We’re pouring more capital into IT modernization than ever before, yet a frustrating paradox remains: technical debt keeps climbing, and tangible business value is getting harder to prove.
A lot of this disconnect comes down to blind, hype-driven investments. It’s easy to fall into the “technology-first” trap of adopting the latest architectural trends without clearly defining the actual business friction you’re trying to eliminate. When that happens, capital drifts away from the core vision and gets chewed up by peripheral tasks. You end up with modernized tech, but a business that hasn’t actually moved forward.
Right now, the industry’s default fix for tracking this value is to bring in specialized software like Apptio, MagicOrange, or Yarken. These tools are built to give you transparency into IT spending, usually by ingesting General Ledger (GL) data, operational metrics, and project logs.
The problem is that these platforms are incredibly rigid and highly opinionated. They enforce a specific worldview, like the standard set of processes and allocation models. If your company’s agile operating model, financial structure, or hybrid cloud architecture doesn’t fit perfectly into their predefined boxes, you trigger a never-ending cycle of data manipulation.
Teams often end up spending more time managing complex ETL pipelines just to appease the software’s mapping rules than they do actually analyzing the data.
The final output of all this rigid mapping is usually a massive suite of Business Intelligence (BI) dashboards. But dashboards have a built-in empathy gap. When a business leader looks at a chart, they are looking at answers to questions they didn’t necessarily ask. To get any real value, they have to reverse-engineer the dashboard creator’s logic to find the specific data point that solves their immediate problem.
Consider a real-world scenario: A CIO notices a 25% spike in cloud compute costs this month and needs to know if it actually drove business value.
The traditional software approach: The CIO has to filter a “Cloud Cost” dashboard, drill down into the infrastructure tower, map the server tags to a specific app, jump to a “Business Unit” tab to see who owns it, and finally cross-reference a Jira board to guess if a new feature launch caused the spike.
The AI approach: The CIO asks the data layer directly: “Did the 25% compute spike for our mobile app this month correlate with higher user transactions, or did we push unoptimized code?”
By forcing qualitative business context into structured cost tables, traditional platforms are essentially just BI tools acting as Value Realization engines. They are great at slicing historical data to show who spent what, but they completely miss the nuance of why the investment was made in the first place.
The future of IT value realization isn’t about building better dashboards; it’s about building intelligent data layers.
With the maturation of Large and Small Language Models (LLMs/SLMs) paired with retrieval-augmented generation (RAG), we can bypass these rigid ETL pipelines entirely. We can adopt a fundamentally query-first architecture.
Instead of pre-building visualizations for every possible scenario, leaders can interface directly with their operational and financial data using natural language, asking things like:
“Which of our infrastructure investments are actually moving the needle on our goal of zero-downtime?”
“Map our Q3 AI investments directly to our core objective of reducing operational latency.”
AI thrives here because it understands semantic context. It can connect the dots between a technical output (a line of code or a server instance) and a business outcome. It shifts the focus from managing opinionated reporting software to actually solving the value realization problem.
Conceptualizing a question-first approach is the easy part; the real value is in the execution. If we want to move past opinionated TBM platforms, we have to look under the hood and build a dynamic, query-driven engine that translates raw IT finance data into actionable intelligence natively.
In the coming scribbles, I’ll be taking this from concept to code, walking through the architecture and step-by-step implementation of how to actually build this out. Mind that I am along the way myself, so please feel free to provide feedback, criticise or let your thoughts out making this journey a worth-while for all of us.