\n<\/p>\n
Daniela Binatti is the founder and CEO of <\/em>Pismo<\/em><\/a>.<\/em><\/p>\n Artificial intelligence is one of the most discussed topics in business today. Across industries, companies are exploring the potential of AI, automation and intelligent agents to drive efficiency, innovation, and growth. According to Stanford University\u2019s Human-Centered Artificial Intelligence Institute<\/u><\/a>, generative AI reached 53% adoption among the global population within three years of ChatGPT\u2019s launch\u2014faster than either the personal computer or the internet achieved comparable levels.<\/p>\n Yet an important reality often goes unspoken in these conversations: many organizations are trying to deploy modern technologies on infrastructure that was built decades ago. According to Gartner<\/u><\/a>, outdated systems consume more than 75% of IT budgets at many financial institutions, crowding out innovation and slowing transformation. I see this challenge frequently in financial services, but it extends across industries. Much of the world\u2019s core infrastructure was designed for an era of limited computing power, bandwidth and storage, often relying on mainframe-based architectures or early-generation technology stacks that are still operating largely as originally deployed.<\/p>\n These systems were built for reliability, and in many cases, they still perform that function well. But they were designed for a vastly different technological environment. Today\u2019s digital economy depends on cloud computing, real-time data access, mobile experiences and large-scale data processing. However, many underlying platforms are still based on assumptions made decades ago. Over time, layers of patches, integrations and workarounds have added complexity, while the institutional knowledge of how these systems function \u201cunder the hood\u201d has faded as original architects retire.<\/p>\n This lack of visibility creates real risk. Even if a system appears stable, it becomes difficult to safely introduce new technologies on top of opaque, tightly coupled infrastructure. AI can deliver powerful results, but its impact will always be constrained if it is built on fragile or outdated foundations. That is why infrastructure modernization is not a parallel initiative to AI\u2014it is the prerequisite that enables it to scale, adapt and deliver meaningful value.\u200b<\/p>\n Many companies are adopting AI by simply layering new technology on top of existing technology. But that approach can create new risks rather than solving old ones.<\/p>\n Yes, AI can make coding easier. As a result, many teams add new layers of code on top of the legacy systems they already have, instead of reviewing and strengthening the foundation. This approach often increases complexity and can introduce new risks rather than resolving existing ones. For example, if data is not well-organized, not easily accessible or duplicated in various systems, it will be difficult for AI technology to deliver accurate outcomes. Take something as basic as a customer\u2019s date of birth stored in different formats across multiple systems. Standardizing that data requires additional manipulation during migration, increasing both complexity and risk. Similarly, if companies do not understand their databases\u2019 ability to share information, they could be creating security issues or unpredictable operational costs.\u200b<\/p>\n This is one of the biggest reasons why companies could feel stuck in pilot mode when it comes to AI technology. The technology is good, but the environment in which it is built is not ready for it.<\/p>\n<\/section>\n When organizations hear the term \u201cinfrastructure modernization,\u201d many immediately imagine a massive migration project, pulling the plug on legacy systems and replacing everything at once.<\/p>\n This is not only unrealistic, but the safest approach is incremental.<\/p>\n In financial services, this approach is particularly important because many industry protocols still rely on standards that were established decades ago. Payment networks, for example, still use messaging protocols that date back to the 1980s.<\/p>\n Modern platforms can interact with those legacy systems through translation layers that convert old messages into new formats. This allows institutions to build modern core infrastructure while maintaining compatibility with existing networks.<\/p>\n Organizations can build modern infrastructure alongside existing systems and migrate capabilities gradually. This allows legacy and modern architectures to coexist during the transition, ensuring stability for customers and operations.\u200b<\/p>\n<\/section>\n Modernization is not so much a replacement of everything all at once but rather a rethinking of how the systems are actually structured. <\/p>\n If you simply take an old application and change the programming language or move the application from an on-premise environment to a cloud-based environment without rethinking the architecture and the processes involved, you are essentially re-creating the same problem just in a different environment.<\/p>\n Instead, what an organization needs to do is to take a step back and think more fundamentally: What actually needs to exist in the modern system?<\/p>\n For example, in financial services, many processes were originally designed around physical limitations that are no longer relevant. For instance, customers used to receive financial statements via mail, necessitating several days to close a billing cycle and generate the printed documents. <\/p>\n Now, with the advent of the Internet and evolving consumer demands, customers can (and expect) access to their financial information via a mobile app, in real-time. If you start to design the system without challenging those original assumptions, you are essentially re-creating the old processes in the new technology.<\/p>\n True modernization requires rethinking the product and the process before rebuilding the system.\u200b<\/p>\n<\/section>\n Infrastructure modernization is often viewed as a technical upgrade. In truth, it is a strategic decision.<\/p>\n Many legacy platforms are approaching end-of-life support, which introduces long-term security and operational risks. And organizations that lack visibility into their systems have limited control over how those systems evolve.<\/p>\n This is ultimately what separates organizations that successfully scale AI from those that remain stuck in experimentation. When we hire and onboard a new engineer, we teach the fundamentals and best practices. The same discipline should apply when using AI to develop systems. Without clearly defined guardrails, AI-generated code can dramatically expand the maintenance surface and introduce vulnerabilities.<\/p>\n AI is powerful, but it needs the right foundations. Companies that truly understand their infrastructure and are willing to redesign it thoughtfully are the ones that will unlock its full potential.<\/p>\n Because the future of AI is about the systems that make those algorithms possible.<\/p>\n Forbes Technology Council<\/u><\/a> is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?<\/u><\/em><\/a><\/p>\nWhy AI Cannot Fix Broken Foundations<\/h2>\n
Transformation Happens Incrementally<\/h2>\n
Modernization Is Not Just Migration<\/h2>\n
Infrastructure Is A Strategic Capability<\/h2>\n
\n
<\/section>\n<\/div>\n