of meeting minutes from notes or recordings, optimizing internal and external communication processes. It is also capable of processing images to create marketing content.
  5. Information Simplifier:  This functionality allows information to be presented in the format that best suits the user's needs, whether in the form of lists, tables, flat files, or even in JSON format. 
  6. Programming Assistant: Developers now have a wizard to help them write code, detect errors, and generate database queries. This speeds up the development of new applications, improves code quality, and reduces the time needed for testing and corrections.

 

A team from the insurance company interacts with John, Pragma's generative AI.

  1. Transcription Tool: From PDF or audio files, it automatically completes forms. This has the potential to speed up processes within the insurer.
  2. Communication with relational databases: The AI's ability to connect to relational databases such as SQL Server allows employees to perform complex queries without mastering SQL. In addition, this functionality can automatically generate graphs, making information even more accessible and visual for non-technical users.

 

How we did it?

Pragma's generative AI assistant responds to questions in natural language

Pragma led the development of a Gen AI solution that has added new functionalities since its launch. We used state-of-the-art technologies such as Azure OpenAI and Cognitive Search to address this challenge.

For the first part of the project, we stored and organized essential documents in Azure's Blob Storage. This first step ensured information scalability and security and laid the groundwork for more efficient data management. We then used automatic indexing tools to organize and tag the information.

With the project's data in good shape, the next step was to select and calibrate the solution's Foundational Model through Azure OpenAI. These models were tuned and tested to process large volumes of data efficiently. 

Another critical piece of the project was the implementation of vectorized searches using the embedding-ada text embedding model. This model allowed searches to be performed based on the meaning and context of the information, ensuring more accurate and relevant responses. Vectorized search technology will enable us to capture the essential information from documents and perform complex natural language queries with highly accurate results.

All of the above allowed us to build a virtual assistant capable of answering queries, taking into account the following: 

  • Conversational context.
    Business tone and identity.
  • Direct integration with company data.
  • Documents uploaded directly from the chat interface.
Thanks to these implementations, we were able to offer an integral solution that initially transformed the way in which the insurance company stores, processes, and accesses its information. In its current stage, thanks to the new functionalities, this virtual assistant has also allowed different areas of the organization to access data-based insights and speed up processes such as onboarding new clients.

Related success stories

banco

Bancolombia_S.A._logo.svg-1

We strengthen the culture of agility in one of the biggest banks in Latam 

We drove Bancolombia's digital transformation, reshaping the way they work and obtaining better results in their innovative initiatives.

Learn more
Carrusel-foto

nequi-logo-color

We developed a 100% digital with more than 12 million users

We helped Nequi reach the category of Best Digital Banking worldwide by changing how banks relate to users.

Learn more
See all Success Stories

Keep Moving. Conversemos
keep moving /