There is a conversation happening in boardrooms right now that goes something like this: “We need AI. Let’s move fast.” And that urgency makes sense. The competitive pressure is real. The promise of AI-driven efficiency is genuine. But somewhere between the executive mandate and the first deployment, a critical question gets skipped over.
Is the technology underneath actually ready?
Most of the time, the honest answer is no.
The Foundation Problem Nobody Wants to Talk About
Organizations have been building and patching software for decades. Applications get added. Systems get integrated. Workarounds become permanent. Data gets siloed into formats that made sense in 2005 and make very little sense now. All of that accumulation has a name: technical debt.
Technical debt is not just a developer problem. It shows up in your business outcomes. It is why a simple change to a customer-facing process takes six months instead of six days. It is why your IT budget is 70 percent maintenance and 30 percent innovation, when it should probably be the reverse. It is why your best engineers are untangling old code instead of building new things.
And when you try to layer AI on top of all that, the debt does not disappear. It amplifies.
This is the core insight behind the McLean Forrester approach to application and portfolio modernization. Before an organization can become genuinely AI-powered, it needs to take a clear-eyed look at what it already has, decide what to keep, what to improve, and what to retire, and build from a position of actual readiness rather than wishful thinking.
Why AI Projects Keep Failing
The failure rates for AI initiatives are sobering. Research from multiple analyst firms puts the number of generative AI projects that fail to reach production at somewhere between 70 and 95 percent. That is not because the technology does not work. It is because the conditions for success were never in place.
Think about what AI actually needs to function well. It needs clean, well-organized data. It needs reliable integrations between systems. It needs workflows that are documented and understood. It needs infrastructure that is stable enough to support something new on top of it.
Legacy environments fail on almost every one of those requirements. Data is scattered, inconsistent, and often trapped in formats that modern tools cannot easily access. Integrations are fragile. Workflows are tribal knowledge. And the infrastructure is being held together by people who remember why decisions were made fifteen years ago.
When you point an AI model at that kind of environment, you do not get transformation. You get a faster way to surface all the existing problems, plus a few new ones the AI invented on its own.
The Portfolio Modernization Framework: Seeing What You Have
The McLean Forrester methodology starts with something deceptively simple: actually understanding your current application portfolio.
That sounds obvious, but in practice most large organizations cannot tell you with confidence how many applications they have, which ones are actively used, which ones are redundant, or which ones represent genuine business risk if they fail. The portfolio grows organically over years of mergers, projects, and workarounds, and nobody ever stops to map it properly.
The REAP framework, which stands for reassess, extract, advance, and prune, gives organizations a structured way to do exactly that. It asks two questions about every application in the portfolio. First, how much business value does this deliver? Second, how technically healthy is it?
That two-axis view produces a clear picture of where you stand. Some applications are valuable and healthy, and you leave them alone. Some are valuable but technically fragile, and those are your modernization priorities. Some are neither valuable nor healthy, and those get retired. And some are technically fine but low-value, and those are candidates for consolidation or replacement.
The output is not just a list of applications. It is a prioritized modernization roadmap that ties directly to business outcomes. You know what to tackle first because you know what is actually constraining your most important capabilities.
What Modernization Actually Looks Like
Modernization is not one thing. The right approach depends on what you have and what you need.
For some applications, the right move is breaking a large monolith into smaller, more manageable services that can be updated independently. For others, containerizing the existing codebase is enough to make it portable and easier to maintain. For still others, the answer is a full rebuild using modern architecture patterns designed for cloud environments.
The common thread across all of these is that the goal is not just to make old software newer. The goal is to remove the friction that slows everything down. Clean APIs instead of brittle point-to-point integrations. Modular services instead of tangled dependencies. Documented logic instead of institutional memory. Data that is organized and accessible instead of trapped.
When you do this work, two things happen. Your development teams become dramatically more productive because they are not fighting the environment anymore. And your organization becomes genuinely ready for the AI investments you want to make, because the foundation can actually support them.
AI Can Help You Get There
Here is something worth knowing. The AI tools you want to deploy can also help you with the modernization work itself.
Modern AI can analyze large legacy codebases and map dependencies in ways that would take human developers months to accomplish. It can generate test coverage for systems that have almost none, which is often one of the biggest barriers to making changes safely. It can identify patterns in old code and suggest modern equivalents. It can document systems that were never properly documented.
This creates a useful flywheel. As your codebase gets cleaner and better documented, AI tools become more effective at working within it. As AI tools become more effective, modernization moves faster. The two efforts reinforce each other.
The important caveat is that AI-assisted modernization still requires human expertise and judgment. Someone needs to validate that the business logic was preserved correctly. Someone needs to make architectural decisions about how new services should be structured. Someone needs to ensure that security and compliance requirements are met throughout the process. AI accelerates the work but does not replace the people doing it.
Starting Without Waiting for Perfect
The temptation when facing a large modernization effort is to wait until you have a perfect plan. Do not do that. The portfolio assessment itself will surface things you did not know, and you cannot plan for what you have not discovered yet.
A better approach is to start with the assessment, identify your highest-priority modernization targets based on business value and technical risk, and begin there. You do not need to modernize everything before AI can deliver value. You need to modernize the things that matter most to the capabilities you want to build.
At the same time, there are quick wins available that do not require full modernization. Retrieval-augmented generation, or RAG, lets you connect AI tools to your existing knowledge and data without retraining a model from scratch. That can deliver meaningful value in the short term while the longer modernization work proceeds.
The goal is intentional sequencing, not perfection. Move deliberately, prioritize based on evidence, and build toward a foundation that can actually support what you want to do next.
The Real Competitive Advantage
The organizations that will pull ahead in the next decade are not necessarily the ones with the largest AI budgets. They are the ones that did the unglamorous work of cleaning up their foundations before trying to build on them.
When your application portfolio is well understood, well maintained, and built on modern architecture, everything else becomes easier. Your teams build faster. Your data is a genuine asset. Your AI investments land where they are supposed to instead of sinking into a swamp of technical debt.
Application and portfolio modernization is not a detour around AI. It is the path to AI that actually delivers on its promise.
Ready to assess your portfolio? The McLean Forrester REAP framework gives organizations a practical starting point for understanding what they have and building a modernization roadmap that connects directly to business outcomes. Visit mcleanforrester.com to learn more.
Top comments (0)