This entity was seeking solutions to address this problem by creating an automated system for  the  structuring  and  mass  processing  of  claims  and  inquiries,  significantly  improving  response times and strengthening its supervisory duties.

To achieve the objective, we developed a system that utilizes machine learning techniques to classify claims received by the market and entities, products, and claimed subjects.  This  process  began  with  data  collection  and  cleansing,  specifically  gathering  claims  along  with  their  corresponding  classifications.  Next,  we  employed  classification algorithms  and  neural  networks  to  analyze  the  information,  testing  alternative  models  and hyperparameter combinations until achieving the desired outcomes.

Furthermore,  our  solution  included  the  development  of  a  web  service  to  enable  the  proposed  system's  interoperability,  meaning  it  can  be  consumed  and  integrated  by  the  client  entity's  systems.  Additionally,  we  implemented  scalability  elements  in  the  solution  through automation and scalability practices known as MLOps, providing capabilities for continuous  model  retraining  and  identifying  issues  that  could  affect  the  solution  in  a  production environment.

The  results  of  our  solution  were  highly  satisfactory.  In  tests  conducted  in  relevant  and simulated environments, we achieved 98% accuracy in claim classification by market  and  92%  accuracy  in  classification  by  entities,  products,  and  claimed  subjects.

In  real-world  tests,  system  users  reported  a  significant  improvement  in  response  time and analysis quality. Claims that previously book an average of one week for classification are now processed in under 5 seconds. This  dramatic  reduction  in  processing  time resulted  in  enhanced  efficiency  in  the  management process and overall financial market monitoring