Agibank Improves Content Management and Reduces Costs by 90%

Agibank is a neobank based in Campinas, in the state of São Paulo, one of the largest innovation centers in the country. Democratizing access to financial services is the central pillar of the company's operations, which was created with the purpose of generating financial and digital inclusion for everyone, especially for the low-income population. As the majority of customers in this profile need support and advice offered in person for the transition from the physical to the digital environment, the institution created a hybrid platform, which combines complete digital banking with a network of approximately 1,000 smart hubs - physical stores designed to serve the non-native digital audience - which are spread throughout Brazil.

“Our goal is to increasingly expand and simplify access to financial products and services for millions of Brazilians. We aim to make people’s daily lives better through credit, insurance and other banking services,” says Marcos Lanzarini, database specialist at Agibank. “Technology plays a key role in ensuring we can handle a high volume of transactions every day while keeping data secure and systems resilient,” reinforces the executive.

Currently, the institution has 2.7 million active customers, manages assets worth around R$17 billion and responds to more than 10,000 loan requests every day.

Agibank is on an impressive growth trajectory, aiming to expand its network to over two thousand branches by 2027. To achieve this goal, the institution needs robust and scalable technology underpinning daily and critical operations.

When a customer makes a financial transaction, applies for a loan or credit card, or takes out an insurance policy, the bank needs to confirm their identity. This is handled by the enterprise content management (ECM) system, which stores customer records such as tax and identification documents, photos, and biometric data.

Previously, the ECM system ran on a relational PostgreSQL database, and as demand and data volumes grew it began to experience performance issues. It was inflexible, with different schemas requiring different databases, and suffered from latency when executing tasks. The database also couldn’t support secondary indexes or full-text search, impacting the sales of core offerings, such as credit and loans, insurance, and credit card approvals.

“We were constantly updating the database and manually scaling, which was time-consuming and prone to errors,” explains Lanzarini. “We needed a resilient system to cope with rising demand and more flexibility to support new services in the future.”

Given the critical impact on customer experience and, consequently, the bank's reputation, Agibank initiated the search for a new database platform to optimize its hybrid multi-cloud environment, incorporating both Amazon Web Services (AWS) and Microsoft Azure.

In 2022, Agibank implemented MongoDB Atlas on AWS and migrated its ECM system. Now, when a customer needs to verify their identity, a request comes into the MongoDB database to be authenticated and validated. Documents can be seamlessly created, read, updated, and deleted through and API integration, while being enhanced with both business and technical metadata. They are stored in line with document management policies, and each transaction is meticulously logged in the system to ensure a transparent audit trail.

With no prior experience with non-relational databases, Agibank developers and database admins worked closely with MongoDB to design and implement the solution in line with best practices. “Through MongoDB University and public training we learned the skills required to manage the platform in-house,” says Lanzarini.

To simplify data management, the ECM team reduced the number of tables and collections by 80%, which is significantly more efficient and boosts the performance of some ECM requests like transaction confirmation and facial biometry. High availability and disaster recovery strategies were implemented to meet recovery time objective (RTO) and recovery point objective (RPO) targets, achieving an RPO of just one minute for critical systems.

Marcos Lanzarini, database specialist, Agibank

“System upgrades now happen seamlessly in the background without disrupting performance, leaving database admins more time to focus on strategic activities,” Lanzarini comments.

Agibank used MongoDB Atlas auditing resources to implement security information and event management (SIEM) tools to improve service monitoring. The Secure Sockets Layer (SSL) encryption protocol was rolled out to protect customer data against cyber attacks.

Additionally, MongoDB Atlas is deliberately designed to integrate with Agibank's existing Prometheus monitoring system, providing the Database Administrator’s team with a unified dashboard for identifying and resolving potential issues before they lead to outages - complementing the strategy in Atlas’ observability offering, including Performance Advisor and Query Profiler. "With the Performance Advisor, we identified and eliminated bottlenecks, while the Query Profiler optimized queries to provide a better user experience. Efficient indexing and proactive insights delivered cost savings, and meticulous configuration tuning streamlined efficiency", says Lanzarini. “Picking up issues quickly means we can proactively prevent service disruptions that could impact the customer experience and damage our reputation,” adds Lanzarini.

With MongoDB Atlas, Agibank now has a high-performing, scalable, and resilient database platform underpinning its ECM system. This ensures customer identity checks and approvals are fast, seamless, and secure.

The architecture has been refined for maximum availability and downtime prevention, resulting in zero outages since the migration. Backup storage has evolved from a single location to multiple regions for enhanced resilience. Moreover, the team employs predictive analytics to anticipate demand surges, proactively scaling the infrastructure to prevent bottlenecks.

“We’ve had spectacular results with MongoDB. It’s 4.8 times faster than PostgreSQL and 90% more cost-effective,” adds Lanzarini. “It’s much more efficient to run, and we’ve saved eight hours of management time per month. Scaling on demand is particularly useful for coping with the growing volume of requests,” reinforces Lanzarini.

The schemaless design of MongoDB Atlas is also more flexible and enables the bank to store more complex and varied data. This flexibility, alongside the cost efficiency, high performance, and scalability, means the vast majority of new developments are built on MongoDB Atlas — and now that Agibank developers are experienced at using a non-relational database, they’ve significantly reduced time to market.

Marcos Lanzarini, database specialist, Agibank