A view from the work
AI transformation starts with what a business could become when its people can do more.
I'm interested in the combination: human judgment and creativity, amplified by the reach, speed, and capabilities of AI. Then the harder, more valuable part - changing how the organization works around it.
Start with the principles01 / First principles
Some of yesterday's work still needs doing. AI can help do it faster and better. What excites me is the space that opens up: deeper thinking, stronger relationships, and projects that previously felt out of reach.
The transformation becomes real when that new capacity changes what the business can accomplish.
Revenue, costs, innovation, and risk give the work a purpose. I define using AI well by its contribution to those outcomes. Speed and proficiency matter through what they make possible.
What becomes better for the business?
Read the principleJudgment, creativity, relationships, trust, and domain expertise give the work direction. AI adds reach, synthesis, coding, and speed. The interesting work is designing how those contributions fit together.
Where does human attention matter most?
Read the principleA great design system makes expert judgment available to an entire team. Every function has its equivalents: examples, conventions, context, and standards that deserve to be written down, modeled, and maintained.
Could someone new recognize good work?
Read the principleA shared vocabulary and training give everyone a place to start. Sharing today's good workflows helps colleagues now; growing tomorrow's experts requires decision practice, feedback, mentoring, and responsibility.
Can the next person make this work?
Read the principleStart with a real problem, a time boundary, and a learning goal. Make it safe to surface what went wrong. Celebrate useful discoveries and successful outcomes, then decide what deserves another round.
What evidence would change the plan?
Read the principleLearning takes effort before it releases capacity. Leaders own that tradeoff. Clear priorities, credible support, and rewards for excellent work help the change become part of how the organization operates.
Who has the authority and support to carry it?
Read the principle02 / The human transition
Different roles experience different stakes. I start by making room for the concerns behind questions like these, then asking what leadership needs to make possible.
The board wants an AI return. Where do I start, and which pilots deserve more?
Leadership names the outcome, chooses among competing bets, and determines how outcomes change strategy and work.
How do I keep delivering while my team learns a different way to work?
Make priorities explicit. Decide what pauses, where support comes from, and when a bounded stretch ends. Don’t leave the tradeoff to individual heroics.
Will I still have a role? How do I learn if the starting tasks disappear?
Create skill building opportunities. Be candid about future staffing, and ensure a path to emerging roles and different responsibilities. Pair access with cultivating expertise.
Does my expertise still matter? Why am I explaining the same corrections again?
Invite experts to model excellent work and challenge weak conventions. Codified subject matter expertise compounds quickly through AI systems.
Staffing decisions differ. I won’t promise jobs are safe when leadership hasn’t made that commitment. I do expect candor about the choices, practical support, and room to name what the change costs. Non-use can also mean habit, friction, poor fit, or a tool that hasn’t earned trust.
Make learning part of the work ↗03 / Meet the actual constraint
The next useful move depends on the situation. These are five places I start looking.
An entry point for a conversation, not a maturity score. Several of these can be true at once.
A useful starting point
I start with a recent piece of real work: where time went, where judgment mattered, and what the team wished it could do. Existing experiments and useful tools belong in that picture too.
A useful starting point
A power user can create something remarkable. Getting the next person to use it well involves a different kind of work: examples, documentation, practice, and a way to ask for help.
A useful starting point
Too many experiments can consume the attention needed to make any one useful. I review the bets together, including the delivery and learning load on the same people.
A useful starting point
I watch someone try the tool before prescribing more training. Habit, missing coverage, repeated corrections, support, and trust can each get in the way.
A useful starting point
A faster first draft can still wait three days for approval. More research can still leave a decision untouched. I look for the step that now limits the whole workflow.
04 / From capability to value
I look for a visible connection between the work that changes and the outcome it serves. These illustrative patterns show what that can look like.
Examples of a method, not reported client results.
Research, scattered notes, and follow-up preparation consume the time available to talk with customers.
AI prepares a sourced brief, drafts the follow-up, and captures what happened. The person brings curiosity, judgment, and the relationship.
People gather numbers and rebuild the same report. Someone else has to work out what needs attention.
A repeatable workflow assembles approved data, flags exceptions, and points to the evidence. An owner decides what to do next.
A promising idea waits for enough research, design, or engineering capacity to make it tangible.
AI helps explore alternatives and build a prototype. People test it with customers, challenge the assumptions, and decide what is worth pursuing.
If a new output adds reading, checking, and reconciliation, it may move work around rather than remove it. I want to know which step it replaces and what it helps someone decide.
Read: From more output to better outcomesWhat this has opened up for me
I brainstorm with AI to sharpen my thinking before I commit to a plan. I build tools and prototypes I couldn't have coded before. I can investigate more deeply, cross-check my conclusions, and keep several independent workstreams moving.
A shared design system lets me turn that work into polished customer materials. A knowledge base lets me answer more of my own questions without interrupting colleagues. Captured conversations free up mental space for the next one.
Those are changes in capability. The organizational question is how to make them available beyond the person who first figured them out.
Explore the craft of making it transferable05 / Questions I’m working on
I bring experience to these choices, and I’m still learning where the boundaries belong. These questions change what I’d build and how I’d lead.
I’m exploring which decisions a learner needs to make, explain, and revisit. Faster production changes the apprenticeship; it doesn’t tell me what should replace it.
Design the learning work ↗I want to distinguish a useful feature from the trust, relationships, and distinctive assets that make a business valuable. AI may strengthen those assets or help create new ones.
A sound boundary today may become needless friction, while a familiar rule may preserve a bad habit. I want evidence for changing the boundary without losing accountability.
Choose authority and oversight ↗Go a layer deeper
Two ways to keep thinking.
What makes this stick
I look for three conditions that help the change last: a sponsor who funds the transition and resolves the priorities it collides with, protected time for people to learn on real work, and an internal owner who can change and repair the workflow after the experiment ends. The field guide covers how to make room for the experimentand how to prove the ownership.
Situational choices about value, standards, authority, learning, adoption, and ownership.
The deeper argumentWhere new capacity goes, what gets in its way, and how to recognize business value and possible competitive advantage.
This worldview draws on my own work and on practitioners whose ideas help sharpen it. Their accounts offer useful patterns, with different contexts and limits.
Let's compare notes
A business priority, a stubborn workflow, an experiment with promise. That's enough to start a useful conversation.