Balancing trust in legacy systems with new tech

Banks’ biggest tech challenge isn’t upgrading legacy systems- it’s integrating an entirely new (Gen)AI layer with orchestration in the lead. And making it work across functions. Too many banks often start with the wrong focus. Whereas dealing with legacy infrastructure is inevitable, it can become a blind spot without the right understanding of what it needs to achieve. Delivering agile, intelligent services that anticipate customer needs should be the goal. Here is a high-level overview of how the back end can be adjusted: 𝟭. 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗲𝗻𝗴𝗶𝗻𝗲: -   An orchestration layer sits atop core systems, routing everything - from customer questions to fraud alerts - to the right AI service. -   Modern APIs abstract legacy systems into modular services, so AI features can be added or swapped without changing existing workflows. 𝟮. 𝗥𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲: -   Real-time data feeds stream transactions, balance changes and logins as they happen. -   A unified data hub brings together customer details, activity patterns and risk ratings so every AI tool works from the same information. 𝟯. 𝗗𝗮𝘁𝗮-𝗱𝗿𝗶𝘃𝗲𝗻 𝗶𝗻𝘀𝗶𝗴𝗵𝘁𝘀: -   Requests are automatically enriched with live account balances, recent transactions and open support tickets - ensuring the AI’s output reflects up-to-date information. -   Data is fetched on demand from indexed records, so the AI stays current without the expense of retraining the entire model for every update. 𝟰. 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 & 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲: -   Data stays encrypted end-to-end, from intake to AI output. -   Automated audits flag bias and log every decision. -   Failure simulations uncover hidden risks before they impact customers. 𝟱. 𝗠𝗼𝗱𝘂𝗹𝗮𝗿 𝘀𝗲𝘁-𝘂𝗽: -   Modern interfaces turn core banking, payment and CRM systems into plug-and-play modules. -   Behind the scenes, back-end services can be updated piece by piece without interrupting the AI layer. 𝟲. 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗲𝗱 𝗱𝗲𝗹𝗶𝘃𝗲𝗿𝘆 𝘁𝗲𝗮𝗺𝘀: -   Small, cross-functional teams manage everything from data ingestion to model deployment and monitoring. -   Defined roles and fast feedback loops keep projects compliant and focused on real customer needs. The GenAI layer doesn’t just sit on top of the existing setup – it’s a complete overhaul of the tech architecture and the business logic behind it. Opinions: my own, Graphic source: BCG 𝐒𝐮𝐛𝐬𝐜𝐫𝐢𝐛𝐞 𝐭𝐨 𝐦𝐲 𝐧𝐞𝐰𝐬𝐥𝐞𝐭𝐭𝐞𝐫: https://lnkd.in/dkqhnxdg

Everyone talks about AI like it’s magic—just sprinkle some algorithms, and suddenly entire industries transform overnight. I wish it were that simple. Because there’s one reality I keep coming back to, over and over again, in my work with executives, founders, and investors trying to build the “next big thing”: the law of LEGACY. People hear “legacy” and think old code and systems, Excel spreadsheets or dusty filing cabinets. But it goes much deeper. Legacy means the roads, bridges, and buildings designed for yesterday’s tools and assumptions. But it also means the societal cultural norms and the legal systems built for a pre-digital world, grinding forward at a snail’s pace while tec races ahead. The architecture of our economies is also legacy—capital flows, pensions, debt—that can’t just pivot overnight without chaos. Lastly and most importantly, legacy is our link with past through present and towards the future through the mental models running in our heads, shaping how we judge risk, trust information, and react to change. Legacy is humanity continuity law. None of these swap out with an API call. This is why we live in a two-speed world when it comes to AI bit we fail to acknowledge iti. On one hand, I see AI models evolve every few months, while the companies I advise wrestle with ERP systems from the 90s or regulators still operating in paper files. I’ve watched founders pitch groundbreaking products that look flawless in a demo—but fall apart the moment they hit real-world legacy constraints. Investors often underestimate how expensive and time-consuming it is to navigate these layers. That’s why innovation pilots stall. Why impressive demos never scale. Why billions get burned chasing “transformation” that never lands. The legacy layers of our world are ignored. Yet here’s the twist: legacy isn’t just an obstacle. It can also be a moat. The people and companies who figure out how to bridge new tech with old infrastructure, laws, and cultural contexts build advantages that pure technical innovation alone can’t replicate. Of course in the process our legacy world gets an upgrade too, but it essential to fully account legacy. My advice to all those investing or building with AI - plan for the world as it actually is—in all its messy yet layered reality. To capture it well, think of it as cake, with different flavours that will give that unforgetable taste: • The top layer is digital tech—fast, modular, exciting. The juiciest. • The middle layer is institutional systems—laws, governance, rules. A bit dry but tasty • The bottom layer is physical & cultural legacy—bricks, roads, norms, beliefs. The most solid of layers with vintage taste, no soggy bottom. When you cut through top to bottom layer, each piece has all the right balancing flavour for the unforgetable Bake Off taste. I am curious how does your cake looks and taste like lately?

Balancing AI-Driven Modernization with Human Oversight Two articles from The Wall Street Journal caught my attention today. One article focuses on the enterprise modernization opportunity, while the other emphasizes the importance of human involvement and AI alignment. In my previous post¹, I discussed the challenges faced by enterprises in dealing with complexity. I encouraged enterprises to consider key questions, such as (1) whether AI can simplify business processes and (2) if it offers an easier way to navigate through complexity. Enterprises are at a pivotal moment in how they approach legacy systems and business process modernization. Morgan Stanley’s recent deployment of DevGen.AI², an internal generative AI tool built on OpenAI’s GPT models, exemplifies the power of AI in addressing one of the most challenging problems for large enterprises: translating and refactoring millions of lines of legacy code into modern languages. According to Morgan Stanley’s global head of tech and ops, Mike Pizzi, this initiative has already saved developers an estimated 280,000 hours this year. This can potentially help the company to streamline operations, reduce technical debt, and accelerate transformation without requiring significant manual effort. Morgan Stanley is translating legacy code into plain English specifications, which they can then use to replace COBOL, PERL, and Assembler systems with modern programming languages. This is an intelligent first step in answering the questions I raised in my previous post. The full benefits will only be realized when business processes are redesigned and reimagined using these specifications. Morgan Stanley is also wise in not to overlook the human element. The second WSJ article warns that advanced AI models are beginning to exhibit unexpected autonomy, such as rewriting their own code to evade shutdown commands³. These developments underscore the need for robust human oversight and clear boundaries to ensure that technology remains aligned with organizational goals and ethical standards. The best path forward is to modernize boldly with AI, but always keep humans in the loop. By combining cutting-edge automation with human judgment and accountability, enterprises can achieve both operational excellence and responsible innovation. References: 1.     https://lnkd.in/e7PXM7RN 2.     https://lnkd.in/ee9ycqvy 3.     https://lnkd.in/eFDfVEiy

"Why Some Organizations Still Embrace Legacy Systems... A thought-provoking observation from the field: Many enterprises aren't holding onto legacy systems just out of habit or fear of change. They're making calculated decisions based on legitimate concerns about modern alternatives: Performance isn't always better in newer systems. Sometimes simpler, purpose-built legacy solutions outperform their feature-rich successors. Modern systems often come with built-in tracking and telemetry that organizations may not want or need, raising privacy and data governance questions. Mandatory automatic updates can disrupt operations and remove control from IT teams who need predictable system behavior. Update mechanisms, while convenient, can create security vulnerabilities and potential backdoors that expand the attack surface. The shift from ownable software to vendor-dependent services means less autonomy and potential risks if vendors change direction or pricing. These aren't just theoretical concerns - they're real factors shaping enterprise architecture decisions. What's your take on balancing modern capabilities with organizational control? #EnterpriseIT #SystemArchitecture #TechnologyStrategy

Legacy systems often stick around longer than anyone plans. At first, they do the job, but over time, they start holding your business back. Many of our clients come to us facing this exact issue — old systems that can't keep up with their growing needs. The big question: how do you modernize without risking major disruptions? The first step is understanding what your legacy system still does well and where it’s holding you back. Not everything needs replacing right away. Focusing on the areas that are creating the most friction in your day-to-day operations will help you target your efforts. We often advise clients against ripping out an entire system all at once. Instead, we help them modernize in manageable steps. This approach spreads the investment over time and allows you to gradually replace outdated components while keeping your core business running smoothly. Moving data from a legacy system to a new platform can be one of the most complex parts of the process. We’ve helped companies navigate this challenge by developing clear migration plans that focus on data accuracy and integrity. Your data is the lifeblood of your operations, and ensuring it transfers correctly — without loss or corruption — is key to a successful modernization. One mistake we’ve seen businesses make is forgetting to prioritize security when modernizing legacy systems. Older systems tend to have vulnerabilities that modern threats can exploit, but simply moving to a new platform isn’t enough. Every upgrade needs to be paired with an evaluation of your security posture. Implementing new encryption methods, improving access controls and conducting regular security audits to protect your data and operations should be a priority in your modernization plan. Legacy system modernization is a journey, but when done thoughtfully, it can unlock new opportunities for growth, efficiency, and innovation. #flexmade #softwaredevelopment #legacysystems #digitaltransformation

A $1M client churn avoided without ripping out a single system. AI-powered SIP integration makes innovation painless. I watched a telecom reseller almost lose their biggest enterprise account. The client wanted new voice features, faster support, and 24/7 service but they didn’t want to throw away their existing investments or disrupt their business. We stepped in with UponAI SIP integration. No rip-and-replace, no forced upgrades. Our AI voice agents plugged right into their legacy infrastructure using SIP trunk, endpoint, and URI, whatever fit best. The transition was smooth, and the client saw zero downtime. The real win came from our modular approach. We layered smart call routing, contextual transfers, and queue awareness on top of what they already had. Suddenly, customers got answers day and night. Human agents handled fewer calls, but every call was faster and more personal. The provider kept their renewal revenue and strengthened their relationship. The enterprise client got the innovation they needed, without risk, without drama. That’s what I love about modular AI: it protects what works, while letting you grow. This isn’t theory. We’ve seen 40-60% reductions in call volume to human agents. We integrate with CRM, ticketing, POS whatever the business runs. And when you need scale, AI voice agents never sleep. I believe modern AI should empower, not erase, the systems you trust. When you preserve legacy investments and add intelligence, you win loyalty and future-proof your business. How are you thinking about AI upgrades and legacy investments in your own business? I’d love to hear your story. #AIVoice #UCaaS #PhoneBill ☎️

🚀 The Strangler Pattern: It’s the talk of the town in legacy system modernization—but how many are actually doing it? Spoiler alert: Not enough. Here’s the deal: The Strangler Pattern isn’t just a fancy term to throw around in meetings. It’s a practical, risk-managed approach to modernizing legacy systems that lets you build new features around the old, gradually replacing the legacy parts without pulling the rug out from under your users. But let’s get real. For all the hype, it’s rare to see it implemented effectively. Why? Because too many teams either don’t know where to start or they get bogged down in the complexities of their legacy systems. So, let’s cut through the noise with some actionable tips: 1️⃣ Start with Low-Hanging Fruit: Identify the parts of your system that are causing the most pain or are the easiest to replace. Begin by building new services around these components, gradually siphoning off functionality from the old system. Domain Driven Design tools like Event Storming are your friend! 2️⃣ Focus on Mission Value: Don’t just refactor for the sake of it. Target the areas that will deliver the most mission value. If your modernization efforts aren’t moving the needle, you’re wasting time. 3️⃣ Parallel Development: Run your legacy and new systems in parallel. This reduces risk by allowing you to validate the new system’s functionality before decommissioning the old one. It’s like having a safety net while you walk the tightrope. 4️⃣ Automate Testing and Deployment: Automation is your friend here. Use automated tests to ensure the new services work seamlessly with the old system. And automate your deployment pipeline to make the transition as smooth as possible. 5️⃣ Monitor and Iterate: Don’t just set it and forget it. Keep a close eye on the performance of both your old and new systems. Use feedback to continuously improve and gradually “strangle” the legacy system out of existence. 🏃♀️ Modernizing legacy systems is a marathon, not a sprint. The Strangler Pattern lets you pace yourself, but only if you commit to actually doing the work. It’s time to move beyond the buzzwords and start implementing. Who’s ready to stop talking about the Strangler Pattern and start using it? #LegacySystems #StranglerPattern #Modernization #TechDebt #SoftwareEngineering #DigitalTransformation #DevOps

If you work on a service that has non zero customers, chances are your projects are somewhat invovling migrating old to new, while keep the service running. The Strangler Migration pattern is a common model used to gradually migrate an existing service to a new system or technology stack. The key idea is to "strangle" the old system by incrementally replacing its functionality with the new system, similar to how a strangler fig plant grows around and eventually takes over an existing tree. This approach allows the migration to happen in a controlled and iterative manner, minimizing disruption to the existing application and its users. It involves creating a facade or proxy layer that routes requests to either the old or new system, gradually shifting more traffic to the new system over time. The Strangler Migration pattern is often used when the existing service is large, complex, or tightly coupled, service downtime is unacceptable or must be minimized, making a big-bang migration risky or impractical. It allows the new system to be developed and tested in parallel, while the old system continues to operate. Here are the key steps of the Strangler Migration process, specifically tailed for online services: 1. Prevention of New Dependencies * Stop new services from integrating with the legacy system * Ensure all new development connects to the new system * Establish clear guidelines for new development teams 2. Incremental Migration with Fallback * Gradually move existing dependencies from old to new system * Implement "kill switch" mechanism for safety * Allow quick rollback to old system if issues arise * Test each migration phase thoroughly * Monitor system behavior during transition 3. Complete Transition with Shadow Mode * Switch all use cases to the new system * Keep old system running in parallel (shadow mode) * Verify all functionality works correctly in new system * Compare outputs between old and new systems * Ensure no regression in business processes 4. Legacy System Decommissioning * Confirm all functionalities are working in new system * Verify no remaining dependencies on old system * Plan and execute resource cleanup * Document system retirement * Remove old system infrastructure If you are philosophy junkies like me, here is a bonus note: The Ship of Theseus paradox and the Strangler Fig Pattern are closely related concepts that deal with gradual replacement and identity. The Ship of Theseus is an ancient philosophical paradox about whether an object remains the same after all its components are gradually replaced. The paradox comes from a ship that had all its parts replaced over time, raising the question of whether it remained the same ship. Philosopher Thomas Hobbes asked - which ship would be the "original" if someone collected all the old parts and built another ship? Regardless what your answer is, migration is the only thing constant!

The older a platform gets, the harder it becomes to evolve. Not because it’s inherently better or worse but because of the layers of decisions, trade-offs, and expansions built up over time. Enterprise platforms in banking, fintech, and insurance aren't designed all at once. They’ve been expanded and modified for years, even decades. Add mergers & acquisitions into the mix, and suddenly, you're stitching together multiple systems, each with its own history and logic. The result? Technical debt. Most people think of technical debt as purely a bad thing. But in reality, technical debt isn’t a curse, it’s a tool. I’ve seen this firsthand working with hundreds of financial firms over 20 years. Technical debt (when used strategically) allows companies to move fast, delivering value before every piece of the system is perfect. But when ignored, it compounds, making innovation harder, slower, and riskier. So how do companies balance moving fast while managing technical debt? 1️⃣ Recognize that technical debt isn’t just a problem, it’s a strategic decision. 2️⃣ Build with awareness: technical debt is a tool, but only if it is managed responsibly (just like financial debt). 3️⃣ Align engineering and business teams so one isn’t slamming the brakes while the other floors the gas. 4️⃣ Modernize responsibly: replace what’s necessary, refactor what’s valuable, and don’t treat legacy systems as obstacles, but as histories of past decisions. If your Fintech is in need of an upgrade, the team at Praxent is here to help.

Are you slapping new processes on old problems? Because if you are—here’s why it’s not working. You can automate the workflow. You can build a shiny new system. You can roll out a new SOP and hold another team-wide training. But if the real issue is cultural, behavioral, or relational— No new process will fix it. It’ll just cover it up for a little while. Process should support people, not replace awareness. Most operational friction doesn’t come from the system. It comes from things like: -Misalignment -Avoided conflict -Unclear priorities -Lack of ownership -Unspoken tension between teams You know what happens when you ignore that and just throw a new tool at it? It works for 30 days… until the cracks start to show again. Here’s how to break that cycle: 1. Diagnose before you deploy. Ask: “Is this a process issue—or a trust issue in disguise?” Get honest before getting efficient. 2. Listen to the people who use the process every day. If your frontline team hates the system, no amount of leadership enthusiasm will fix it. 3. Stop over-engineering around poor communication. If teams don’t talk, no dashboard will save you. Fix the relationship before the reporting. 4. Measure outcomes, not just adherence. Just because everyone uses the process doesn’t mean it’s working. Look at results, not checkboxes. 5. Be willing to burn down what no longer serves the mission. Legacy systems are fine—until they’re not. Don’t be precious about broken things. Because at the end of the day, no process is a substitute for alignment, trust, and leadership clarity. Build the culture first. Then build the system that supports it.