Retail bank employs neural networks in the 90’s
A critical operation for a retail bank needed to be improved in order to deliver its full potential for profitability of the operation, for attending repressed demand and for preparation for the next cycle of innovation on digital services in an industry where information technology is key for success.
The third biggest Brazilian bank at that time.
A very expensive operation for a retail bank
Bank cheques are very popular in Brazilian culture; far more popular in the 90’s, well before the arrival of smartphones. At that time, millions of bank cheques had to be processed every day, in a window of only 4 hours, according to Brazilian Central Bank regulations. Not only the operation is expensive from the perspective of human resources involved, the operation is expensive when the window for interbank communication is missed and the bank has to put its own money to cover these interbank operations.
On top of that, every document has to be microfilmed and kept for later retrieval, another very expensive operation, involving a lot of staff and real estate. Customers are happy to pay for retrieval of documents, but the bank cannot attend the demand due to operational limitations imposed by microfilm equipment and the overall workflow involved.
Reduce costs, improve productivity, attend repressed demand
Processing cheques need to be automated to the next level, in order to avoid missing the window for processing and in order to attending the increasing popularity of bank cheques among the general population. Alongside of that, cheques must be scanned only once, electronically, and the image should be recognized by a neural network, reducing the need of staff reading the figures involved. The entire process deserves to be redesigned, bringing new opportunities for more efficiency, new opportunities for cost reduction and preparing the bank for a cycle of electronic services which would benefit customers and open up new opportunities of revenue.
Enabling a solution
Mathminds was hired by a French consultancy group responsible for the implementation of the solution. Mathminds played a central, key role in the process, being engaged in all steps involving prototyping, implementation, deployment and production support of the solution. A team of nearly 20 people was involved directly or indirectly as part of Mathminds’ team responsible for guaranteeing the correctness of the solution, its performance, other functional and non-functional requirements.
- A complex project, but plenty succesful and delivered according to plan;
- Project awarded international recognition for its innovation, relevance, results and size.
- Reduction of costs with equipment and real state for operation;
- Reduction of costs with staff necessary for operation;
- Reduction of costs involved to missing processing windows;
- Preparation for a new cycle of augmented digital services.