AI Is Holding a Mirror to Your Organization — Are You Ready?

David begins with his product practice inside organizations: understanding what employees need, why business processes behave as they do, and how a daisy chain of internal solutions supports client-facing value. The work is less about delivering a preboxed tool than designing around the behavior that keeps an organization functioning.

That perspective shaped his ProductCamp PHX session. Speaking to product developers, managers, and owners, David argued that AI makes human-computer interaction newly important. Product teams need to examine conscious and unconscious behavior, question their initial assumptions, and design for how people naturally communicate and make sense of work.

Storytelling becomes a product-design input. A form produces controlled data, but it can lose the upstream experience, downstream consequences, and workarounds behind the entry. David proposes voice-first interactions—such as a person dictating a project update—so AI can organize the raw account into status, risks, and actions without forcing the person to compress the story prematurely.

The conversation then turns to the organizational behavioral record: emails, chats, meetings, files, and collaboration data that preserve how work actually happens. AI makes patterns in that record easier to examine, creating an opportunity for organizational self-realization while raising questions about access, interpretation, intent, and accountability.

David explains that An Inbox Between Us began as a private journal written while he was designing AI solutions. A conversation with an employee who feared losing her job clarified the book’s purpose: explain why people still matter, why jobs contain far more ambiguity and judgment than their task lists show, and why AI adoption is a relationship rather than a replacement event.

For individuals, the starting exercise is deliberately simple: dictate the honest story of how work happens, including frustrations, adaptations, and recurring questions, then use AI to identify patterns worth investigating. Dave describes a related practice—building a personal AI thinking partner—while the group stresses that AI is not a therapist, boss, doctor, lawyer, or final decision-maker.

The boundary becomes consequence. AI can identify signals and suggest paths, but it has no lived experience or lived consequence. Low-risk assistance may be delegated more freely; as impact grows, interpretation, judgment, and accountability must stay visibly human.

For product teams, David offers a deliberately contrarian test: use AI to figure out how not to use AI. Apply probabilistic models where ambiguity requires them, but prefer deterministic steps where the outcome can be made reliable. Use capable models for exploration and planning, test whether smaller models can handle production volume, preserve repair knowledge, and design for repeatability rather than novelty alone.

The episode closes with authorship and craft. David describes using AI to organize raw notes and expose structural choices, while reclaiming the point where his own voice must take over. The same principle applies to product work: tools can reduce friction, but teams must protect the thinking, foundational knowledge, and human originality that make the result trustworthy.