AI Doesn't Fix Broken Workflows — Understanding the Problem Still Comes First

Right now, it feels like every software conversation starts with the same assumption: if something is inefficient, the answer must be AI. But in practice, adding AI to a workflow that hasn't been clearly understood can actually make the experience worse instead of better. Automation layered on top of unclear processes often amplifies confusion rather than removing it.

Research in process improvement and systems engineering has long shown that technology alone rarely solves operational inefficiencies without first mapping the underlying workflow. Studies in organizational design and digital transformation repeatedly emphasize that unclear processes, fragmented data, and manual workarounds are typically the primary causes of productivity loss — not the absence of advanced tools. When new technology is introduced before these root issues are understood, teams often experience increased complexity and adoption resistance rather than efficiency gains.

AI is incredibly powerful when used in the right places. It can help draft communications, summarize information, surface patterns, and automate structured decision paths. But it works best when applied to clearly defined steps inside a well-understood system. Used prematurely, it risks becoming just another layer of software people have to manage.

That's why my approach starts with understanding workflows first and tools second. The goal isn't to "add AI." The goal is to reduce friction. Sometimes AI is the right lever. Sometimes the best solution is simply clarifying the process. Getting that order right makes all the difference.