Robert Cooke, CEO of 3forge, recently shared some interesting insights with CI. The team at 3forge pointed out that currently, global banks run on billions of lines of legacy code, some even written half a century ago, and nowhere is that inheritance more visible than on Wall Street. This seemingly leaves CTOs and their dev, security, and IT teams spending their time trapped in managing fragile legacy systems just to stay afloat.
In fact, 70% of bank IT spend globally is tied up in maintaining, patching, and reconciling legacy systems.
Robert Cooke, CEO of 3forge, takes a historical perspective on legacy systems that few consider: In the last 100 years, every major software breakthrough was built on what came before. Recognizing that banks are trapped by decades of layered legacy systems and technical debt, he claims to have developed an approach now said to be trusted by some of the largest banks.
As AI enters the picture, it’s critical that it interacts with data that is governed, auditable, and reliable, providing the foundation for meaningful automation in regulated markets. After all, while AI is a major advancement, it’s also another step in the century-long evolution of the software stack.
During our conversation, Robert talked about how banks can safely modernize their legacy systems without losing the lessons of the past. Robert touched on how banks can integrate legacy and modern systems seamlessly though data virtualization and real-time streaming, enabling modernization without disruption.
Our conversation with Robert Cooke is shared below.
Crowdfund Insider: How can digital transformation and digital technology continue to enhance user experience?
Robert Cooke: In finance, “user experience” (UX) is more than just sleek interfaces. It’s the ability for clients and employees to see and act on accurate information instantly. The most significant UX improvement comes from real-time, governed data that unifies systems long separated by silos.
By virtualizing access to legacy platforms and streaming data in real time, banks can replace overnight batch cycles with dashboards that refresh continuously, letting traders, risk teams, and executives share a single trusted view. This approach not only removes the friction of reconciling multiple systems but also reduces latency and errors.
Traditionally, financial institutions have separated back-end and front-end development, which slows the delivery of new functionality and locks the user experience to what legacy technology can support.
A unified full-stack model closes that gap by allowing frontline users, developers, and even business teams to co-evolve interfaces and data logic in real time. That convergence produces experiences that evolve with how teams actually work: collaboratively, securely, and under the same governed data model.
In practice, institutions using 3forge have built unified interfaces that handle millions of updates per second with 99.99% uptime, enabling smooth interaction even during market surges.
A hedge-fund client leveraged this model for the largest hedge fund launch since 2018 (over $5B AUM), creating more than eighty custom dashboards updated in real time and augmented with new user-driven features several times per week, which meant faster decisions and better collaboration across global teams.
Crowdfund Insider: What challenges do legacy systems pose and how can they be overcome?
Robert Cooke: Every decade of computing has left behind an architectural fingerprint, and financial systems have inherited them all. I recently wrote the foreword to a book called The Stack: How Layers of Software Built the Modern World, which discusses how, in the past 50 years, almost every major software breakthrough, including COBOL, UNIX, C, C++, and Java, has not been a clean replacement of the last layer but an addition to it.
Grace Hopper’s compiler in the 1950s made code portable; UNIX in the 1970s introduced shared, multitasking environments; and Java in the 1990s abstracted hardware differences through the “virtual machine” concept.
Each solved the problem of its time by building on what came before. That same layering instinct made today’s software possible, but it also explains why banks now run on billions of lines of code, some written fifty years ago.
In banking, that accumulation manifests in an “inheritance trap.” Mainframes, adopted for their reliability, still process most global settlements. COBOL, chosen for its clarity, now spans hundreds of millions of lines across just one institution. Forty-five of the world’s fifty largest banks still rely on IBM Z somewhere in their core stack.
Stability, not inertia, keeps those mission-critical systems alive: they deliver the “five nines” of uptime regulators demand, but the side effect is brittleness. Every new cloud deployment or mobile interface must reconcile with half a century of architecture beneath it. Even trivial changes, such as altering a counterparty risk limit, can ripple through middleware written before the Internet existed.
So, how do we solve the inheritance trap? The answer is not a “big-bang” rewrite. History proves those efforts collapse under their own weight. The solution lies in shielded modernization, a concept we’ve engineered directly into our architecture. Think of it as applying the lessons of software history—abstraction, modularity, portability—to the enterprise level.
The challenge isn’t only technical; it’s structural. Without proper abstraction, new systems tend to inherit the fragility of what came before. Shielding innovation from legacy through virtualization layers prevents this ‘choking effect,’ ensuring modernization frees resources instead of reproducing old constraints.
The first step is to shield the new from the old with a high-performance data-virtualization layer. Rather than connecting every new app to legacy directly, you standardize access through a single, governed interface. That preserves reliability while enabling modern APIs, real-time streaming, and audit controls.
Next, replat wholesale. Progressive decommissioning, retiring bounded clusters or workflows, creates a virtuous cycle. Each sunsetted system frees budget and talent that can be reinvested in the next. A global money center bank currently deploying this technique illustrates this pragmatism with a six percent-per-year retirement cadence.
The final step is to architect for rapid scale so modernization compounds instead of recreating fragility. Through real-time streaming instead of overnight batch, unified observability of data lineage and latency, containerized elasticity, and cell-level governance built in rather than bolted on, financial institutions can take their first step in their modernization journey.
Revisiting the historian’s perspective, each generation of software didn’t erase the previous one: it abstracted it. In the same way, banks can’t escape their inheritance by starting over; they must transform it into an asset. With data virtualization, real-time processing, and composable dashboards, modernization becomes continuous rather than catastrophic. That’s the only sustainable way to break free from legacy lock-in while keeping the systems that make modern finance possible.
Crowdfund Insider: What is your key focus area as we approach the end of 2025? And why?
Robert Cooke: As we close 2025, our focus is squarely on enabling financial institutions to modernize safely in parallel with their AI ambitions. The industry’s challenge isn’t a lack of algorithms; it’s decades of legacy infrastructure that make data fragmented, inconsistent, and hard to govern.
We’re concentrating on the layer beneath AI: real-time, auditable data systems that unify operations across trading, risk, and compliance. That foundation of governed data, observability, and resilient architectures is what ultimately makes AI both possible and trustworthy in regulated markets.
These priorities mirror what CTOs across the industry are identifying as the most significant obstacles to modernization: addressing budget gravity, stack fragmentation, and resilience risk. Modernization isn’t just a technical foundation. It’s a business driver that improves productivity, profitability, and risk management.
At the same time, we’re continuing to harden our platform’s security and governance, already SOC 2 Type II certified, and address demand for end-to-end encryption, and granular permissions. That way, modernization never compromises compliance. In short, 2025 is about building the foundation that allows regulated institutions to use AI safely, efficiently, and at scale.
Crowdfund Insider: What are your main objectives for the coming year and how do you plan to achieve them?
Robert Cooke: Our priorities for the year ahead are closely aligned with what CTOs across global financial institutions are already signaling in their roadmaps: shifting from “AI-first” hype to modernization-first, AI-enabled reality.
IT budgets are tied up in maintaining and reconciling legacy systems, and most leaders now view modernization as the critical precondition for meaningful automation. You can’t run reliable AI on unreliable plumbing, so we’re helping these tier-1 banks free budgets from “run the bank” maintenance to “change the bank” innovation.
Our focus is therefore on helping firms modernize beyond AI, creating the real-time, governed data infrastructure that makes AI productive and safe rather than aspirational. That means continuing to build technology that helps CTOs execute three key steps for incremental modernization:
Shield the new from the old through data virtualization and abstraction, allowing innovation to occur without constant friction from legacy dependencies.
Replace iteratively, freeing up resources in measurable, low-risk phases rather than in one costly “big-bang” rewrite.
Architect for rapid scale, where performance, governance, and observability are embedded from the start instead of patched in later.
We’re also seeing CTOs reframe modernization not only as an engineering task but as a business efficiency strategy. When banks can shorten delivery cycles, improve developer productivity, and retire redundant systems, return on equity can rise by one to one-and-a-half points over two years. That’s where our energy is concentrated, helping Wall Street measure progress not just in uptime or latency, but in tangible business outcomes.
Finally, we’ll keep deepening the areas that modernization immediately impacts: real-time risk oversight, data lineage and auditability, and governed analytics.
Those are the operational foundations that make it possible for AI to be truly useful, explainable, and regulated in markets. Our goal for 2026 is to make those capabilities so seamless and reliable that AI adoption feels like a natural, incremental next step, not a leap of faith.