Promoting AI Transformation at All Levels

Where You Are

How useful is AI in your work today? How much has it changed what you actually do? Hold up a hand and a number of fingers that reflect your level.

What You'll Leave With

If you could only walk away with three things, this is my short list.

Access is not adoption

Buying licenses doesn't magically cause AI transformation.

Start little, start fun

Find the smallest value add to get motivated, and have fun with it.

Wring out simplicity

Our world is complex, and our expertise is the only way to simplify it.

We Bought the Licenses

Permission Isn't Participation

If you don't go to each individual and focus on real people and their real problems, you might get usage, but you won't get change. Granting permission opens the door with technology. Does your culture open that door, too?

Giving 1 person the means to drive change is more valuable than crafting the perfect policy or directive.

Fun, Little, Dumb Things

We've all been reluctant to try something new. Ruth Paarman writes content for businesses. She was one of the first people to worry what AI would do to her livelihood. She resisted using AI for years — it wasn't as good at writing, it didn't know the context, and it destroyed the spirit of the work. who said the way in is to find a fun little dumb thing to work on. It stuck with the room harder than anything on my slides, and it's right for a reason: it moves the incentive from money or promotion to curiosity, which is the only motivation you can supply yourself.

1. One annoying task

   Something that irritates you personally. Not the company's top priority.

2. Ten minutes

   Bound the risk before you bound the scope. Ten minutes cannot fail expensively.

3. One question

   Follow the curiosity, not the roadmap. Let the result tell you what to try next.

A Stack of Ungraded Paper

How Long It Took

2011

Took one coding class. Other people were better at it, and that scared me, so I quit and majored in languages instead.

2022

A childhood friend described a language box you could talk to that talked back. Nobody was sure exactly why it worked. We started building LexiLift.

2023

LexiLift failed.

2024 · spring

The grading automation. The first thing I built that changed my own working life.

2024 · summer

Left teaching for a curriculum role. No AI in the job description, so I did it at night.

2024 · nov

Automation consultant for law firms. Finally it was the job.

2025 → now

Innovation Lead for AI transformation at a firm of about a hundred people. Then Mosaic.

It took 15 years to try coding again. It took 18 months to turn that into a career.

A Bag of Cans

Iowa has a bottle deposit: five cents on every container. For decades, getting that nickel back worked like this.

1. Sort

   Separate every category at home.

2. Transport

   Fill a car with loose containers.

3. Count

   Feed or hand over every single item.

4. Wait

   Get a receipt. Redeem the receipt.

{
  "What are we actually paying for": "the deposit back. not a receipt, not a count.",
  "Who verifies, and why": "both sides, twice, because the unit is one can",
  "What has to be true": "the count has to be trusted. it does not have to be watched.",
  "What could stand in for counting": "anything with a known, standard quantity"
}

{
  "The new unit": "one Can Shed bag",
  "What you do": "fill it, drop it off, pick up an empty one",
  "What you get": "$12, on the spot",
  "What disappeared": "sorting, hand-counting, waiting, the receipt",
  "Why it works": "the container is the count"
}
  // Standardization was the innovation.

Nobody automated counting. They changed what had to be counted.

What Must Be True

Can Shed is what the move looks like when somebody gets there by instinct. Here is how to get there on purpose. Four questions, in this order, and the order is the whole technique.

The most common failure isn't refusing to automate. It's automating precisely what the person does today — align, trace, cut — and shipping a faster version of a process nobody ever designed.

Chesterton's Fence

Every one of those questions costs you something. Somebody at Can Shed had to understand deposit law, container types, transport costs and fraud before any of it could be made that simple. There is no version of this where you get the simple answer without doing that work first.

That is the guardrail. It is slower, it is less satisfying, and it is the only thing standing between a redesign and a very confident mistake.

Sixty-Three Elements

Here is what that work buys, from somebody who did it in 1869.

This story was told to me by Juan Aranda.

Make Your Own

Three squares he left empty, and described anyway

Within seventeen years all three had been found, each close to the weight and the behavior he wrote down for a square that was, at the time, blank.

It's on the wall of every chemistry room on earth and nobody looks at it. Almost nobody asks the only question about it that matters: how would I make one of these for my own work?

You don't find simplicity. You wring it out of the complicated thing.

What This Asks of You

Enablement isn't a program you run. It's a condition you create, and each seat in the room creates a different part of it.

If you are anyone

Pick one fun, little, dumb thing this month. You don't need approval to spend ten minutes on something that annoys you.

If you lead

Protect small experiments. Your job isn't to approve the roadmap — it's to make it safe to have a bad idea in public, and to notice out loud when one of them works.

If you are on a team

Ask what the work must accomplish before anyone asks what tool to use. You're the only people who know which fences are load-bearing.

Own your transformation, then go help somebody own theirs. That is the whole of leading it.

Systems, Not Heroics

Here is the part that gets left out. Experimenting constantly is tiring. Learning something new every week, on top of the job you already have, at night, with nobody asking you to — that is a real cost, and it comes out of the rest of your life. If the transformation runs on you being heroic, it ends when you do.

Are you running on systems or heroics? Only one of those can work without you. Debbie Foster

I have been on every rung of that. I was the hero who could fix anything. Then I was the leader of heroes, teaching people to be smarter about their heroics. Then I worked out that a system which lifts your colleagues gives time back to all of you at once — including you. That last step is the only one that scaled, and it is the reason Mosaic exists.

Three Things

Access is not adoption

Give people a reason, not a directive.

Start little, start dumb

Practice making your own life better, then build from there.

Wring out simplicity

The simple version only exists once somebody has understood the complicated one. Be that somebody.