Why Application Rationalization Matters More Than Your AI Strategy

All right, so I had this conversation last Tuesday with a healthcare CIO (I think they were evaluating their AI roadmap), and we got into this debate about implementation costs versus value. And I realized something while we were talking: they were trying to layer AI on top of 47 different applications. It has got me thinking about why application rationalization actually matters more than your AI strategy in the short term.

Why This Conversation Keeps Coming Up

Here is what I keep seeing with customers: the AI conversation starts with use cases. Case summarization, predictive analytics, conversational insights—all the sexy stuff, right? And look, those capabilities are absolutely transformative. That being said, nobody wants to talk about the application portfolio underneath.

So when I am working with customers on AI implementations, we eventually get to this uncomfortable moment. They have got Service Cloud, Field Service, Health Cloud (maybe some legacy systems from an acquisition), and now they want to add Agentforce or Einstein capabilities across all of it.

But here is the thing: every additional application in your portfolio increases your AI implementation cost by roughly 15-20%. Not because the AI is more expensive. Because your data is fragmented, your processes are not standardized, and your change management complexity just multiplied.

The Hidden Tax Nobody Calculates

Let me unpack this a little bit. When you are running AI models (whether it is predictive scoring or generative summarization), those models need clean, consistent data. If you have got customer records scattered across multiple systems, you are looking at data integration work before you even get to the AI implementation.

I was looking at this healthcare client's architecture, and they had patient data in four different places. Four. Which means before we could even pilot case summarization, we needed to build integration layers, establish data governance, figure out which system was the source of truth.

Your AI is not just analyzing data. It is trying to make sense of your application chaos. And that is expensive.

A Two-Stage Approach

Okay, so the way I normally frame this with customers is sort of a rationalization-first approach:

One: In the first phase (and I am talking 60-90 days here), we audit the application portfolio. Not just what you have, but what is actually being used. I typically see 30-40% of applications getting less than 10% adoption. Those are rationalization candidates.

Two: Once you have consolidated to your core platforms, then you start layering AI capabilities. Because at that point, your data is cleaner, your processes are standardized, and your implementation costs just dropped by 40-50%.

The Constraint Nobody Talks About

Now, I have seen organizations try to skip this step, so let me be transparent about what happens. If you implement AI without rationalizing first, you are probably going to hit a wall around month four or five.

Either your data quality issues surface and kill adoption, or your integration costs spiral out of control, or (and this is the tricky one) you build AI capabilities that only work in one application while your users are working across three.

The math does not work. Your ROI projections assumed clean data. Your change management plan did not account for training users on AI across multiple platforms. And honestly, your IT team is probably spending 60% of their time just maintaining integrations instead of improving AI capabilities.

But your rationalized portfolio? If you have done that work first, your AI implementations move faster, cost less, and actually get adopted.

So Here is My Take

Before you invest heavily in AI capabilities, spend 8-10 weeks doing an honest application rationalization assessment. Identify your core platforms. Sunset or consolidate the rest. Build your AI strategy on top of that clean foundation.

The Question: If you layer AI on top of 47 applications without rationalization first, which month does your integration cost spiral begin?