The previous unit established that the real barriers to AI adoption are bandwidth, literacy, and trust — not willingness — and that leaders must clear the lane before demanding adoption. But even once people have time and support, there's a second trap that catches most organizations: rushing to automate processes they haven't actually mapped. The panel was blunt about this, calling detailed process mapping the "do not pass go, do not collect $200" prerequisite before any AI deployment begins. This unit digs into why that mapping is so hard, what to do about the institutional knowledge holders who make it uncomfortable, and why the real opportunity isn't faster processes but fundamentally reimagined ones.

In the session, the panelists' vivid description of what lurks beneath every workflow you want to automate: "a lot of processes are held together by tacit human knowledge and implicit human knowledge that holds these processes together. And nowhere in organizations are they mapped accurately or comprehensively." Alexis Gonzales-Black crystallized this with the example of "Brenda" — the person everyone relies on but nobody has documented:

Gonzales-Black: "Brenda just knows how to break our pricing structure for our clients. She just knows. We just trust Brenda."

The point wasn't that Brenda is a problem — it's that unmapped Brenda is a problem. You can't automate what you haven't documented, and you can't document what only one person understands unless that person has a reason to share.

Building on this, Gonzales-Black asked the question every leader should be asking:

Gonzales-Black: "Who does this problem serve? Who wants it to stay this way?"