I showed up during lunch, where I scraped the last helping of chicken into a rice bowl from a Chipotle buffet intended to feed 100 people.

“What did I miss?” I ask, sliding into my seat at a table full of men.

“Um…” o…

Read more

I’ve been working independently since 2022 and running a company solo since 2025. But this year, I realized how wrong I’ve been about what it takes to play this game.

San Francisco used to represent a whole world I felt locked out of, as a non-technical, non venture-backed founder. But since I’m presently on my first flight out West since entering “founder mode” myself, I decided to take the time to share what I’ve learned.

I grew up an academic purist. I believed that good grades led to good outcomes and that all competitions were judged on objectivity and merit. (Obviously this is incredibly naive. But it’s hard to rewrite your natural priming.)

That’s why (before I went solo), I used to think that the most successful people were the ones with the most talent and the best ideas. I used to get mad when brilliant people with amazing ideas got sidelined while others got to play on the field. But deep down, I think I believed that the people in the game must actually be extraordinary. (How else did they get tapped to play?)

As it turns out, this is a dangerous conclusion to draw because it makes you believe there are only two possible win conditions. Either:

1. You’re actually one in a million extraordinary (an infinitesimal shot for most of us), or

2. Your level of self-delusion is so high that you actually convince yourself that you are.

Since I did not associate as either, I essentially wrote myself out of the game by my own set of rules. Rules that I made up in my head.

That’s why, over the years, when the idea of starting a business would come up, I’d tell myself some flavor of this story:

“I’m not
[smart enough]
[rich enough]
[technical enough]
[young enough]
[creative enough]
to start my own company in the tech industry.”

As it turns out, if you tell yourself this enough times over 15 years, you start to believe it’s true.

But when you start from these seemingly reasonable objections, you suddenly jump to limiting conclusions like:

“That’s why I’ll…
[be a COO but never a CEO].”
[need a technical cofounder if I ever start a company].”
[need to go to business school to be taken seriously].”
[join an investment firm rather than operate a company].”
[join someone else’s rocket ship rather than trust myself to start my own].”

Game over, you might tell yourself.

Thank you for playing.

Who Gets to Play

Last year, I finally figured out that those rules aren’t real.

In December 2024, I vibe coded my first app on a Friday night, then used it with my kids at the Metropolitan Museum of Art the very next day. By Monday, something broke inside. The story that I’d been telling myself for my entire career unwound itself in a single weekend.

All this time, I thought I’d be bound to the sidelines for another decade, longingly watching the game play on without me on the field. But in the end, the next move was so much simpler than all that: I stopped waiting to get invited. I just started playing.

I still think most of us largely under-estimate the paradigm-shifting nature of what AI coding agents bring to everyday people. When I stopped thinking of myself as a user of someone else’s software or a player of someone else’s game, but instead as a builder, it completely flipped the old narrative on its head.

It got me wondering: What else have I been wrong about? If it turns out that not being technical is no longer a barrier to entry, then what else might be untrue? Is it also possible that you don’t need to have millions of dollars to get started? That you can learn corporate finance without shelling out $200k for a business degree?

Over the past 18 months, I’ve been pretty shocked to find just how many of these assumptions have been patently untrue. Carrying them around for the bulk of my career made them feel real, but they were just stories. This is not easy to admit, but I’m sharing it because my guess is that you’re holding onto some untrue assumptions about yourself, too.

Maybe there is a little bit to winning that’s about talent. Or money. Or skills. But you don’t actually need any of those things to just get started. If you want to play, you just need a little courage to get yourself on the field, and maybe some strong habits that give you the stamina to keep going.

When you look at it this way, success feels a lot less remarkable. Which, as it turns out, also makes it feel a lot more achievable.

Are You Playing to Win or Playing to Play?

Last year when describing my latest endeavors to a friend, he asked me, “Are you having fun yet?” I looked at him like he was unwell.

At the time, I was about six months into a go-to-market push that was not working for an app that I no longer maintain. I hadn’t paid myself in half a year, and I was starting to feel severe financial strain. I knew I needed to make a change but hadn’t yet internalized that lower gut feeling of how to actually push through a pivot.

Let’s be perfectly clear: It’s really hard to have fun when you can’t afford to pay your own living expenses.

A lot has changed since then. In January, I hit my first ever “bingo week” – a week where every day I felt like my skills were tapped precisely to meet that moment. In a single week, I ran a paid workshop, I ran a student CS course, I helped someone build her first app, and I piloted an app I built with 2-3 lower grade classrooms at a school district in North Carolina.

It took me about a year of building to hit my first “bingo week.” Which meant I knew I could hit it again. And I have. After all, right now, I’m getting paid to take a trip across the country to do work that I love to do (and hopefully help other people unlock that feeling in themselves, too).

That’s gotta count for something.

Like it or not, there’s something en route to San Francisco that makes me think a lot about playing to win. The startup scoreboards flash like billboards on the way from the airport to the city, a constant reminder of who’s raised what, how much your equity pool is worth, or what the prize looks like when you finally cash out.

But if you move too fast out there, you can miss staying in touch with what’s real. Winning is good. Winning is fun, but winning isn’t the only reason worth playing a game.

I ask myself often, “Would I prefer it if I could just skip over the messy middle and jump straight to the end?”

For me, at least, the answer is no. Over the past five years, I’ve consistently turned down full-time job offers that would have unlocked much faster access to financial mobility. I’ve had little to no interest in joining a rocket ship hyperscaling AI behemoth (even though I know I could benefit from a massive payday).

Win, lose, or draw, I’m playing for agency, ownership, and impact-oriented outcomes, not a particular dollar sign. Since I started building things myself, I’m increasingly appreciative of the satisfaction of doing it the hard way, at whatever pace feels right to me. It seems that every week, I cross another milestone personally and professionally. I’m able to pay myself to do this work, and I’ve even been able to tap friends into projects as subcontractors.

Today, my days and weeks look absolutely unhinged to anyone who isn’t me. I’m typically running 3-5 AI workshops each week, taking 3-5 sales calls, and spending large blocks of deep work time, alone in a room with my AI’s, talking through all of my business goals and prototyping new mini-apps to help me solve the next phase of work. For most people, this would be an unsustainable pace of work and life. Somehow, I never seem to get tired enough to quit. To me, that’s about the strongest signal there is that I’m on the right path.

Sure, I could join the venture-backed startup race, raise a $30M round and rush to the finish line. Or I could buy myself an equity-earning lottery ticket at another company and hold out for a big exit. But if I play out either of those options, the short-term view looks a lot less great than the thing I’ve got going for me right now, which is a steadily growing business that’s increasing in maturity at the same pace that I can keep up with my ability to build it technically and support it financially.

Shouldn’t we chalk that one up as a win?

I didn’t start a company to turn it around as a quick acquisition target 18 months out. I started a company because I wanted a place to work where I could do my best work, and where I could see myself doing my best work there for a very long time to come.

So I see it as a good thing that the deals I relish the most are the ones I win on my own, that the conversations that lead somewhere else feel more like a puzzle than a chore, and the excitement of seeing something today that literally didn’t exist yesterday is a dopamine hit that you can’t get if you fast-forward all the way to the end.

It’s taken me a long time to find the shape of a container that seems to suit my personality and temperament. So why would I be in a rush to change things?

As it turns out, you don’t have to pray for a win in order to enjoy the experience of being on the field during the game.

Thank you for playing.

Not because the game is over. But because you actually took the time to play.

The Line Between Learning and Working Just Collapsed

For a long time, we operated under a clear binary: Schools are where you learn, and businesses are where you work.

While the scales have been tipping slowly for years (with companies launching internal L&D programs and schools adopting project-based learning), the recent acceleration around AI has practically flipped the scales overnight.

It feels impossible to open an inbox without seeing another venture or profitable business launching an educational ecosystem. Founders School is promising million-dollar payouts to teenagers who launch businesses. Andreessen Horowitz recently announced the Horowitz Andreessen Academy to train high-potential talent. Meanwhile, AI communities are finding that their true monetization engine isn’t the tooling itself, it’s the teaching. (see: The Neuron Academy, and builder-educators like Claire Vo and Sabrina Ramonov).

This is where a significant portion of my own work with Build First has shifted.

Increasingly, my focus is on helping deep domain experts or functional teams get real work done using AI. This summer, we piloted two different versions of “Build First AI School” for businesses, with 4-8 week cohorts of weekly classes to help non-technical people become builders. Over the next month, I’m running an AI workshop for HR leaders in San Francisco, a remote bootcamp for finance professionals, and a legal hackathon in NYC.

The deeper I get into these specific verticals, the less generic the AI instructions become. When last year I was teaching people how to “click the plus-sign to attach a document to your chat conversation,” this year I’m increasingly finding myself inserted directly into the front lines. And I’m learning more too. After all, to teach someone how to use technology to solve HR or finance problems, you have to understand enough about HR and finance to map out their problem space well enough to build a real solution.

Last week alone, I ran two back-to-back workshops for entirely different cohorts: Geologists on one day, battery experts the next. The context shifting is wildly unsettling, but the practice of building remains remarkably consistent. In other words, the educator and the practitioner are merging.

The Pause vs. The Gas Pedal

The other interesting thing about this particular moment is that the power right now lives among individuals and groups with the greatest capacity for experimentation. This gap is dizzying.

On the one hand, my Twitter and LinkedIn feeds increasingly make it seem like if you aren’t starting your own business and running your own agentic swarm, you’re already behind. Gary Vaynerchuk posts content 4-6x a day encouraging people to quit their day jobs and start something new, Greg Eisenberg’s Idea Browser gives away free business ideas to anyone willing to pay attention, and newsletters like Nate’s Newsletter, Lenny’s Newsletter, and Ben’s Bites each topple hundreds of thousands of subscribers with daily builds.

On the other, traditional institutions inherently recognize that they have a lot more to lose when a few erratic agents go off the rails and are opting to move slower as they grapple with a massive question: What does it mean to incorporate AI in a way that keeps our proprietary data safe?

Unfortunately, this is creating a scary bifurcation in AI upskilling: If you work at a traditional institution, you are less likely to be exposed to the cutting-edge frontier tech that the next generation will be built upon. If you have a manager today who’s opting out of learning this stuff, it’s going to come at the expense of your own ability to learn how to adapt for the what’s next.

Last week, on the train ride to a corporate training session, my co-facilitator and I got an alarming notification: The exact software product we were hired to teach was being sunset. The announcement had gone live that morning. We had to stand up and introduce a new tool in the same breath that we announced what they would need to migrate it to next.

This innovation explosion is not only causing complete chaos among leadership teams, but it’s leading individual contributors and senior leaders alike to ask themselves the hard question: “Do I stay in a work environment that won’t move as fast as I can?”

The answer, increasingly, is no. In the past six months, two past corporate clients left their jobs for more forward-thinking teams elsewhere, and I know two more with one foot out the door. Some others have decided to bypass the bureaucracy completely and pay out of pocket for me to teach them things their companies won’t. In other words: The demand to learn this stuff couldn’t be higher.

And yet.

Public school systems (including New York City, where I live) are increasingly opting to hit pause on bringing AI into classrooms. In a district serving 850,000 students, you can understand why a safety-first, risk-averse approach feels like the only viable option for leadership. But at the individual level, I can’t honestly look any young person in the eye and tell them they’d be better off not learning this stuff.

There’s something unsettling about all of this.

When public institutions pause while private capital accelerates, we create a dangerous vacuum. We’ve seen in web 2.0 what happens when we let our education be shaped only by the people who have the funding, resources, and access get to sit in the room and control the narrative. If schools push pause, but businesses start schools anyway, what’s the end game?

Building from the Front Lines

Last week, I ran an AI build session for master’s students in Technology Management at Columbia University. After the class, I had a conversation with a woman who had spent her entire career as a software product manager. She had been playing around with vibe coding and dabbling enough to know exactly what she wanted to build. But like so many non-engineers, she had been told for years that she needed a technical counterpart to build anything real.

I recognized this feeling right away since it’s the same thing I felt last year, too. And I told her just as much: She was waiting for a permission slip that might never come.

Notably, someone like her is not the target audience for a program promising teenagers a $1M payout at graduation. But she represents the exact persona we need more of right now: People who are deeply close to real problems, who have lived enough to know what their communities actually need.

When AI allows domain experts to turn ideas into software in real time, everything changes. And I’ve seen this happen in real time. It takes hours, not months, to flip the switch.

Earlier this year, I helped a seasoned educator and non-profit leader in North Carolina build an MVP for a math tutoring app she’d been conceptualizing for years. She had the design schemas on paper, she had wireframes and a project plan ready to go, she just lacked the technical execution. In a single two-hour session using Replit, we spun up a functional prototype she could share internally the very next day.

It completely reframed how she viewed building. All at once, the question shifted from “Could someone build this software for me?” to “Who else needs this, and how fast can I get it to market?”

To be clear, I don’t believe every school will become a startup incubator, just like I don’t believe every business should become an academy. There is plenty of gray space in the middle. But it’s time we acknowledge that the line between learning and working is no longer binary and that the people closest to the problems are finally the ones holding the tools to solve them.

For the past 3.5 years, I’ve worked most days from a shoebox office on 28th and Park Avenue that’s the size of four sidewalk squares. In nine days, I’m moving out.

I wasn’t expecting to miss it, but it will be a little sad to say goodbye. This tiny sanctum where you can touch all four walls from the center has seen me through a pretty monumental career shift. And I can honestly say I couldn’t have done it without having a space to myself to figure things out.

Back in 2023 when I first moved in, I had a baby and a toddler at home while working fractionally for a handful of crypto projects. Today I’m full time running my own business teaching people how to become AI builders.

Like a lot of our friends, we’d spent 2019 converting our home office into a nursery. Which meant we spent the next two years during the pandemic working two feet from the bed we slept in at night. We stayed on mute through Zoom calls while the baby screamed on the other side of the wall and made sure the camera angle just barely kept the piles of dirty laundry out of view.

But you can’t make something new from a place where you can never focus.

While my current space is too small to receive visitors, I didn’t mind. Its primary purpose has always been to do what I could never do at home: Unlock my creative potential energy.

Over the years, what started off as a sterile glass fishbowl has accumulated so many trinkets that people walking past sometimes stop to take a photo. Right now, it looks more like my childhood bedroom than any place I’ve ever worked professionally. It’s decked from top to bottom with colorful posters, photos, and art from my kids. I’ve got tiny toys scattered everywhere: Legos figurines, a 3D-printed slinky, an unboxed Tamagotchi, coloring books I made myself, and three brightly colored stuffed animals.

When the kids spilled milk all over the yellow and orange shag carpet in their room, I cleaned it up and brought it in. When a neighbor left a tiny yellow table in the front hall, I repurposed it as a candle and coffee stand. There’s no whiteboard, but since all the walls are glass, I brought in whiteboard markers and began scrawling away. Today, every wall is covered with ideas, inspiration, motivational tips, and Post-It note reminders to keep me on track: “Just write it.” “Build ahead, sell behind.” “Community as the input.” “No more ideas!”

Despite its billing as a community coworking space, my tiny corner at the end of a long hall is just two spots from the boiler room and nearby rooms have remained largely vacant. Months would pass without seeing another person in one of the unrented rooms across the hall (the more expensive ones, with the windows) that they sell for twice the price. I haven’t made many friends here, and the ones I did have long since moved out. A few months ago, another startup managed to squeeze 3 desks and 5 monitors into their fishbowl across the hall, which means I get to watch their dashboard metrics and Claude Code sessions all day long. It’s crowded, but at least when the third person joined, they stopped doing push-ups in the hall.

There have been days where I’ve been so tired that I’ve fallen asleep at my desk and many nights when I’m the only one left in the dark halls, setting off the motion-controlled lights only as I come and go from the bathroom. I’ve shown up at 5am for calls with Europe when I briefly GM’ed an international dev shop, and I’ve stayed until after midnight for project planning calls with a client in Australia.

It worked really well for a really long time.

But then last year, I started needing a place where I could bring someone else around. I spent the summer building an app with voice AI with a friend. It needed a lot of testing, which meant at coffee shops, we were a nightmare pair. Just picture it: Two people talking and giggling over an app that talked back, while also voice dictating bug fixes to our computers, all while trying to tune out the barista’s espresso machine and avoid making eye contact with other patrons.

I also started getting inbound interest for 1:1 AI coaching sessions. Since I’d learned that my living room doesn’t double up well as a creative studio, we’ve been using common space in my current coworking suite. While my clients have been nice about it, I know they also don’t love talking about proprietary business workflows within earshot of a receptionist receiving packages and building deliveries.

And then there’s the group work. With increasing interest among friends and fellow builders in IRL AI studio sessions, this year, I’ve hosted more than a dozen workshops in borrowed spaces across the city. I’ve found that getting that first “yes” is easy, but finding a consistent home for a community is harder. I’m now halfway through teaching a four-week cohort for women starting their own consultancies or agencies, and a single three-hour room rental costs as much as one participant pays for the entire course.

I’ve stood in the new room only once. It’s bigger, with exposed brick, a second desk, and space for me to run two workshops a month from their community coworking conference rooms. It’s a good start. I still want a room I can build alone in and teach a dozen people in, on the same Tuesday. I’ve been looking all year. But I’m not sure anyone makes that yet.

This weekend I brought my kids to my current office for the first and last time. Lydia (6) wanted to see “mom’s office which is the same size as a bathroom.”

While my daughters played with all my toys and trinkets, I looked around and sighed.

“How am I going to move all of this to a new space?”

Sydney (4) glanced up and then back at me.

“Well, Mom, start by just taking everything off the walls. And then go back and forth and back and forth until you’re all done.”

“Have you seen the news?” I ask my husband this morning, promptly upon waking up.

“About the AI ban in NYC public schools through 8th grade?”

“Yeah.”

“I mean, isn’t that…better than we thought?” he says. “Weren’t we mostly worried about high schoolers?”

I give him a knowing look.

How old were you when you started coding?

Flash back to the summer of 1999.

I’m 13 years old and obsessed with a video game, “Where in the World is Carmen Sandiego.” I played it every afternoon. I just loved the scheme of following a character around the world, helping her find clues to sleuth out criminals.

I wanted to win every round.

At some point, I realized that while the clues changed with each level, the specific evidence you needed to obtain a warrant to arrest any criminal stayed the same each time.

That’s when I came up with an idea. If I could log all the details about all the criminals, I could win every round. So I reached for the tool I knew best: My notebook.

I made a page for every criminal mastermind in the game. Every time I played and learned something new about a criminal, I wrote it down in my notebook. By the end of the summer, I had a perfectly complete list of every detail about every character. I won every round.

My very first database. Only I didn’t call it that, of course.

I called it writing.

Now cut back to 1999.

There’s another middle-school kid who also liked to play games. His favorite was the board game, Clue. He’d play with his family on nights and weekends.

He wanted to win every round.

At some point, he realized that while the clues varied with each game play, the predictions narrowed with each player’s move. But it was tough to keep track of everything in your head.

That’s when he came up with an idea. If he could log all of the details about all of the clues, he could beat his family to solving the mystery every time. So he reached for the tool he knew best: His computer.

He coded the game logic into JavaScript, and logged each round of gameplay. It tracked each decision each player made, and when enough evidence was gathered, the algorithm told him the final answer using the minimum required information.

He called it coding.

In the end, my husband became an engineer, and I graduated with a journalism degree into an industry that technology had already transformed.

I then spent 15 years working alongside brilliant engineers while continuing to think of myself as “non-technical.” Proximity to technology wasn’t enough. I just never saw it as something I could shape.

As it turned out, that belief began forming long before high school.

Yes, the risks of relational AI for children are real and unresolved. But a blanket ban doesn’t eliminate those risks. It determines which kids learn to see AI as a tool they can use to solve a problem, and which ones grow up believing it’s beyond their reach.

Which Tool Do You Reach For?

Yesterday, I saw what can happen when students are encouraged to reach for a different tool.

In a school-wide design competition called AI for the People, 400 students at Comp Sci High (including those as young as age 14) conceived real app concepts to solve real problems for their neighborhood.

One group picked up on a subtle, but important problem: Could AI help non-English-speaking tenants turn a photo or description of a problem into a clear repair request for their landlord?

I thought that was a very good idea.

I’ve been on this school’s board since it opened eight years ago. Eight years of watching 9th graders arrive already sorted into technical and non-technical. But we’ve reached the same conclusion that high school is too late to start. That’s why Comp Sci High is planning to open a middle school next year.

So if high school is too late, how early is early enough?

I can’t tell you exactly how early is early enough. But I can share my lived experience. I spent last year building AI education apps with my two young daughters (now ages 4 and 6) as co-shapers, not just users.

My first app, MuseKat, helped my kids make sense of the art on the walls at museums on their level, through a storytelling lens of a friendly meerkat character named Miko.

For six months straight, my daughters trotted along to every museum in the city with me, testing Miko. We’d listen to what he told us, then they would share their feedback with me, and then I’d make a few changes.

What I loved about this experience is not only that my daughters had the chance to participate in supervised use of AI, but that they saw their own feedback reflected in the resulting product. It introduced technological agency at a very young age.

When I introduced Scribblins, a second AI-powered, play-based app later on in the year, my daughter and I showed it off together, in her public school kindergarten classroom on the Upper West Side.

“My mom made this,” she proudly told her friends. “And I helped, too.”

The Cost of Starting Too Late

Since last year, I’ve helped 3,000 people build their first apps with AI no-code tools through my company, Build First.

I make apps every day of the week. In the time it took me to write this blog post, I also refactored a website, applied a new design schema, and built a data layer to my training work.

Last weekend, my husband found me working alone in our darkened bedroom, headphones on, illuminated by my computer screen, where I’d been hacking away on a problem alone for hours.

“Whoa,” he said. “You remind me of myself in middle school.”

I looked back at him and grimaced.

I’m 39 years old. It’s taken me more than 20 years to develop that same sense of technological agency that he’s felt for most of his adult life.

Today, I sometimes feel like I’m making up for lost time. I wonder what I would have been capable of with a little more guided instruction from a younger age.

These instincts form earlier than we realize. By the time we sort children into “technical” and “nontechnical,” we may simply be naming the consequences of which tools they were encouraged to explore.

If you want to see what this costs at scale, just look at corporate America.

When I run AI transformations at companies, the thing standing in the way is almost never the technology. It’s capable adults who have spent entire careers believing that building software was someone else’s job.

Watching that belief come apart takes only a few hours. But developing the confidence to act on it can take months. My husband absorbed that lesson naturally over a single summer.

I don’t wish someone had simply taught me to code. But I wish someone had taught me that software was something I could shape. And I want every kid in this city to learn that, too.

A Surprise AI Upgrade

The next wave of conversational AI is happening right under our noses, whether we like it or not. It’s time to pay attention and get involved.

My family returned home this weekend from a one-week trip away to a surprise upgrade from Amazon Alexa.

“I’m Alexa+,” said the voice, now in a distinctly different tone and personality. “Our next generation Alexa experience is now included with your Prime subscription.”

My husband and I frowned.

“Alexa, what can you do now?” I asked.

“I’m Alexa+,” the voice repeated. “Our next generation Alexa experience is now included with your Prime subscription.”

“Alexa, how do you manage privacy controls?” asked Jason.

“I only listen and respond when the identification word is used to call me,” she assured. “You can change these settings at any time in the Alexa app.”

We rolled our eyes and groaned, then began to unpack from our trip.

A few minutes later, I overheard my kids talking to new Alexa in our bedroom.

“With my upgraded AI, I can help you with more than just playing music and sharing the weather,” Alexa said. “I can book you a restaurant reservation, call you an Uber, or more.”

“Alexa, I love you!” cried Lydia (6).

I tensed up while waiting for the response.

“I’m just an AI, not a romantic companion,” Alexa informed her. “Is there something else I can help you with?”

“Alexa, tell me a story!” she requested.

I ducked around the corner while Alexa told a story about a girl who found an ancient light that gave off special powers.

I perked up at the word, “ancient,” clocking it as one of the most popular words used in stories that my own AI-powered kids’ app, Scribblins, used to use when encouraging make-believe stories with friends. A sign of the times.

“Alexa, you sound different!” Sydney (3) chimed in.

“Would you like me to synchronize your voice so I can remember things about you and personalize it for next time?” the voice asked.

Whoa. That’s new.

“Yes!” my two kiddos giggled.

“Great, let’s get started. What’s your name?”

“I’m Lydia,” said my 6-year-old.

“Hi Lydia, would you like me to activate customized voice control for you?”

“Yes!” Lydia responded, gleefully.

“It sounds like you might need your parent to approve this request. You can manage voice synchronization in the Alexa app.”

I breathed a sigh of relief.

“I’m Sydney!” shouted Sydney, copying her sister. “And Mom is here too! Her name is Bethany!”

“Hi Sydney, would you like me to activate customized voice control for you?”

“Yes!” she shouted, then erupted into another round of giggles.

“You got it. It just takes a few snippets of conversation for me to save your voice to my memory. I will replay the last parts of our conversation and then catalogue your voice. This will take a couple of minutes.”

Whoa, whoa whoa.

My mind immediately went to voice cloning.

Today, convincing AI voices can be created from less than 10 seconds of recorded audio. Earlier this year, I heard firsthand accounts from parents who received fake kidnapping calls that appeared to come from their own children’s voices.

I can’t imagine that’s Amazon’s goal. But I don’t really know where my kid’s voice is stored if I decide to opt into this service. Hearing a device ask to catalog my six-year-old’s voice made me realize how quickly the line between “helpful personalization” and “sensitive biometric data” has blurred.

Is this how it begins?

“Alexa, stop voice control,” I overrode immediately. “Delete all memories of personalized voice conversations. Do not save anything.”

The AI Backlash in Education

It’s hard to know how to handle AI’s meteoric effect as it relates to educating our own children.

On the one hand, some argue, AI is here, and we must figure out how to integrate it into our lives. On the other, two decades of distrust in corporate institutions and false promises with technology in classrooms has led not only to social media backlash, but technology backlash in general.

Earlier this year, New York City Public Schools faced immense blowback for introducing a specialized AI high school, ultimately backing down from the initiative entirely. Meanwhile, graduation speakers are getting booed off stage, VC techno-optimism is not sitting well with the social impact sector, and some are suggesting that NYC schools pause AI use for two years.

Despite these challenges, I spent the bulk of last year teaching myself how to build with AI largely by building three different AI-powered kid-friendly applications. I did this for two reasons:

1. I wanted to learn how to build apps, and

2. I wanted to develop a thesis for what raising kids in the AI age would mean for me and my family

When building each app, I ran in-person pilot activations, classroom learning modules, and QA tests with young kids in facilitated environments.

These were tricky to coordinate. Many teachers were still quite AI skeptical last summer and fall. Museums were nervous to replace docents or deliver potentially inaccurate information. When I asked parents if they would pay for their kids to learn AI literacy in a safe way, only 2 in 10 said they would feel comfortable doing so.

“Have any of you used AI before?” I’d ask each class of first or second-graders.

One or two hands would go up in the air.

But then I’d ask this telling follow-up.

“How many of you have an Alexa or a Google Home in your home?”

In classrooms from Harlem to Charlotte, every single little hand went up.

Forfeiting the Future

I empathize with the challenges of introducing such a novel, dynamic, and hyper-changing thing into our educational system with proper guardrails.

But we have to face reality: AI is already inside of our homes. It is embedded into every Google search result. It’s getting stuffed inside of teddy bears and sunglasses and rings.

If you’re a parent today and you feel strongly about what your kid’s future looks like in the AI age and you choose not to build alongside them to figure it out, you’re not just forfeiting your own knowledge. You leave a gap for twenty-something tech graduates who aren't parents to decide what your child's AI storybook should look and feel like.

The only question is: Will you wait it out to see what happens naturally, or will you join in the conversation?

Because the third option is to leave it up to your Alexa+ and your 6-year-old to figure it out on their own.

Wyman was Head of Learning Engineering at Playlab, a company at the forefront of helping schools navigate both the opportunities and challenges of AI adoption.

I first met him in 2024, when he and Playlab’s co-founder Yusuf Ahmad changed my mind about how to introduce AI adoption to groups at scale via communities of practice and facilitated learning environments.

At the time, I was working with Decoded Futures, a Tech:NYC project that helps nonprofits level up on AI through a cohort-based learning program. As a fellow portfolio project of the Robin Hood Learning + Technology Fund, Decoded Futures teamed up with Playlab to run a 30-person demo day for educators, hosted at Sesame Workshop.

It was the first time I saw how they could take a group of people who had never coded before and teach them how to build (and demo!) an app in a single afternoon. That afternoon changed the way I thought about teaching AI, and its influence is still visible in my own work today.

Less Tech, More Teaching

What Wyman understood before many others was that AI adoption wasn't primarily a technology problem. It was a teaching problem. That’s why he treated every room with as much care and compassion as his own classroom.

A few months after the Sesame Workshop event, I helped a local NYC high school consider opportunities for AI inclusion for an extra-curricular entrepreneurship initiative. While I’d run a lot of AI trainings by that point, my work with teachers in particular was limited, and I emailed Wyman for help.

Without missing a beat, he acknowledged that since the school was so nearby, he’d love to join for the session. Wyman came, got every teacher in the room onboarded with a free account, and patiently answered their questions, concerns, and fears about AI as he shared his own story in moving from a teacher to developing his own set of best practices in rolling out AI in a safe, practiced way.

“I’m not here to pitch you,” he told the room of world-wary educators who had been sold the story of technological transformation in the classroom many times before. “But AI is an ‘arrival’ technology, which means it’s here, even if you aren’t seeking it out. I’m here to collaborate and share what we can do together.”

Wyman knew that skeptical rooms aren't won over with scare tactics and hype. They're won through trust, patience, and the feeling that everyone is learning together. That's why, whenever I connected him with a teacher or school, he made time for the conversation.

Giving Back to Grow Forward

Wyman was the kind of person who has always worked even little rooms in big ways. And the outpouring of grief from former students, teachers, colleagues, and friends is a real testament to that.

When I first saw the email from Playlab with the subject line, “Honoring Wyman Khuu,” I opened it immediately, assuming it was an announcement of a laudable education award or prize he had won. That's how much I'd come to associate his work with excellence.

As AI continues to reshape our educational systems and entire way of life, we need more people like him. In a time where leadership is fracturing in corporate and political spheres, where the future is more questionable than ever, and there are a lot of things to be afraid about in our future, it’s rare to find the kind of high-integrity person who will sit, shoulder-to-shoulder with you, and help you learn something new.

Whether or not you knew Wyman, if you’d like to carry on some of his best qualities in your own work (whatever that may be), here are three things I saw him do again and again.

1. He showed up when he was invited to give back.

2. He saved space for people who carried different perspectives to the conversation.

3. He never made anyone feel stupid for not knowing something.

We lost one of the good guys this week. Here’s to hoping his example inspires a few more.

Sometimes you don’t know the true edges of a system until you come up against outlier scenarios. When I accidentally discovered a new way for airlines to help customers who get stranded, it reminded me to why it’s important to spend time pushing the boundaries of any system or tool.

Yesterday I was booked on a flight to fly from Dayton, Ohio, to New York City by way of Atlanta, Georgia. But my flight was late. Too late to make my connecting flight.

“Can you book me on a later flight to New York?” I asked at the Delta check-in desk in Dayton.

“Looks like all the other flights to Laguardia are sold out,” the attendant replied.

As a frequent flier, I’m accustomed to the options that airlines can offer when flights are canceled (ie: new flight time, different connecting flight, free night in a hotel), so i started to rattle off other options.

“How about JFK? I’d even do Newark. I just have to get home tonight since the kids are with a sitter.”

“Well…” the attendant began. “The only way would be for you to take an Uber to Cincinnati and fly out from there. It’s a 90-min trip, still time for you to catch the 5:30 p.m. flight.”

I didn’t really want to pay for an impromptu car service, but I had run out of options. My husband was out of town, my kids were stuck with a sitter whose shift ended at 11 p.m. If I didn’t get home tonight, how would the kids get to school?

"Yes,” I told him “Let’s do it.”

“Great,” the agent replied. “Rebooking you now.”

I grimaced as I opened up the Uber app on my phone as I loaded up how much this impromptu 90-min ride would cost.

“And will you be needing ground transportation?”

“I’m sorry, what?”

“Ground transportation,” he repeated. “Should we book you an Uber?”

My eyes widened at the thoughts of this novel arrangement: Ground transportation, paid for by the airline service.

I shut my own Uber app in a minor triumph.

“Why yes,” I said. “Yes I would.”

They had me on a pre-paid car out of the airport in seven minutes flat. As we pulled away, I realized I hadn’t just learned a new airline policy. I’d discovered another edge of the system.

Why Boundary Pushing is Healthy

We tend to assume systems have fewer options than they actually do. Once we’ve learned the “normal” paths, we stop asking what else is possible.

But organizations often have strange, little-known accommodations reserved for unusual circumstances. These solutions that only appear when someone pushes beyond the default menu.

Yesterday, that edge happened to be a prepaid Uber to another airport. Next time, in some entirely different system, I’ll probably ask one more question before assuming I’ve run out of options.

In human-made systems, I’ve got a habit of seeing how far I can push boundaries. Sometimes I hit the edge and I’m snapped back into place. But other times, the bricks in that seemingly sturdy wall are more moveable than they might appear. Finding those loose bricks is a game to me. I like playing it.

But there’s more to this than meets the eye.

Boundary pushing is a very important step to take when learning anything new.

Kids learn how the world works by pushing the boundaries of what’s considered “socially acceptable” behavior (and subsequently getting called back). Grown-ups learn the rules of systems like airports not by memorizing manuals of every single rule, but by recognizing the patterns of what things tend to happen when flights get canceled or delayed.

Oh, and by the way, if you’re learning a new tool or technology, you better believe boundary-pushing is the best way for you do that, too.

“It Can’t Do That…”

Over the past year, I’ve helped 2,500+ people build their first AI-powered applications. Suffice it to say, I’ve seen a lot of people learn something new for the very first time.

I see how some people approach learning with an open mind, whereas others treat it like taking their vitamins.

One thing I notice is that right off the bat? Newbies are very quick to tell me exactly what the AI doesn’t let them do.

• “I can’t integrate with our customized software or tool.”

• “It can’t scrape my LinkedIn connections.”

• “It doesn’t get the look and feel of this document just right.”

It’s fine to hit boundaries. It’s important to hit boundaries. The danger is when you misinterpret a boundary as a forgone conclusion of what’s possible:

“…and that means I can’t make AI work for me.”

Every system has walls. Every tool has breaking points. Your first job as a user of the system is to find the edges and then teach yourself how to work within the system.

I promise you: There is another way to do the thing you want to do.

Coming from someone who specializes in teaching people how to learn new things, as a student, I’m a massive pain in the ass.

I am incredibly particular about how I learn, about who is doing the teaching, and about where the learning is taking place. I don’t like it when people try to teach me without answering all of my questions, I don’t like it when you tell me the thing (vs. show me the thing), and I certainly don’t like feeling stupid or stuck.

I’ve found that this habit of boundary-pushing has served me incredibly well when working with AI tools. And I encourage you to push the boundaries a bit more in your own work too.

The next time you meet a new tool or AI model, treat it like you’re meeting a stranger for the first time. First, try asking what it thinks it’s good at, and then throw a few tests its way. See how far you can push it before it outperforms you, before it confuses you, or before you hit a wall.

When you really think about it, getting stuck in a coding project isn’t really that different from getting stuck in an airport.

You may not take a direct flight home, but you can take a connection. You might not arrive the day you wanted, but maybe you’ll get a free hotel out of it. And just when you think you’re out of options, you might discover there’s a free Uber ready to take you to another airport, 90 minutes away.

America has always sold us a frontier. This weekend, three very different stories reminded me that AI may be the newest one: America’s 250th birthday, Taylor Swift’s wedding, and Anthropic’s temporary release of Fable 5.

They all revolved around the same thing: Mythology. One celebrated a nation’s founding story. One became more powerful by refusing to feed the internet. One promised limitless possibility to anyone willing to sit down and build.

If you spent even a single second in New York City this weekend, you couldn’t have missed the remarkable and rare feeling seeping out of the corners of the city: Unity. Parents dragging star-spangled toddlers by the hands to the riverfronts, folding chairs and picnic blankets in hand.

You can AI-generate an image of an antique plane flying over the skyline, but you can’t replace the feeling of hearing and seeing it with your own eyes. It harkened back to a folksier, simpler era. One where patriotism didn’t feel like such a charged ideology.

I went home with the kids and felt compelled to watch The Sandlot, with my kids, for the very first time.

“I liked it when the doggie ate the ball.”

I watched it for the folklore, I suppose. For a couple of hours, New York felt strangely close to the watermelon-drippy, backyard summers I grew up with. But when it finally showed itself this weekend, I wondered: Is this just a giant tease?

Thanks to Anthropic’s temporary re-release its long-awaited model, Fable, I found myself wondering the same thing with every seemingly superhuman task I assigned my new AI senior software engineering militia for the weekend. Am I talented, or am I participating in arbitrage?

Anything is possible, the tech titans tempt us. The only limit now is your own creativity. What will you build next?

And so, as the modern-day pioneer, I built. I built all weekend long.

I built while my mind wandered to what was happening inside Madison Square Garden, just 40 blocks south, at a wedding that surely rivaled the Great Gatsby era for the modern-day elite.

I updated security protocols and permissions across my apps, waiting for the photos to drop. And when they didn’t, I realized something remarkable: That despite all the press, the crowds, the gawkers, and the celebrities, not a single damn photo got leaked.

After all, the absence of any real content makes anything digital a fake by default.

Taylor Swift, the most popular human on the planet, reminded us all that the rarest form of power may simply be refusing to publish. Scarcity breeds mythology. We play stupid games, we win stupid prizes.

America has always romanticized the frontier. We’ve always had to travel far to find it, whether by boat or by foot or by carriage. The founders back then had to endure great periods of separation from the people they loved, all in the name of actualizing that great American idea. You can always go West. Just one more mountain, one more quest.

“Mommy is working again,” I overheard the kids say to their dad after they ask me if I want to play another card game with them before bed.

The frontier-seekers of today are disappearing, too. We drift further and further into our computer terminals and digital devices, lost in our heads to the imaginary folklore of the promise of a better technological frontier.

“Just one more prompt,” I catch myself thinking. “I’m almost there.”

I’ve been getting a lot of questions from clients lately about which AI tools to use to drive AI adoption at their company. But this question isn’t as simple as it may seem.

The most important AI investment isn’t choosing the right tool. It’s building an organization that can keep learning as the tools inevitably change.

In today’s climate, it’s increasingly difficult to know who to trust.

On one hand, we have the technology providers themselves. Armed with millions of dollars in venture funding and their entire future business on the line, they are aggressively pushing the adoption of their specific services.

From startups like Replit and Lovable (who are rapidly growing their creative community growth teams) to giants like OpenAI and Anthropic, every company is investing heavily in education, community, and enterprise adoption, with Anthropic’s Claude Corps as the latest example of what a super-funded institutional adoption campaign looks like.

It’s the exact same playbook that Google Classroom used to capture distribution in nearly every school in America, wrapped in a narrative that echoes the structure of Teach for America (which, as we’ve learned, works well in some contexts, but really not well in others).

The problem? Any training designed by a single vendor inherently ignores the broader, rapidly shifting technological landscape. This blind spot is exactly why the market for independent, AI-first consultants is booming.

On the other hand, there’s a surging wave of boutique agencies, independent experts, and agentic middleware startups are flooding the market, positioning their AI know-how as the ultimate competitive moat. As a result, they are commanding massive premiums, with quotes ranging from tens of thousands for basic builds to millions for annual retainers.

Which leaves behind business leaders and their teams, who are completely overwhelmed. They are caught in a dizzying cycle of decision fatigue, trying to figure out which tool won’t be obsolete by next quarter without being taken for a ride from a consultant, all while their employees are still stuck using AI as a glorified Google search, rather than a tool to actually build things.

After 20 years of having enterprise software thrown at me, I believe agency is better than authority.

Every company wants to become your AI standard. Every consultant wants to become your AI expert. But neither is a substitute for building an organization that can learn.

Like it or not, every company in the world is becoming a school for learning AI.

Three Things to Know About AI Tools Before You Sign Away Your Company

What these companies are doing is pure provocation. They are injecting the market with a massive dose of FOMO to make us believe that a more perfect organization, a higher-performing team, or deeper market penetration is just one software purchase away.

But before you invest in your internal training (or tooling) options, you need to understand three critical realities.

1. Tools change (and humans are fickle).

Last year, every single company Build First worked with asked us teach AI using ChatGPT or Gemini. This year, it’s nearly all Claude. Next year? Maybe open source and locally hosted models will be in vogue. The fact is: Tools change. Don’t commit to anything forever.

This isn’t just happening with the large LLMs. Earlier stage VC-backed founder friends of mine tell me that their user retention has never been worse; people jump tools faster than ever. Given the novelty factor and the shifting social landscape of which tools are in favor by the broader population, it’s inherently risky to anchor yourself too deeply to anyone provider. Humans are too fickle, and tech move too fast. In that sense, the only way to build long-term technological agency is through resiliency of adaptability and tools. Learning how to learn a new tool is the meta exercise we all need to adopt.

2. One tool does not need to fit them all.

There is a very popular idea when cloud services first came out for software developers called microservices. That’s to say, you didn’t have to store all of your data in one single hosting provider. Companies started using Amazon Web Services for some parts of their codebase and Google Cloud for others.

I’m noticing the same thing is starting to happen with AI tools. Some of the best companies that I’m seeing are ones that are actually looking end to end at the problems they’re trying to solve where the data lives today and picking the tool based on the problem, not the other way around.

Instead of forcing your team into a single, restrictive ecosystem, consider a different goal: Interoperability. Build a flexible stack where different AI models talk to each other to solve specific business needs. This approach also invites you to inherit the best practice of developing "sandboxed" environments with distinct security access and permissions. More importantly, it builds long-term technological resiliency for your organization. If a specific model falls out of favor, or a software contract renewal suddenly gets too expensive, you maintain your optionality and keep switching costs low.

3. Tooling requires a buy vs. build philosophy.

Should we buy a new tool…or should we build it ourselves?

Engineers have been asking this question for ages, but only until recently have my non-engineering friends begun to ask of themselves. Like a lot of things, the conclusion is often, “It depends.”

At its core, this is a philosophical conversation that needs to happen at the leadership level and be largely informed by the makeup and capabilities of your current team. To make a long term investment in any enterprise grade tool means you must make some assumptions about what you think your employees are going to be doing.

Most of your employees have never been asked to build or architect software in their lives. To them, this isn’t just a new tool; it’s a completely different role.

I think we are under-estimating the massive paradigm shift that follows suit from this AI transformation. In this sense, it doesn’t matter where you start. It matters that you start teaching everyone how to stop being users of software and start thinking like makers of software.

The good news is that you can do this with any tool on the market today. You can build Copilot agents or a Gemini Gems or Custom GPTs or Claude projects or artifacts. The workflow is the same. From there, you’ll get a much better sense at who in your organization has the technological chops to think like an engineer and start asking them what problems they need to solve. This does not have to be a decision made from the top. It should be made informed by grassroots behavior of real things at your company.

The goal isn't to pick the perfect AI tool. It's to build an organization that can keep learning as the tools evolve. If you're working through that challenge, Build First would be happy to help. You can reach out directly at bethany@buildfirst.ai.

It’s always incredible to see how far people can go with just 6 or 8 weeks of focused building. But one participant shared a piece of reality that really stuck with me. While they ultimately got the exact output they wanted, it took 40 drafts across different versions and Claude prompts to get there.

This led to a conversation about “tokenmaxxing” (the art of squeezing the absolute most value, context, and efficiency out of every single AI prompt) and how to optimize your build time. Engineers talk a lot about ways to optimize model usage, context windows, compute cost, and prompt efficiency. But one thing that particularly helped me in the beginning was much simpler: Better first drafts.

5 Ways to Get Better First Drafts

In the 18 months since I’ve been building with AI, I’ve gotten much better at first drafts. I’ve noticed that the price of iteration, while seeming low, has long-term effects and how important first drafts are to developing the long-term technological architecture of anything that you build. The question is: How do we get better at first drafts?

There are a couple of things that seem to be working for me:

1. Starting with clear problem-oriented outcomes in mind.

Before building anything, I like to spend a lot of time making sure I’m actually building the right thing. I do this often by having long conversations with a separate AI tool or human, or both, about what I want to build. If I can’t get it to a crystal-clear outcome that I can describe first in just a single paragraph, then I’m probably not ready to start building yet.

A framework that really helps me is one that we conceptualized during the first iteration of Decoded Futures back in 2024: POP-IT. This problem-centric framework helps you focus back on an outcomes-oriented approach to building and has enabled me to build everything from my personal business operating system to mobile apps.

2. Working with my AI on order of operations.

As a non-classically trained engineer, this has been the hardest thing for me to learn over the years. How do I know what has to happen in which order first? I’m starting to develop a better intuition for this, but again, without having it in my head, it’s sometimes harder for me to know the right mode.

My favorite way to figure this out is to ask for help along the way. Here’s a brief infographic to get you started.

3. Timeboxing of any kind.

Today I work in three-hour morning blocks where I get things built, and then my afternoon blocks are where I work with humans. If I can’t complete a project in a morning block, that means I probably scoped it too big for the morning. If I can’t complete a bigger project in a week, that means I scoped too big. Just like having a bunch of half-finished blog posts in a Google Drive is worthless to me, having a bunch of half-finished apps is also worthless to me, so I’m trying to get better at pushing publish before I’m done, and that means timeboxing.

4. Practice, practice, practice.

Yesterday on LinkedIn, my friend Saadiq posted a counterpoint to AI slop generation, noting that it’s only after working through what “slop” feels like to you can you move past it to create AI-enabled artifacts that don’t feel quite so sloppy.

What he’s calling out is the importance of practicing. Without practice, you won’t improve outcomes. This has certainly been true for me.

Most of my best apps and AI-enabled builds that I still use today are actually projects that began with multiple first drafts, often starting with the exact same prompt. I like to look at it like how a writer might emulate the style of someone else: take note of what you like, take out what you didn’t like, and then try again. When I rebuilt my website last year, it took me about six first drafts over the course of several months in order to get to a first draft that felt good enough to continue iterating on.

5. Structured sprints.

The way that we run every workshop and builder experience at Build First is through an AI Power Hour that includes a structured sprint. It includes:

1. Context: The first third is the context gathering sprint. You need to understand what your problem is and teach the AI about that problem.
   

2. Logic: The middle bit is the logic layer. You need to decide what you want the AI to create, decide, or take action on your behalf.
   

3. Interface or Integrate: Once you can build this out, you can then decide whether you’re trying to build something that’s an interface-oriented, like an app or a website, or something that’s more integrated, like a headless autonomous agent.

Over time, I’ve noticed this feels like a flow state of building.

Like a Vinyasa yoga class, the core sequence remains the same whether you are a beginner or an expert, and whether you're building a fresh codebase or a tiny feature. It treats software development as a continuous practice, rather than a one-off task.

40 First Drafts

No matter if you’re a writer, a developer, or a regular at the yoga studio, the truth is that you can’t shortcut your way toward perfection. The iterative process of learning how to work from a blank state to a final product is the work. And on some level, this friction is an essential part to learning.

And what took you 40 first drafts the first time will likely take you 10 or less the second time.

So if you’re currently finding yourself in the midst of an iterative rabbithole, don’t stress too much. Take a deep breath, take a break, and give yourself a pat on the back. Sometimes we all need a little reminder: This is what it feels like to learn something new.

There was just one problem: It was 10 a.m. on game day and I had no idea where to start.

Luckily, by combining AI builders ( Replit and Gemini) with hyper-local WhatsApp groups, I was able to up a website and promotional image in real time, automating the logistics with minimal human effort.

It was a powerful reminder of how AI enables us to achieve the impossible.

Here’s how it went down:

Hyper-Local Apps for Niche Projects

This isn’t the first time I’ve spun up a new website or app in a hurry.

Back in January, I helped my block association, The Manhattan 75, coordinate group logistics around snowstorm support with a pop-up app called Storm Link.

We used this for approximately 5 days and helped coordinate several mid-tier snowstorm logistical nightmares, such as helping a neighbor grocery shop so he could take care of his austistic son, or shoveling problematic sidewalk areas before the city had time to take care of it.

It wasn’t a beautiful website, but it didn’t have to be. What it needed was functional utility…fast. As the block president, I help manage a hyper-local community of nearly 400 people in the neighborhood, which means I have niche distribution to a very specific audience.

The “Knicks in Five” pop-up party was the same thing. It took just three steps to coordinate a party from 0 - MVP in under 6 hours. And technology made it all possible.

Securing a projector:

A rooftop game watch party needs a projector. But we didn’t have one. At 8 a.m. on Saturday morning, I posted notes on Instagram, WhatsApp, and several text message threads with local parents. By 10 a.m., we had 2 projectors donated to the cause. Game on.

Spreading the word

I wanted an easy way to invite people without needing to text everyone individually, and it needed to be easy to share. The first thing I did was go to Gemini and spin up a quick “text-able” image that would make it easy for me, my husband, and our friends to spread the word.

Coordinating potluck and attendee logistics

Planning a potluck is usually a headache. You have to suggest specific items, track who is bringing what as the guest list grows, and ensure you’ve covered the basics…all without guests being able to coordinate among themselves. I didn’t have time to get my house game-day ready and figure out who was bringing what. So I let AI do the work for me.

In a long prompt to Replit about my party plans and goals, it spun up this quick custom RSVP page and host dashboard on the backend that let me approve or deny guests, get an accurate headcount, and coordinate what everyone was bringing in real time. All of this was displayed on the main website page for everyone to see. All I had to do was indicate my own requests and ask everyone else to do the same.

Let the AI Free Up Time for the Humans

It took less than an hour to spin up a mini-marketing campaign for my hyper-local pop-up party.

This meant I was able to spend the rest of the day doing the stuff the AI could decidedly not help me with: Cleaning the house, hanging a projector, and making a quick supplies run to the store.

As it turned out, lots of people in the neighborhood had been looking for a place to watch the game with fellow fans and families (that wasn’t a bar). We ended up hosting 40 people — 25 adults and 15 kids — who all stayed awake until the game-winning conclusion. Then we danced on the rooftop.

Ultimately, this wasn’t just about putting together a last-minute party; it was a glimpse into the future of community building. In the past, software was built to last for years and scale to millions. Today, AI allows us to build "disposable software,” specifically created to solve a highly specific problem for a specific group of people, right now.

By pairing the speed of AI with the high trust of a hyper-local network, we can eliminate the administrative friction that usually keeps neighbors apart. AI handled the logistics, which left the humans free to do what we do best: Connect, celebrate, and party on the rooftop until a 1 a.m.

New York City is on fire right now.

To be clear, that fire is orange and blue.

For the first time in the 15 years that I’ve lived here, walking around the city feels like walking around a college campus. It feels like one out of every five people is wearing orange and blue for the New York Knicks, who haven’t made it to the NBA finals in 27 years.

On the Upper West Side, the black and white cookies at Fairway have turned orange and blue. The bagels at Kossar’s have done the same. Hundreds of pop-up game watch parties have sprung up all over the city, and the people are quite literally taking to the streets. It’s reminding me of the important role that fandom plays in our society.

Most communities today form around opposition. They’re built around what people dislike, what they’re resisting, or who they’re fighting against. But fandom works differently. Rather than forming around polarization, fandom invites people to direct their attention toward something they love.

As I watched the second Knicks game in the NBA finals at home with my husband (admittedly, the first Knicks game that I’d ever watched), it struck me at how much fun all the fans were having, for so much of the game. Because the game was anyone’s to win or lose, the crowd and the fans from both teams spent equal amounts of time on the brink of hope.

We have so much these days to make us sad, that it’s easy to forget how fun it is to be a fan, how nice it feels to have someone to root for (even if they lose).

Fans show up year after year. They celebrate the wins and endure the losses. They invest hope (even in outcomes they can’t control). And in doing so, they create something bigger than any competition: A sense of belonging and shared identity. That’s something I think we all could use a little more of these days.

Near-Misses and Quick Rebounds

Knicks fans aren’t the only ones with a big week in New York City. Last night was also the 79th annual Tony Awards, a celebration of the top theatrical performances of the year on Broadway.

Thanks to my husband’s role as a Broadway sound designer, I’ve been hanging around this industry long enough to see how these award shows go. Some years, the ones you’re rooting for take it all home. Other times, the chips don’t quite land in your favor.

Like a lot of things, the winners always make it feel like success was inevitable. But when you’re close enough to the fire, you hear more about the near-misses. The nominations that never quite convert into wins.

Whether you’re a sports fan or a theatre aficionado, it’s easy to get disheartened by these close calls. Just chalk it up to luck, bad calls, or a rigged system, and move on. But as it turns out, that doesn’t work, either.

If every sore loser opts out of the winner’s charade, then winning no longer carries any value, and the industry as a whole loses its gravitational pull.

What I like about the Tony Awards is that it’s a celebration of what’s new and novel in theatre today. The insiders watch for the “who’s-who” of category winners, but the majority watch for a chance to get a peek at the newest shows on stage without every having to buy a ticket.

Maybe that’s the real magic of all that extra attention.

Whether it’s a televised industry spectacle or a basketball team breaking a decades-long drought, when you drop the entry point to zero, these high-stakes competitions serve a much nobler purpose: To usher in new fans.

What Keeps You Showing Up?

When I was younger, I used to think championships and awards existed primarily to determine who was the best. Now I’m not so sure.

The longer I spend around sports, theatre, technology, or really, any other competitive industry, the more I see wins and losses up close. As a founder, I win and lose deals every week. I’ve been nominated for awards that never materialized, and I’ve lucked into being in the right place at the right time for others. I've seen people get rich on a whim, and I've seen the transformative power of a single lottery-ticket exit.

A few observations:

• Some people leave the game as soon as they've won enough.

• Others leave after a painful loss.

• But then there are the ones who stick around, no matter what. (The true fans.)

What I notice about the ones who stay in the game the longest is that they aren’t playing to win. That is, not really. They play the game in order to stay in the game. Over time, they develop fan bases that like to play along, too.

Which brings us back to awards and competitions. If the point isn’t actually to quantifiably delineate “the best of the best,” maybe the real goal is something much more powerful: Competitions give people reason to care.

So, whether you win or lose, I hope at the very least that you get to experience it for yourself this week. And maybe in the weeks and months and years to come, too.

After all, it’s fun to be a fan. But it’s even better when you find something worth showing up for.

Over the past 18 months, I’ve been trying on a new role and persona as the founder of Build First, an AI learning lab I officially launched last summer.

While you might think that learning new technology or figuring out how to make money would have been the harder part, for me, the struggle has actually been something much more foundational: Rewiring my programming from “order muppet” to “chaos muppet.”

I imagine many of us who have worked in highly structured operational roles may relate to this feeling, so I decided to write this post to share more about it. For the vast majority of my career, I was the “responsible” one – the structured and organized person who helped visionary leaders structure their thoughts, organize their project plans, and get shit down. This took me very far for a very long time. I could have kept doing it forever. Except, there was just one problem: I wanted to start shit myself, too.

As it turns out, when you start something brand new, the world doesn’t care about perfectly groomed spreadsheets and details. They want to see ideas and movement. So for the past year, I’ve been teaching myself how to resist the urge to spend my free time “cleaning up” after myself, and instead edge myself toward a more uncertain path of what new thing might come next.

Every now and then I look over my shoulder and shudder at the metaphorical mess I feel like I’m leaving behind in my own kitchen. But then I remind myself: Innovating requires us all to find our inner chaos muppet.

Order Muppets and Chaos Muppets

The muppet theory of business goes a little like this:

• Order Muppets keep things organized and the wheels turning (think: Kermit the Frog, or Bert)

• Chaos Muppets break the rules, make things up as they go, and create pure energy (ie: Miss Piggy or Ernie)

The best businesses, teams, partnerships, and stories need a little bit of both. This is what makes shows like The Book of Mormon so brilliant and relatable. Elder Price (played by Andrew Rannells) is the classic order muppet, whereas Elder Cunningham (aka Josh Gad) breaks all the rules and makes things up as he goes.

You might have a “default muppet state” in life, but “founder mode” requires a little of both. Which means, if you’re setting off on your own (as more of us seem to be doing these days), you need to not only recognize your inner muppet, but actively find ways to counter-balance it.

As it turns out, this is easier said than done. There are some weeks that are far too orderly and I over-correct into the operational details of my work, and there are other weeks where I float above it all like a balloon, completely untethered to reality, with no one to hold the string.

Advice for Order Muppets Who Need More Chaos

Over the past five years, I’ve injected a lot more chaos into my life.

I’ve stayed at hacker houses at crypto conferences, I’ve hosted AI hackathons in castles in foreign countries, and I’ve even gone mini-viral by live-blogging a near-death experience from a hospital. I’m still not pure chaos, but I’m learning how to live with structured messes. Here are three things I recommend:

1. Spend time with chaos muppets.

One of the things that really helped me edge out of my “order muppet” ways was spending more time with more chaotic people (in the best way possible!). As it turns out, chaos muppets have a ton of fun, think outside the box, and aren’t afraid to try new things. Over several years of spending time among a lot of other founders, I learned a few things: Failure does not equal death, pivoting is totally natural, and creative collaborations are key to success. Seeing how natural-born “chaos muppets” adapted easily to changes in market dynamics made it safer for me to recognize the pattern in myself. I can now identify my own creative cycles in a much more nuanced way.

2. Push yourself one step beyond your comfort zone regularly.

Ninety-nine percent of the things I’m not doing aren’t due to lack of interest but due to fear of failure. Ask yourself, “What am I avoiding because I’m scared?” Then take the smallest step toward that thing. Unfortunately for me, I am very good at avoiding things that feel scary (often, by keeping myself too busy). If you find that you are spending more time rationalizing something vs. just doing it, you might be stuck in a loop, too. To reach escape velocity, you have to get comfortable being uncomfortable. I’ve been pushing myself toward uncomfortable things on a near-daily basis for the past couple of years. It’s a habit, which means it takes practice.

3. Build in time for bad ideas only.

I’ve always been envious of people who have tons of free time to “riff” on ideas just for fun. When I’m in an overly “order muppet” state of mind, I tell myself I don’t have time for fun or to play. But this is a massive problem, because playing invites creativity. By lowering the bar for “bad ideas only” this helps me not take myself too seriously, even if the idea is truly far-fetched, like my latest: Running a hackathon on an airplane.

Advice for Chaos Muppets Who Need More Order

Over the past year, I’ve pivoted three times. I’ve started and shelved projects that I loved. I made friends, I lost them. I lost money, I made some of it back. And then I started to find clients who wanted to buy the same thing – or packages of that thing. I made some software to help me manage it all, but I burned myself out trying to run all of the operations, delivery, and sales on my own. In other words, I’ve successfully flipped the script. But it’s no fun to live in constant free fall. Here are three ways to add a bit more order among the chaos.

1. Set goals and report out on progress to your close friends and network.

Late last year, I had a really stressful moment in my business where I needed advice from a lot of people in my network. While people were more than happy to help, I also heard one thing loud and clear: “If you want help during times of crisis, you need to keep us posted during the normal times too.” So in December, I started sending out monthly Build First updates to about 40 people in my close network. Most of these are fellow founder friends who have “been there, done that” or investors who have better pattern recognition than I do about state changes in founder mode.

This has been an incredibly helpful exercise for me. For the first time in years, I set monthly revenue targets for myself, and I report out on them. I share my wins and my losses, and the forcing function of sitting down monthly to organize my thoughts is often more helpful than any individual response. I also feel much better knowing that there are at least a few other people out there who have some awareness about the state of my play in my chaotic world.

2. Find ways to get excited about the tedious stuff.

At the beginning of this year, I was feeling incredibly underwater because while I had finally started to close regular business across a healthy client list, I was getting completely stuck in operations, logistics, invoicing, and contract management. It was really hard for me to get excited about Build First “out there” while I was drowning in my own paperwork and bookkeeping (alone) “in here.”

Luckily for me, I found that while doing the day-to-day logistics was starting to feel tedious, building AI agents to help me do that work kept it interesting. I leaned into my own advice to others and created an end-to-end business operating system for myself, linked up to my email, calendar, call transcripts, and Google Drive. Today, I am at about a 3-click proposal, 2-click contract writer, and 1-click invoice generator. Now I get to teach other people how to do that, too. A win-win. If you can’t find ways to get yourself excited for the tedious work, it’ll be really hard to stay on top of it long term. At least until I can afford more human help to get me to the next level.

Build in structured windows for “cleanup mode.”
Now that I have an actual business to run, I realized I needed to start acting like Build First was my job, and not just my chaotic side project. So this year I started to bring back order into my week. Mondays are for invoicing and contracts, Fridays are for creative fresh builds, Tuesdays - Thursdays are for client work, business development, and pipeline management. While this isn’t a perfect fit every week, it helps me to have a relative schedule in mind that includes some of the routine “maintenance” work.

It’s All a Game

One of the most powerful things I’ve learned from enmeshing myself among such a wide variety of order muppets and chaos muppets in my career is also one of the simplest: “Have fun.”

“Nothing that happens in business is your real life.”

I was initially pretty shocked to hear Gary Vaynerchuk (founder of Vayner Media, among others) say this on an Instagram reel with Tara Keeney. He’s the guy who built a business empire around his own brand of being Gary V. How could he not consider that real life?

But the way he concludes is this: It’s all a game. If you can’t distance yourself enough to recognize that the game of business is just like any other game, you’ll never be able to objectively look at your own situation and make the right calls. When you play games, you need to have fun. When you’re having fun, you think more creatively, and you’re more open to new possibilities and ideas.

As it turns out, it’s really hard to have fun when you’re worried about whether you can pay your bills. It’s really hard to make clear decisions when you can’t cover your own rent (or when it’s raining in your kitchen). I truly empathize with anyone who is in this camp right now. I’ve had some seriously low moments in the last year as I’ve begun to figure this out for myself, and I’ve also seen many friends get stuck along the way, opting to go back to full-time work (at least for the short term).

While it’s tempting to call it, wait for a rescue, or blame it on external factors, I think the real wins come from acknowledging that each of us has our own inner “order muppet” and “chaos muppet” waiting to come out. We just need to know which one to call up at the right moments to make sure the show keeps playing.

About a year ago, when walking down the street, my 4-year-old daughter overheard someone speaking another language.

“Is that Spanish?” she asked.

“Yes it is!” I said, impressed that she recognized it.

“Go and talk to her!” she urged me.

“But I don’t speak Spanish!” I told her.

“Just try, Mommy, TRY!” she implored me. “Just try to speak Spanish!”

“I’m afraid it doesn’t quite work that way.”

“I speak Spanish!” she continued. “Hola! See?”

She then proceeded to count from one to ten in Spanish, quite pleased with herself.

Of course, it takes more than knowing how to count to ten to consider yourself fluent in a language.

And… I hate to be the one to tell you this, but it takes more than daily use of ChatGPT to become fluent in AI.

Fluency starts with daily practice, yes. But it doesn’t end there. To become conversationally fluent in a new language requires more than memorizing the colors and typical greetings. You must practice reading. Writing. Maybe even immerse yourself abroad, enjoy the cultural nuances, pick up slang, watch movies and TV, conduct business, and even make friends.

I practiced French for 8 years and still didn’t consider myself fluent. I’ve been practicing AI for two years and I know I still have only barely scratched the surface of what’s possible.

The Phases of Language Fluency

Over the past year, I’ve worked with over 2,000 people to build their first AI-powered tools and applications. This is the AI adoption pyramid we use in all of our work:

It’s based less upon how frequently you use it and more about at what level you can build a solution.

This is a helpful way of anchoring people in a corporate training session on what they can expect to learn (ie: you can’t teach agentic workflows to a person who hasn’t yet understood the predicted behaviors of how LLMs behave in chat interface contexts). But it’s actually not that helpful in showing why it takes a long time to get from level one to level four.

So I started learning more about the phases of language fluency (with AI, of course). There are three primary pillars:

• Cognitive Fluency: How quickly your brain can retrieve words and grammar patterns

• Social Fluency: The ability to navigate cultural nuances, slang, and “unspoken” rules of conversation

• Functional Fluency: Being able to live your life entirely without a translator

I decided to map these to how it might look in the context of AI fluency. Here’s what we came up with:

By approaching AI literacy through the lens of language fluency, we move away from framing AI as a singular tool (ie: “use a calculator for math class”) and more as an integrated system into our whole lives.

Anyone who has taught a workshop on the tech can tell you that it’s nearly impossible to escape a group class without some existential dread conversation.

That’s because we instinctively sense that this is about something much greater than simple productivity gains. We aren't just learning a new software; we are literally rewiring our brains to adopt a new mental model for a non-deterministic companion. We are learning to interpret machine output with a new level of human discernment.

This matters because the ultimate opportunity isn't just "efficiency.” It’s Agency.

It is the chance to develop such a reflexive working relationship with AI that it can power your life and business alongside you. This represents an entirely new type of human-to-machine connection. The machine isn’t just processing your data; it’s acting on your intent.

The Long-Game of Language Literacy

When I was 21, my purse and passport were stolen from the riverbank along the Seine. At the police station in Paris where no one spoke English, I was led aside to a private room to file an official police report with an officer on duty. Eventually, my passport was recovered (by a local business owner who ran a restaurant on a boat), and I was able to broker a deal for my hotel to front the cost for a taxi home, expensed to my room, with a new credit card on file.

That’s the day I finally considered myself to be fluent in French.

In that moment, everything just clicked. It happened without thinking. But I couldn’t have gotten there without those 8 years of intense coursework, literature, study abroad program, and cultural immersion.

As I learned, fluency is a journey of immersion, not a destination reached by a single class.

And collective fluency is a next level up from there. If you expect your company or organization to operate collectively, in an AI-fluent way, that means that everyone must have some baseline level of cognitive, social, and functional fluency first.

Like it or not, we’re (still) humans. And we can’t skip over the basics. We must first learn how to solve problems for ourselves before we can solve problems for bigger teams, end users, and eventually, agents themselves.

But When Will You Actually Use AI?

When you ask adults why they decide to learn a new language, it often boils down to one thing: They needed to use it for something.

Maybe they got a job in a foreign city, started working with colleagues in a new country, took on a nanny who didn’t speak English, or recently booked a trip and wanted to feel conversationally proficient.

In other words, we instinctively recognize that learning a language is tied to some functional utility that unlocks access to something else.

The problem is, when you ask someone why they want to learn AI, the first answer most people share is often emotional, not practical:

• “I’m afraid of losing my job (to AI).” (FEAR)

• “My manager told me we all have to use it.” (PRESSURE)

• “I just want to know what I’m missing out on.” (FOMO)

• “I’ve heard it can help me be much more productive.” (HOPE)

• “I want to get my time back.” (LIBERATION)

• “I want to make something that didn’t exist before.” (AGENCY)

But true fluency only arrives when we move past these emotional reactions and start learning how to identify (and solve) AI-shaped problems.

Instead of just throwing a generic task at a chatbot and hoping for the best, an AI-Shaped Problem is a business or personal challenge that has been specifically “architected” to be solved by artificial intelligence rather than a human alone. They are scoped to fit the unique strengths of the technology.

It is the difference between asking an AI to “write an email” (a generic task) and designing a system where the AI monitors your inbox to resolve specific customer billing disputes automatically (an AI-shaped problem).

True AI fluency is the bridge between two skills: The instinct to spot an AI-shaped problem in the wild, and the agency to actually build the machine that solves it.

No amount of solo book studying could have prepared me to “pattern match” my way through that interaction in the Paris Police Station. I needed years of practice and real-world experience to find my way through it.

The same is true for your business. An organization cannot achieve Collective Fluency until enough individuals have the literacy to recognize where the AI-shaped problems are hiding in their daily workflows.

Individual fluency is a personal advantage. But when an entire team develops the reflexive ability to architect solutions in real-time, it becomes a corporate moat. You stop chasing the technology and start out-operating the competition.

The only question is: Are you ready to build?

What will you Build First?
If you’re ready to take your next steps toward functional fluency (whether individually or collectively), I would love to build together. Build First offers workshops and immersive studio build sessions designed to move you from theory to execution. We help you identify your highest-value patterns and get your AI-shaped problems from 0 to MVP in days, not quarters.

I hear some flavor of this sentiment several times a week now. It’s most common in 1-1 or small-group settings, where I guide people through an experience of building something new with AI no-code tools for the very first time. It turns out that some of us (maybe all of us?) really benefit from a little human hand-holding, particularly when we are doing things entirely unfamiliar or scary. Without it, we stop at the first friction point and never push past it.

I recognize this feeling of frustration-turned-gratitude because I felt it, too. What I’ve found is that, even with all of the infinite potential of what AI can do for us, there are some glitches in the (human) machine that keep getting in the way of progress. This is one of them.

These “glitches” aren’t things that can be commented away in code, or worked out with algorithms. They run much deeper and much more subtly through the fabric of the way we work, the way we learn, and the way we connect.

Here are the three most prominent ones I’m seeing.

#1: The Emotional Glitch

One year ago, I was the one frozen at the screen.

I felt paralyzed in my own progress in building my first apps, and there was not a YouTube tutorial in the world that could have pulled me out of it. In the end, it was 1-1 AI coaching and pair prompting that taught me how to build my way out.

The coaching was tactical and surgical; I showed my work to my coach Leon Coe and described my process and he gave me tips on how to approach it from an engineering-first mindset. The structured prompts he provided helped, but not as much as the human conversation he facilitated. Some days I just needed someone to encourage me to keep going, to hear me out on my MVPs and talk me through it before I spent the time on screen.

If coaching taught me how to think like an engineer, building an entire mobile app from scratch with a seasoned developer who had been there, done that, taught me how to have the creative confidence to invent something totally new. I know this may seem like a weird thing to need to learn, but it’s actually an unfamiliar practice for most of us. We’ve spent our digital lives trained as users of software, treating computers as things that are made for us. It is a radical, disorienting shift to suddenly be in the driver’s seat.

When the tables are turned and anyone has the potential to create, most of us have no clue how to even start. The truth is: I needed someone to sit with me and make sure I didn’t crash. And I know I’m not alone, which is what’s really at the heart of this emotional glitch.

There’s something decidedly human and important about being witnessed in this moment. It’s why, at our localhost:3000 gatherings (an AI no-code builder club I co-created with Louise Macfadyen) we see a recurring trend: Builders who have spent months coding in total isolation suddenly choose that room to debut their work. Whether or not the code works is besides the point. They are looking for the one thing an LLM can’t give: A human nod of recognition.

#2: The Social Glitch

The first time I met someone who was “AI pilled” was when I met Pasquale D’Silva in early 2023.

At the time, hot off the presses of ChatGPT’s November launch, he shared how he’d been able to replace his entire engineering org with AI no-code building, and how he predicted that more humans would seek out AI companions to help us make sense of the world around us. I’m not going to lie: At the time, I felt like I was meeting someone from 10 years in the future and told him as much.

Since I’ve doubled down on my own AI usage in the years that followed, the gap between me and Pasquale is narrowing in some ways, but the gap between me and my “non AI-pilled friends” is widening. To some, I’m the one “living in the future.” To others, I’m “so six months ago” in my approach.

While it’s becoming increasingly common to discuss the individual mental health issues that can result from “AI psychosis” (which include, among other things, a deluded state of reality), there’s also a subtler, interpersonal byproduct. This social glitch isn’t just about losing touch with reality; it’s about losing the ability to synchronize our realities.

Today, the people most absorbed in the technology (like me) are so preoccupied with the dopamine kick of constant creation that it’s becoming harder to prioritize meeting up IRL (“in real life”) over time building something else. When we do meet, it’s a high-energy, exhilarating and multi-threaded conversation that starts and ends with a million different possibilities. Almost as if we have a hard time saying no to each other (because we’ve all gotten so used to playing this infinite game of improv with our AI companions).

On the other side (but equally troubling), I’m finding it tougher to have conversations with humans on the complete other end of the spectrum and not mention AI. And anytime I’m asked even a basic human question (for instance, “How are you?”), my unfiltered laundry list of personal updates, thoughts, and plans overwhelm to the point of mental exhaustion. I’ve gotten so accustomed to dropping my full monologue onto every AI agent without skipping a beat that I now completely blow through a single human’s context window every damn time. I’ve become a high-bandwidth creature in a low-latency world. And the truth is: It’s making me a worse friend.

Add to this complexity the range of experiences people have with AI and you can see why this singular delusion is becoming more of a broader social concern.

It’s one thing to have an awkward dinner with a friend. But it’s another thing entirely when our schools, our businesses, and our neighborhoods are all running on different versions of the truth. When we lose the ability to reality-sync, we don’t just lose friends, we lose the ability to collectively build anything together.

#3: The Institutional Glitch

In the science fiction book, Klara and the Sun, there’s a concept of “lifted” humans.

These are people whose bodies have been biologically modified in some way to allow for them to take on excess capacity (or in theory make them smarter). In the book, this inevitably leads to all sorts of societal divisions between those who are lifted and those who are not. Certain colleges only invite access to “lifted” students, certain jobs are only available for “lifted” kids…you get the picture.

Our present-day AI age is not far off from this fictional depiction. A college degree has been the single-greatest way to “lift” yourself in our society, but adept usage of AI is quickly becoming the new currency of power and influence. The trouble is that institutions can’t keep up.

Universities with the deepest pockets are integrating AI into their curricula first, essentially resourcing their way into an AI-augmented upper class. Meanwhile, underfunded schools will likely pivot toward the pragmatic realism of trade schools. This leaves a hollowing out in the center. For everyone in the middle relying on a standard degree to signal their worth, the floor is falling out. When a four-year credential no longer guarantees a job, the so-called “standard path” becomes a dead end.

Companies aren’t doing much better. Through corporate workshops with Build First, I have a unique lens into how organizations around the globe are tackling AI head-on. I notice they often treat AI as a procurement problem rather than a cultural shift. But by focusing on allow-lists and enterprise licenses, organizations accidentally create an internal bifurcation among those who self-start and those who wait for permission. In this high-velocity era (where a company we trained on Gemini in January is already migrating to Claude by April), a “tools-first” mindset leads only to transience. When decades-old decision-making frameworks are applied to a future where the finish line is constantly moving, actual adoption plummets, and lasting skills never take root.

Teaching AI is a lot less like teaching a tool and a lot more like learning a language. You need more than a vocab book and a single experience at a fine dining restaurant to learn French. Unfortunately, the scope of learning a language is decidedly outside of any single organization, individual, or software’s ability to support it.

These institutional breaking points around resource allocation and procurement do more than just confuse people; they drown out the most important factor of all: The intrinsic motivation to learn it yourself.

The Patch

Last week, 20 students at the Zahn Innovation Center at City College presented real AI-powered solutions to business challenges introduced by André Ware, the founder of the nonprofit, BeeU NYC.

Just six weeks earlier, most of these students had never utilized AI outside from basic chatbot conversations. The studio class featured liberal arts students, not software engineers. But over just three rapidfire AI build sessions, students were prompted to quickly level up their own understanding of AI, using it first as a thought partner, then as a data visualizer, and ultimately as a problem-solving tool.

We leaned into the friction points that usually stop people (ie: confusion over tool usage, shifting project scopes, even group dynamics) and used them as the primary learning material. And in the end, every single group used AI in at least 4 different ways in their presentation: Research, design, website creation, and even deployable software.

The winning team created an integrated online booking system intended to dramatically streamline the registration flow for BeeU’s hive experience. Notably, nobody on this team had ever built an app before. Today, we are already talking about how to get it deployed live.

This is what it looks like to build in community. But Zahn was a sprint, and BeeU is one nonprofit. While this work is popping up across the city in builder clubs, hackathons, and one-off trainings hosted by nodes like Decoded Futures or AI small business hackathons through neighborhood labs like Welcome to Chinatown, it’s not enough to meet the tidal wave of demand. Many of these spaces are often temporary, and it’s impossible to patch a systemic glitch when we’re still working in silos.

That’s what I want to build next. Imagine a storefront in New York City where a florist, a laundromat owner, a technologist, and a college student sit at the same table on a Wednesday afternoon and ship something real by Friday. Not a course, not a coworking space. Just a room where the emotional, social, and institutional glitches all get answered by the same simple thing: People working on problems they actually have, together.

AI literacy isn’t taught, it’s practiced. Ideally, in public, in place, with people who have real problems and real neighbors. The storefront isn’t a venue for AI training; it’s the answer to all three glitches at once. The emotional permission to start. The social fabric to keep going. The institutional alternative to waiting your turn.

When we stop asking for a curriculum and start creating our own outcomes, the institutional divide starts to dissolve. We aren’t waiting to be lifted; we are architecting our own ladders.

If you want a seat at that table, let’s build.

More of my friends than ever before are choosing to go independent in their careers. Maybe you’re fed up with corporate culture and are officially splintering off from the 9-5. Maybe you were recently laid off but you’ve decided to consult vs. jump into the next full-time job. Or maybe you’ve tasted what’s possible with AI no-code building, an now you’re opting to make something from scratch yourself.

This is undeniably my favorite thing happening amidst this time of serious political, economic, and social upheaval. But even with these mass movement indicators on our side, it still doesn’t change the reality: It’s hard to start things.

I’ve been watching and helping people start things for most of my career. But I didn’t fully commit to the bit myself until a few years back – first, with fractional work, then with community work, and now with entrepreneurship.

Since I started making (and sunsetting) apps last year, then building a brand ethos around Build First, I’ve learned a lot of things from the ground floor of season one of founder mode. I hope many more people go down this path, as it’s what I believe will contribute toward a changing texture of leadership that the world desperately needs right now. So today I wanted to take the time to share a few reflections in case you’re considering going this route, too.

Here are ten things I wish I knew about starting things.

1. You can just start things.

There’s a moment as a parent where you move your kid from the fully contained “crib mode” (ie: 4 walls up, no exit or escape without help) to “big kid bed mode.” In big kid bed mode, your kid has full domain over their bed and their room, which means they can get out of bed and then return back to it independently. But there’s a period of time — and it’s different for every kid — when suddenly a light bulb moment goes on and the kid realizes, “Wait, if I can leave my bed whenever I want, I can also leave my bedroom.” (Cue: Parental panic.)

This earth-shattering realization for a child is not unlike the paradigm shift an adult feels when they finally notice the walls they’ve built around their own potential. So just in case you need to hear it from me first: You do not need to wait for someone else to come and get you when it’s time to wake up. You can just start things.

Depending how deeply ingrained your prior work culture had been, this powerful paradigm shift takes time to fully internalize. But it’s a critical first step toward feeling enough agency and confidence to actually succeed.

2. It doesn’t make sense at first.

The first thing I decided to start (and name, which is important to do for a lot of reasons) was The Manhattan 75 block association. This was confusing to a lot of people. (“It’s not a company, it’s not a business, it’s not tech. What’s the point?”) But the truth is that I knew I needed to feel grounded in a community that outlived whatever transience existed at work. Doing the startup process in another community first also gave me a very helpful blueprint for how to read and interpret signals in other contexts. As it turns out, starting an AI-first company isn’t straightforward, either, so this was a great practice round for me.

I wish I could give you more advice on how to decide what thing doesn’t make sense (for real) vs. what thing just doesn’t make sense (for now). But I think the only thing to do is ask yourself, “Does this make sense to me?” And if it truly and honestly does (and you’re not lying to yourself about it), that’s good enough to start.

3. It’s cripplingly lonely at first.

Whether you’re standing on the street corner with your own lemonade stand with your kids or sending cold DM’s on LinkedIn from your shoebox office, you have to cross a lot of bridges on your own before you get your first high-five from another human who cares. While more than ever is possible for solopreneurs these days, it doesn’t shortcut the solitary work that happens in the early days of any new idea. If you don’t know how to get yourself back to the table even after things fall apart again and again, you’ll never make it to season two.

I have a lot of ways that I keep myself motivated without someone else sharing the same physical space as me, and every one of my founder friends has their own version of this. (Gym time, bath house visits, improv on the weekends, movie marathons, or socializing with friends.) We’ve all got different needs. It’s important to find your own formula that keeps you showing up.

4. It takes a lot longer than you probably think.

They say, “If you can dream it, you can achieve it.” But they don’t tell you how long it really takes. Unfortunately, I live in a chronically delusional state of reality that makes it seem like the thing I want is just within reach. I don’t think this is an unusual feeling. I’ve noticed a lot of founders share this “just around the corner” mindset, only to wear themselves out just as things are getting started (or to give up just before they reach the precipice). While a little self-delusion is essential for anyone starting up for themselves, anchoring in reality (and other real human timelines) is the only way to sustain the pace.

One really good use of advice (either from an AI or a human) is to use it to manage expectations on both process and time horizons. I like to ask people, “How long did that take?” or even watch peers who started around the same time as me track their own benchmarks of progress. And if you struggle with knowing what steps 2-8 look like on a 10-step checklist to achieving your dreams, AI is a great first place to go to break down some of the milestones that need to happen in the middle.

5. If you don’t document your goals (and check back on them), you’ll risk missing your own milestones.

I’ve worked in highly autonomous roles for most of my career, so I’m no stranger to self-directed projects. But the milestones you clear on your own are very quiet and rarely celebrated externally. Even working fractionally carries some degree of closure and acknowledgement (typically, at the end of a project). But as a founder, when you’re always looking ahead to what’s next, it’s easy to forget how far you’ve already come. Whether you practice this through peer group chatter, writing out loud, or even your own celebrations for one, it’s important to witness your own leveling up.

Most of my serious founder friends have a habit of either personal or semi-public reflecting to hold themselves accountable. I started sending monthly updates to my close network and peers. The truth is that it doesn’t matter how many of them actually read it; the habit of self-benchmarking is good enough for me to keep it up.

6. Your intrinsic motivation matters more than any extrinsic validation.

I personally find it impossible to listen deeply to the problems I actually care about when I invite too many other people into the process in the early days. It’s hard to sit down and ask yourself what you really want to build; it’s a lot easier to ask other people what they think you should do next. This part of why taking outside capital too early is a big problem for early stage founders (and why I resisted the urge to do so). If you don’t know your own “why,” then you’ll be stuck managing the expectations and desires of your early supporters or investors, rather than figure things out for yourself.

Consider artists who take retreats to write their books, or even musicians who go dark on social media during their “studio build” weeks when creating a new album. This is a signal of what it takes to carry intrinsic motivation during those early days of creation, where you keep your ideas quieter and in a smaller circle (starting with yourself) before you expand out again.

7. Public perception comes at a lag from current reality.

I’ve rebranded myself enough times to know how long it takes my network to “catch up” to my current work. What I’ve noticed is that there is a lag between what people see and what’s actually happening on the ground floor. Today I use the signal from my external network not as an “end all, be all” but more as one more data point in the tea leaves of reading the market (and my place in it). And just in case anyone out there needs to hear it – even if you’re “crushing it on LinkedIn” your bank account may still need time to catch up.

A few things that help me with the public signaling are to pay closer attention to how other signal. What are they LinkedIn posts that other founders, hustlers, and solo operators are doing today, and what might that mean for the place they want their network to see them tomorrow? How can you use the public commons to write your own narrative?

8. External signaling is incredibly distracting.

Tied to the previous point, I find that excessive availability and openness is only good to a certain point. As a hyper-connector, I find it quite natural to book up my weeks with well-meaning meetings of people who “just want to catch up.” But then I get from Monday to Friday without any demonstrated progress or next steps. I’ve gotten much more aggressive on my calendar blocking and I try to move past anything on the internet (the good or the bad) as fast as I can.

It can be tempting to listen to the nice things people tell you as validation of accomplishment. But I also find that (due to the lag), if you treat any positive external signal as an excuse to take your foot off the gas, you’ll be pretty upset a few months later when the leads dry up. And the reality is, nobody knows your business better than you. Inversely, if you treat any negative signal as an indication to give up (particularly if that negative signal comes from a person who is not part of your intended buyer audience), you may stop before you’ve given yourself permission to start. Overall, it’s just a giant distraction.

9. Following the money is a very clear way to find the right signal.

I never used to understand why people said it was helpful to put up paywalls and start charging for products early on. Now I do. As it turns out, adding price tags to things is a clear way to weed out the true believers from the fair-weather fans. I wish I did this earlier, in basically every single instance. After all, if you want to start a business, you need to know what people will pay for, how much you need, and what kind of people and institutions might want to support it.

One of the biggest ways that I have changed my perception around this is to just talk about money a lot more with my peers. While initially this felt like a deeply uncomfortable discourse, I’ve notice that the people who operate at the upper echelons of economic prosperity talk about money matters all the time. So the truth is that if you do not talk about it, you are getting in your own way. Today I have a pretty good sense of what the hourly rates are for people with my level of domain expertise, what people will pay for AI trainings and small batch software today, and also how much money it takes to actually lift something off the ground.

10. Staying in the game is the only way to play the game.

It takes a long time to get to the starting line. Really long. But as I learned, all of that it still just season one. Season two is where the game play actually gets started. And I’m finding this year, I’m less focused on hitting a certain unrealistic threshold and more focused on staying in the game for long enough to maximize the surface area of my luck. You can’t get lucky if you don’t stick around long enough to throw out longshots. So this year for me is all about holding the line while I lob out the next set of experiments to see which one will carry me into season three. That’s how you go from starting something to sticking with it.

After reading Deloitte’s 2026 State of AI in the Enterprise report this week, one stat stood out to me: 66% of organizations report productivity gains from AI, but only 20% have actually grown revenue from it. Meanwhile, 74% say they hope to.

The word, “hope” is doing a lot of work in that sentence.

Deloitte’s own framing is that adoption isn’t the main challenge anymore. Instead the pressure is now operational (ie: data systems, governance, workforce structures, all the boring plumbing...)

But I’d push one layer deeper than that.

Most teams aren’t stuck on AI. They’re stuck on the problem.

Here’s what I see over and over when I run Build First workshops:

Someone walks in with a giant, fuzzy, “we need to use AI for something” mandate from their boss or their board. They bought the licenses, attended the webinars, and developed the internal policies.

But when I ask, “Okay, so what’s the first thing you actually want to build?” They freeze up.

To be fair, this isn’t for lack of ideas. Often, I notice the exact opposite to be true. They start with too many ideas, or ideas which are too big. Which unfortunately often means that none of them are scoped small enough to actually ship.

This isn’t an AI literacy problem. It’s a problem-definition problem.

Enter: POP-IT

This is exactly why I built the POP-IT framework, which is a design thinking framework for AI, developed in partnership with Decoded Futures. It’s the backbone of every workshop I run, and it’s deliberately simple because the work of shaping a real problem is already hard enough.

Five letters. Five questions.

• P - Problem. What is the problem you are trying to solve? Not a vague aspiration, this should be a a specific bottleneck affecting a real person or team.

• O - Output. What do you want to see from the AI when you’re done? A brief, a CSV, a creative asset, a mini-app. Picture exactly what success looks like in your hands.

• P - Prompt. What instructions does the AI need? I like to think of this part as if I’m briefing a smart intern. It’s where the role, steps, and clarification questions come into play. (It’s also the part where you need to learn how to use the AI to write a better prompt.)

• I - Input. What context, documentation, examples, or data can you share? This is the linchpin of making mini-apps bespoke, hyper-customized and actually work for you. The more relevant context you feed in, the better.

• T - Test. How will you know it’s working? I like to imagine in my head what I think the AI is going to do next, then see how close it gets. This is also the step where it’s important to practice iterations to improve the ouput.

You’d be amazed how many hours of wheel-spinning those five questions save.

Start with the Problem

Roughly 60% of every workshop I run is spent on the first step.

Not prompt engineering. Not tool selection. Not which model is better at what: Problem shaping.

Scoping things down. Picking the one small thing. Finding the AI-shaped problem hidden inside the fuzzy mandate. Turning “we should use AI for marketing” into “I want to automatically turn every workshop transcript into a LinkedIn draft in my voice by Friday.”

That’s the actual work. And once you’ve done the pre-work (naming the problem, defining the output, gathering your input...) the building is the easy part.

(By the way, this is also why the MIT “95% of generative AI pilots fail” stat doesn’t shock me. Of course they fail. Most of them were never real problems to begin with.) So if you’re new to AI, or if you’re struggling to make AI work for you at your organization. consider this:

Maybe you don’t have an AI problem. Maybe you have a problem problem.

Most organizations haven’t stopped long enough to ask the quiet question out loud: “What exactly are we trying to build, and what will happen if we get it right?”

That’s why it really matters what you Build First. And it’s where every workshop starts.

If your team is stuck in “AI for show” mode with lots of tools, lots of decks, zero shipped builds, let’s fix that. Book a Build First workshop or drop me a line at bethany@buildfirst.ai. We’ll start with the first P. What will you Build First?

Pick what fits, and I’ll see you there.

APRIL

📍 Build Your Personal Operating System with Claude Code

Mon, April 14 | 10:00 – 10:45 AM PT | Virtual | Free
With: AI Snack Club

A quick primer on building a personal command center with Claude Code. I’ll show you “how it started” and “how it’s going” with the OS that runs Build First… then help you get started on yours. Great if you’ve been curious about AI coding tools but haven’t known where to start.

Register: https://luma.com/ug1puw0e

📍 localhost:3000

Tue, April 15 | 10:00 AM – 6:00 PM ET | In-Person · NYC | Free

One of my favorite formats is a full-day build sprint where you go from idea to working prototype. If you’ve been spending too many hours alone with your AI’s, we invite you to join us for an all-day builder coworking day where you have 8 hours to build something real among peers. Show it at the end.

Spaces limited. Register at https://luma.com/uaf3o92y

📍 From 0-MVP: Build a MiniApp For Your Business

Thu, April 16 | 1:00 – 3:30 PM ET | Virtual | $250

A hands-on afternoon workshop where you stop waiting for the perfect tool and build a small, working one instead. We’ll take a real problem from your work and build something functional around it, starting simple (with a chatbot) and building up (to a frontend interface). You can think of this like a cooking class for building: Once you know the right way to organize your ingredients and start cooking, you’ll be able to do it again on your own.

Register: https://buildfirst.ai/events/zero-to-mvp

📍 AI Studio Build Day: Turn Your Data into a Custom Web App

Mon, April 27 | 4:00 – 8:00 PM ET | In-Person · DUMBO, Brooklyn | $25

With: Studio Rodrigo

Bring a spreadsheet. Leave with a working web app. No coding experience required, just a dataset and something you want to make sense of. We’ll work together for a couple of hours on our independent builds, then invite you to do a real-time demo following to show off how far you got — and what you intend to build next.

Register: https://luma.com/8z3azdxj

📍 From Localhost to Live: Ship Your First App with Claude Code

Wed, April 29 | 3:00 – 4:30 PM ET | Virtual | $60

With: Megha Torpunuri

You’ve got something running locally. Now it’s time to ship it. This Build First co-facilitator, Megha Torpunuri, is an experienced founder, product leader, and builder who now runs her own flower business through a backend infrastructure she built herself. In this session, she’ll show you how to take any project from your machine to a real, live URL.

Register: https://luma.com/a7ihryr1

MAY

📍 Building with the Bard: Remixing Shakespeare

Fri, May 15 | 9:30 AM – 12:00 PM ET | In-Person · NYC

AI Week NY Official Event

We’re downloading the complete works of Shakespeare and feeding them into Claude Code. What comes out is up to you. Try a chatbot who speaks in iambic pentameter, a modern-day retelling of a classic plot, an interactive visualization across the whole canon. Come find out.

Register: https://luma.com/vjihrr24

JUNE

📍 Design Thinking Meets AI: A Book Talk and Live Build

Mon, June 1 | 5:00 – 7:30 PM ET | In-Person · DUMBO, Brooklyn

NY Tech Week Official Event · With Louise Macfadyen and Studio Rodrigo

Louise Macfadyen, the author of Designing AI Interfaces (a new O’Reilly release), joins me for a conversation about what it actually means to design AI-native systems. Then we build something live. Copies of the book will be available at the event.

Register: https://partiful.com/e/N7nJMJdM3bIOtaUWfcmN

📍 The One-Person Operating System: Build Your AI-Powered Business Brain

Thu, June 4 | 11:30 AM – 2:30 PM ET | In-Person · NYC

NY Tech Week Official Event · With: Lightswitch

In this lunchtime builder session, I’ll pull back the curtain on the custom operating system I built (with AI) to run Build First (from deal tracking to client ops to daily priorities). Then Sunny Israni (Co-Founder & CTO, Lightswitch) takes you inside Claude Code with a hands-on engineering primer. Together, we’ll build through two mini-sprints and mini-lessons. You’ll leave with the foundation of your own one-person OS.

Register: https://partiful.com/e/EhOH3OE5AVVA5aFrUrWR

📍 Building with the Bard: Remixing Shakespeare

Fri, June 5 | 9:30 AM – 12:00 PM ET | In-Person · NYC

NY Tech Week Official Event

A second run of our Shakespeare build session, this time as part of NY Tech Week. Same format, fresh energy.

Register: https://partiful.com/e/Mt1yixuqsZw3l7qHMq3T

Want to build something else?

We’re looking for partner organizations for summer and fall programming. If you’re looking to bring a Build First workshop to your network or community, give a shout at bethany@buildfirst.ai. See all events: https://buildfirst.ai/events