Perspectives
Written by practitioners.
For people doing the work.
These are not trend pieces. They are working arguments about problems we encounter repeatedly — on cost modeling, AI governance, federal acquisition, and the operational discipline that separates programs that deliver from programs that report.
Featured
The Allocation Model Problem: Why Most FinOps Programs Fail Before They Start
Organizations invest in FinOps tooling and then discover the real problem was never the tool — it was the cost model underneath it. A look at what actually separates programs that produce defensible numbers from ones that produce reports nobody trusts.
The Governance Gap Is Not a Technology Problem
84% of AI initiatives fail to scale enterprise-wide. The technical capability exists. The models work. What's missing is the accountability infrastructure that makes deployment defensible to leadership, auditors, and oversight bodies.
Read the piece Federal AcquisitionOTA Prototype Authority: What Program Managers Get Wrong
Most program offices treat OTA as a contracting workaround. It isn't. It's a congressionally designed acquisition strategy for exactly the kind of commercial AI and software capabilities agencies are trying to deploy now.
Read the piece StrategyThe Best Solution Is the One That Fits the Problem
Not the most sophisticated solution we know how to build. Not the simplest one we can justify billing for. The calibration between those two poles is where most technology programs succeed or fail before a line of configuration is written.
Read the piece ITFM · FederalWhat the FITARA Scorecard Measures — and What It Misses
Federal agencies have spent years chasing FITARA grades. The scorecard measures CIO authority, data center consolidation, and cyber hygiene. It does not measure whether leadership can answer a basic question: what does our technology portfolio actually cost?
Read the piece Automation · AIAutomation Without AI: The Case for Keeping It Simple When Simple Works
Not every automation problem needs an AI component. Some of the highest-value workflow improvements are built on rule-based logic and conditional routing — no model required. The compulsion to add AI where it isn't needed is its own form of gold-plating.
Read the piece AI Adoption · FederalWhy AI Pilots Don't Scale: The Program Management Argument
The 84% failure-to-scale rate isn't a model problem. It isn't a data problem. It is almost entirely a program management, organizational design, and governance problem — and the fix doesn't start with the technology.
Read the piece