Fintech Startup Ank Reduces Manual Reconciliation Transactions to under 3%

Ank is an Argentine fintech that developed a product for money transfers between banks and virtual accounts. It allows you to manage several accounts from different banks from a single app and facilitates quick transfers between them.

Challenge

Ank’s team needed to fulfill three main objectives within the scope of this project. Create an automated flow of reconciliation transactions to reduce manual work while improving the accuracy and trustworthiness of the process. Second, increase the frequency of the reconciliation process to obtain updated data. Finally, implement a batch data processing architecture based on best practices and technologies that could serve as a model for future pipelines. Together these needs could be solved by the implementation of a Modern Data Stack.

Solution

Mutt Data developed a system for automatic reconciliation on an hourly, daily, and weekly basis. The system extracts the information from the different sources, centralizes it in a Data Lake in Amazon Web Services, performs the relevant reconciliation checks and, if it detects inconsistencies, corrects them automatically or creates tickets forcasesthat require manual attention.

For more information on the implemented solution, check out Ank’s and Mutt’s case study featured on Amazon Web Services.

Client saw manual conciliation processes reduced to under 3% of their total monetary transactions.

“In Mutt Data we found a partner with extensive experience building data systems from design to production. They worked seamlessly with both our technical and business teams.

Their know-how in building modern data architectures on Amazon Web Services (AWS) was fundamental in developing a reconciliation system which has proven to efficiently scale our operations.”

Gustavo Arjones
Chief Technology Officer at Ank

The system is developed on an Airflow 2.0 installation on Amazon EKS and uses Amazon Athena, DBT, Great Expectations and PostgreSQL. Other tools include Kubernetes, K33, Google Auth for Airflow Login, and Alembic.

Impact

The client saw manual conciliation processes reduced to under 3% of their total monetary transactions.