Historically, the industry has evolved from trust based on the documents of the Medici Bank to the digital immediacy of systems like Pix, where trust is now digital and verifiable. However, for many institutions, progress slows when it collides with legacy systems that carry decades of outdated design decisions and hard-to-trace dependencies.
Legacy systems are often thought of as problematic simply because they run on outdated technology. But the real issue lies in the accumulated complexity they sustain. Business rules embedded in code and critical knowledge concentrated in a few individuals make modernization projects slow, costly, and often perceived as a leap into the unknown. A clear example of this comprehensive transformation can be seen in the digital onboarding of AFOREs in Mexico, where regulatory pressure forced a redesign not only of the digital layer but also of core systems to remain competitive. This involved rethinking channels, operational processes, cybersecurity controls, and, most importantly, the core systems that support the business. Organizations that understood this transformation holistically are now leading the sector, while those that did not continue operating on a foundation that limits their ability to evolve.
This is where artificial intelligence emerges as a catalyst for results, helping to understand and unravel the deep complexity of these systems. AI models can analyze large codebases and map dependencies, eliminating the uncertainty of whether a small change will trigger a cascade of errors. However, this technology should not be seen as an additional layer that masks existing systems, but rather as a tool to truly understand them: how these legacy structures work, why they exist, and how they can be replaced with more modern solutions.
At Baufest, for example, we use AI to extract business rules directly from source code, enabling the documentation of critical knowledge and reducing risk before any transformation begins. This approach turns modernization from an uncertain event into a guided and objective process.
Successful legacy modernization efforts are rarely full replacements; they are incremental processes of refactoring and progressive migration. AI proves particularly effective in critical tasks such as test generation. A recent Baufest case showed how a bank increased testing coverage from 10% to 70% in just four weeks through the use of agents.
Additionally, modernization impacts platform operations. In regulated environments, AI can analyze metrics and logs to detect anomalies before they escalate, improving resilience and business predictability.
A strategic approach
Despite its potential, only 5% of companies are able to capture value from AI at scale. Many remain stuck in the experimentation phase, with pilot programs that never reach production due to a lack of clear architecture or proper governance. For these initiatives, it is essential to design processes that allow for rapid progress while maintaining the auditability and security required in the financial sector.
A practical AI-driven modernization path typically follows three steps. First, understand the real system—the one actually in operation, not just the documented version, which often fails to reflect patches and workarounds built over time—and the role of each of its components. Second, define a clear order of priorities when intervening, choosing how to implement changes based on their impact to avoid unnecessary disruptions. Third, execute the transformation incrementally, generating value from the earliest stages. This becomes an iterative cycle that continuously refines itself to align with the organization’s operations.
The lesson from multiple AI adoption cases is clear: the value does not lie in moving fast, but in doing it right. It is about leveraging AI in the best possible way to turn it into a strategic asset for continuous modernization.
Ultimately, it is about using AI to reduce the uncertainty of change. Because in the end, the organizations that will lead are not those that experiment the most, but those that succeed in transforming complexity into a foundation for evolution—with control, speed, and a clear business focus.
