My Story
The Gift of Flight
I was nineteen years old, selling printing services door-to-door in Metro Detroit, when I learned the only sales technique that ever mattered: that nobody buys what you're selling. They buy how you make them feel about what they already need. And they buy from someone they have a relationship with.
My father had founded a printing company and I was his first employee. I drove a used car through industrial parks in the winter, knocked on doors that smelled like machine oil and paper dust, and sold pens and mugs and whatever else would move. Door-to-door sales strips away pretension and teaches you about rejection. I was an art kid. BFA from Oakland University, not a BS in computer science. I had no business being in a sales call, except that I was good at it. I could feel what people needed before they said it.
That instinct — reading the room, sifting consumer demand — would carry me through the next twenty-two years. But I didn't know that yet. I was just a girl from Detroit with a chip on her shoulder and a work ethic that would outmatch my peers.
My father was a solo entrepreneur and a product inventor with an endless capacity for reinvention, and I absorbed his curiosity like a genetic trait. Before I understood what entrepreneurship was, I understood what it looked like in practice: my father and his ruled yellow memo pad, scribbling away in capital letters, working through problems - problems that were simply businesses that hadn’t been built yet. The observation that a person who pays attention, stays curious, who refuses to stop learning, can build a life entirely on her own terms.
My dad’s company wasn't a venture-backed startup with a pitch deck and a burn rate. It was a small business in the most traditional sense. I learned about supply chains by managing distributors. I learned about customers by servicing them directly, face to face, in a city that understood what it meant to make things with your hands. I learned the difference between a sale and a relationship — the lesson that the easy first transaction is meaningless if the second one doesn't follow. The lessons I absorbed in those early years — about cash flow, about the dignity of building something tangible — would prove a durable framework for what I would later encounter in the world of technology startups.
Detroit’s blue-collar ethos influenced me. Growing up in a region defined by its work ethic, from the automobile industry, to the complicated American story of industrial rise and urban reinvention, I developed an intuitive understanding of how physical products move through the world. I understood merchandising not as an abstraction on a slide deck but as something I could see and touch: shelves, packaging, point-of-sale displays, the human interaction at the counter. That literacy in physical products would matter at every company I would ever build.
—
I was twenty-three when Henri Seydoux hired me.
Henri is the co-owner of Christian Louboutin shoes, a French nepo baby and thus a rebellious self-made engineer. When I met him, he was in the process of building Parrot into a €350 million technology company in Paris and expanding to North America. I was employee number three in the Americas office — thanks to an internship I got from my dad's best friend.
Henri was the quintessential eccentric Frenchman and an intimidating boss. He would casually climb onto the conference table, lie flat like a corpse and nap between press briefings. He would hand me his credit card in Vegas with a discreet directive to help find the engineering team ‘entertainment’ after dinner. HR practices tend to be lost at startups anyway, and I learned early how to advocate for myself — though the directive was innocent; this kind of request wasn't culturally faux pas in France. Henri often demanded that I only speak French with him. I learned the language and it served me well over the seven years I spent traveling back and forth to Paris.
Those years shaped my perspective on international business. It was the first truly multicultural, globally built company I'd worked inside — and where I was particularly inspired by the je ne sais quoi of my older female European colleagues, who approached leadership, age and beauty with confidence and ease. I couldn't wait to one day be like one of them.
I had humble beginnings at the company. Originally I was brought in to compile market research for the VP of communications. I attended more than a dozen press events a year, hosted exhibitions, organized trade shows. One advantage of running North America was that my territory was ahead of the technology fads and trends before headquarters in Paris. I served as the early-adopter beacon — traveling frequently to present my findings, proposals, and opinions about what might work in what promised to be the company's largest market.
Then 2008 happened.
The financial crisis hit Parrot's North American office like a wrecking ball. Henri walked through the office and let two-thirds of the company go in a single day. When the dust settled, it was me, the sales director, and a quirky CFO. His directive contained no mention of promotion, no raise, no bonus: keep the company alive; keep your jobs.
I kept the company alive.
Two years later, I launched the world's first consumer-grade augmented reality drone — the Parrot AR.Drone — at CES 2010. The timing was cosmic in its inconvenience. The product that would come to define an entire market of consumer drones arrived at the same moment my life was being remade by motherhood. My only son was born during that launch.
On a whim, I called my CMO about a "new startup called YouTube" where I'd already set up a UGC channel. We collected and reposted more than 12,000 user-generated videos of drone footage in the first year — mostly from GoPro hobbyists duct-taping cameras to homemade rigs, posting shaky breathtaking footage that nobody had seen before. That demand drove our first ten percent of global annual sales online, while my colleagues in Paris were still skeptical that "ecommerce" would be successful. I understood, before most of my peers, that the hobbyist community wasn't just our customer base. They were our evangelists.
Great leaders understand the psychology of what drives consumer behavior. I realized that our customers weren't just buying a product.
They were buying the gift of flight.
That insight taught me everything I needed to know about marketing bleeding edge technology. It's never about the specs. It's about the feeling. We engineered the AR.Drone for augmented reality gameplay. But it sold on nostalgia — the dream every little boy carries into adulthood, the one human beings have had since we first looked up at the sky.
The personal cost of the launch is something I deeply remember. I returned to a rigorous travel schedule throughout the Americas no more than eight weeks after my son was born — negotiating and preparing my market, retailers, partners, and stores for the coming years of sales. The company didn't have a maternity policy. I was one of only a few women on the North American market team.
I launched more than fifty products in my seven years at Parrot, across four product lines. Most of them, nobody has ever heard of — they were packed with emerging tech that hit the market too early or too late. I learned how to design and market technology that added real value to a customer's life, and why it failed when the product made the tech the hero instead of the human.
I managed a $7 million marketing budget. I grew the North American subsidiary from nothing to $40 million at its peak. I drove more than $350 million in drone sales for the parent company. My retail buyers used to say, "If Kelly's selling it, I'm buying it." It didn't much matter what it was. I knew how to make them money. Parrot went public on Euronext during my tenure – my first IPO experience. Not the last.
Parrot taught me something I'd carry into every room for the rest of my career: I was the translator between what engineers built and what humans actually wanted.
Falling in Love with Robots
In 2012, I joined Sphero in Boulder, Colorado, as employee number twenty-five and the fourth member of the executive team. Sphero made the little rolling ball you might remember from an Apple store, or the BB-8 droid from Star Wars. It had come through Techstars' class of 2010 and recently closed its Series A. The investors believed. The press was interested. The company was still trying to figure out what it was.
Was Sphero a toy company? A robotics company? An education company? My answer — which I would help prove — was all three at once. The magic was in refusing to choose.
I arrived as VP of Marketing. Within a year, I'd been promoted to VP of Sales and Marketing. By year three, I was also leading product development and design. I was responsible for the company P&L, margin structure and global pricing strategy. I managed forty-five employees. I sat in a boardroom with Brad Feld — the legendary Foundry Group partner and Techstars co-founder — and his beloved CEO for four years, writing quarterly board memos and absorbing, by osmosis, what it meant to build a venture-backed company at scale.
I grew our startup from $10 million to over $150 million in gross revenue in roughly three years. I expanded our reach to the top forty global markets. I convinced retail buyers across more than 20,000 points of sale — Apple, Colette, Harrods, Nordstrom, Target, Amazon, Hilton, Macy's, every major carrier in the world — to carry our products. I negotiated the contract that made Sphero the exclusive top-selling toy in all 420 global Apple retail locations. I architected partnerships with Disney and Lucasfilm that would produce one of the most memorable consumer electronics launches in modern memory.
On September 4, 2015 — Force Friday — we launched BB-8, an app-enabled droid modeled after the character from The Force Awakens. Authentic movement. Adaptive personality. Holographic communication simulation. Voice recognition. A toy that felt alive. We used Snapchat as our key launch channel, capturing the joy and excitement of the global event. BB-8 sold out everywhere within hours – fortunately we had stock waiting.
But the lesson under the launch was the same one I'd learned at Parrot: a product's emotional resonance matters at least as much as its technical specifications. Children didn't buy BB-8 because they understood its Bluetooth protocol. They bought it because it felt like having a friend from the movies in their bedroom. The most successful brand collaborations aren't extractions. BB-8 didn't water down Star Wars and Star Wars didn't dilute Sphero — both got more of what they were, and the customer got the friend from the movies.
We won the Auggie Award for Best Game at AWE in 2014. CES Innovations Award. CES Hot Stuff Award. Popular Science Product of the Future. Shorty’s. The trophies stacked up.
—
But the work I'm proudest of from those years isn't a revenue number or a product launch.
I launched forty iOS games at Sphero, including software that teaches young women and men how to code. The SPRK program I helped initiate is now in 40,000 schools worldwide. Sphero EDU — the platform — gave girls a chance to see themselves as engineers and programmers. Building a product experience that results in more female engineers has been one of my proudest accomplishments. Partly because I missed that encouragement at a young age myself. I could have just as easily used my full-ride academic scholarship to pursue Computer Engineering at Oakland University.
The cost of building at this altitude is real, and I paid it. Over the course of two years, I left my partner and my newborn son at home for stretches of seven to fourteen days at a time while we expanded across EMEA and APAC. The pace was relentless. The pressure of exponential growth — across teams, time zones, partners, and geographies — taught me how to manage stress at a level I had never experienced.
My marriage ended in divorce.
Incredible growth is not always marked by personal sacrifice — and for women, it doesn't have to be — but for me, it was. Anyone who tells you they built something extraordinary without losing something personal is either lying or lucky, and I am neither.
And still — I was a voracious growth leader, consuming everything the leadership would allow me to take ownership of. It was hard to know where to sort me between brand, marketing, sales, and product. One day, our CEO, Paul Berberian, was intuitive enough to sit me down and ask me about my career ambitions. Did I want to become a CMO or a CEO? He explained that if I took ownership of sales and owned our P&Ls, I'd be on track to become a CEO. The distinction between those two paths was subtle, but it was the difference between functional expertise and general management.
Without hesitation, I told him I wanted to be a CEO.
Two years later, Paul recognized that it was time for me to run my own company. He began searching for a role inside the Foundry Group family and Techstars portfolio.
Then a phone call changed everything.
Boarding a Rocketship
I was in a meeting with Jony Ive's design team at Apple when I stepped out to take a call from someone high up at Snapchat.
I had taken an interim CMO role at ROLI — a London-based music technology company backed by Foundry Group and Pharrell Williams. I got the NOISE app to number one on the Apple App Store for three weeks. I was about to sign for a flat in London in two days. I was not interested in advertising companies, not interested in social media platforms, not particularly interested in Los Angeles.
Then Snap called.
I was impressed by Snap's VP of Engineering. His resume included Google, Facebook, and Microsoft. They told me they'd exhausted candidates from Apple, Amazon, and Meta. That I was the ideal leader for Evan Spiegel's pet project – even though he didn't care for "marketing and advertising people." Frankly, I don’t either. Buzz words make bad friends.
I graciously declined.
They called again. And convinced me to get on a plane and fly down to meet Evan. Short story: I met Evan. I realized that a rocketship was taking off and that I had been offered an executive seat on it.
I took the job. Halted my British visa process. Convinced my ex-husband to move from Boulder, Colorado to Los Angeles without knowing exactly what I would be working on at Snap. I became one of three executive leaders inside SnapLab — a startup within Snap. Our job: convince teenagers around the world to wear a face camera.
—
Spectacles was my secret assignment. Camera sunglasses priced at $129.99 that recorded ten-second circular video clips from a first-person perspective. We put them in cute one-eyed yellow vending machines we called Snapbots. We sent the bots around the world by teasing Twitter followers about pop-up locations. Lines before dawn. A vending machine drop at the bottom of the Grand Canyon. The "wacky bot scavenger hunt" that mandated our buyers join the outdoor adventure theatre instead of queueing in a retail store for a new gadget.
This was not a product launch. It was a cultural event before most people had even held a pair. I had taken the lesson of the AR.Drone — that adoption arrives through delight, not specification sheets — and elevated it into an art form.
Taking the LAX red-eye weekly, I single-handedly oversaw the five-month construction of an eight-million-dollar pop-up retail shop in New York City — towering over the Apple store near Central Park. I managed operations for three retail pop-ups across NYC, Venice Beach, and London. I developed the circular content brand identity. I built Spectacles Live Stories. I worked directly with Evan Spiegel in weekly meetings to shape the product roadmap. The campaign was co-created with my longtime collaborator Rhenee Bartlett.
We sold 100,000 units — the same quantity as the iPod First Generation — with zero online direct-to-consumer sales. We won six Cannes Lions at the 2017 festival. Three Gold.
Privately, the campaign was a massive success. Evan often rewarded employees from least to greatest with fresh flowers on Monday mornings for a job-well-done. Hand-written notes and flowers came no fewer than four times during my tenure, recognizing my efforts.
Behind the scenes, I was learning something else entirely. I discovered Evan's deep anxiety about "intruding" on user stories with ads and swipe-up links. He believed — correctly, I think — that the moment a piece of content stopped feeling like content and started feeling like a delivery mechanism for advertising, the trust would evaporate. I worked tirelessly to create a frictionless Snap store, learning firsthand the extraordinary difficulty of turning a content platform into a commerce platform.
I watched, from the inside, as a company navigated the transition from startup to public company. The atmosphere inside the cabin of the rocketship was changing. Some people who had thrived in the scrappy pre-IPO years didn't recognize the company that emerged on the other side. The Spectacles team had been one of the last places where the Snap I'd been recruited to still existed — an executive sandbox where you could move fast, take risks the broader company couldn't, and ship a product that bordered on art. That sandbox was closing.
Having helped review the S-1 materials for the IPO, I knew that our camera glasses had convinced our investors that Snapchat could be more than just an app for sexting. The company culture was a mess. But Evan had a strong intuition for consumer demand, product design, and acquiring money and talent. The virality of the Spectacles launch created the hype-ramp for that spring's IPO. I stood in the bullpen as the bell rang.
Then the layoffs came.
Post-IPO, Snap let go of hundreds. I was among them. I found myself staring down the barrel and feeling paralyzed because I didn't understand how I got there. I had built. I had executed. I had delivered. We sold every product we made. And it didn't matter. Then publicly, the hardware story at Snap was muddied. The subsequent product launches were never the same.
The leadership conflict at Snap was not the kind that makes good copy — no feud, no coup, no single dramatic moment. It was the slow, structural friction between a founder who saw the company as a creative studio and a finance organization that needed it to behave like a public ad business. Evan ran on product instinct — sharp, opinionated, allergic to the compromises that look reasonable in a board deck and gut a product in the field. Other C-level adults in the room ran on the entirely sane belief that a publicly traded company has to have one clear story for Wall Street. Both of them were right. They were also pulling in directions that could not coexist forever. The hardware team — Evan's pet, by emotional default — was where that tension played out first, because we were the easiest thing to cut without breaking the rest of the business.
For me, the Spectacles experience crystallized convictions that would shape everything that followed. The go-to-market strategy is not separate from the product — it is part of the product. Timing matters enormously: Spectacles launched at the exact cultural moment when it could become a phenomenon, and a year earlier or a year later it might not have. And creating a wave is even harder than catching one.
Founder; Round One
After Snap, I had forty new executive roles lined up. Sweetgreen. Tamara Mellon. Lyft. Drive.ai. Melissa Cha, VP of Emerging Devices at Amazon, offered me a role leading go-to-market for an entire suite of projects. The comp packages were meaningful.
I turned them all down. The idea of building massive growth for another product didn’t feel as challenging anymore.
It was at that moment that I realized I didn't want to work for Amazon. I wanted to try my hand at building the next gen Amazon, myself.
I declined the offer and got to work.
—
The company started in 2017 as Petalfox — a consumer brand with a unique GTM strategy. The thesis was simple: what if you could buy anything with just a text message?
I'd been buying flowers for myself for years. Not for special occasions. For Tuesdays. I arranged them every week, a small ritual of beauty in the chaos of a career that moved at the speed of product launches and red-eye flights. I had noticed the same ritual in the places I loved most: Tokyo train stations where commuters grabbed bouquets on their way home, Paris streetside markets where flowers were as ordinary and essential as bread.
I also noticed an industry statistic that no one seemed to be acting on: 51% of all purchases in the multi-billion floral industry — were made by women, for themselves. This impulse buy happened most frequently at the grocery store, because no digital experience had been tailored for this behavior. The entire floral industry was optimized for Valentine's Day and Mother's Day — for guilt and obligation — when the actual majority of the market was something quieter and more personal: women investing in their own small moments of happiness.
I validated the MVP without a single line of code — using spreadsheets, manual process, and off-the-shelf software. SMS-based ordering. An emoji interface that felt more like texting a friend than navigating a website. A hundred thousand dollars in revenue. Two thousand users. Proof that the idea had legs. This led to an accidental first round of fundraising from Kate McAndrew and Charles Hudson. I had never run a fundraising process - which would prove to be extremely difficult as subsequent rounds came along.
By 2019, Petalfox had run its course as a standalone consumer brand. The unit economics of direct-to-consumer commerce with perishable inventory are unforgiving. But the idea at its core — that commerce could happen inside a conversation — had achieved early product market fit. And there were signs that conversions could become a powerful sales channel. Now I needed a way to experiment and scale our technology solution.
—
We rebranded to Whym in early 2020 and pivoted to B2B. At the time, every consumer product had attempted to adopt a monthly subscription model. Consumers were getting burned by having too much stuff, forgetting to skip their shipments or getting charged when they didn’t want to purchase. We created a solution for brands that solved this problem by letting customers simply sign up for a monthly text message for an individual product. They could reply GET, GIFT, or SKIP. Our software operated like a Shopify in a box for messaging — letting merchants engage consumers through text, with in-thread carts, frictionless payments, and a shopping experience that felt like talking to a friend.
Founded with Snapchat teammates Rhenee Bartlett and Ryan Hornberger, we assembled a venture capital and angel roster that read like a directory of Silicon Valley's consumer technology elite: individuals from Apple, Airbnb, Microsoft, Stripe, Venmo, Salesforce, Snap, Facebook, Red Bull, Spotify. Over the life of the company, we raised under $10 million to build payments, carts, messaging, digital shopping identities, and sell product intent data to over 500 merchants.
Then the world changed.
Beginning in 2021, the e-commerce landscape underwent a seismic shift. Apple's App Tracking Transparency changes disrupted advertising effectiveness - the bottom dropped out of Facebook Ads for a while. CFOs mandated reductions of up to ninety percent in SaaS spending - onboarding systems like Triple Whale in an attempt to track first and last touch attribution contributors. Shopify became even more protective of its lucrative payment products while their app store became a freemium race-to-the-bottom littered with off-shore solutions. Workforce reductions of twenty-five percent swept through the retail ecosystem. The market we had built Whym to serve was suddenly retrenching, and no amount of execution excellence could overcome macro headwinds of that magnitude.
We pivoted. And pivoted again. We released new features that generated a 350 percent increase in traffic to the consumer wallet. Consumer account creation surged from thirty-five new accounts per month to 1,250. At its peak, the platform reached 750,000 monthly active users with 77 million total user actions. Accounts paid from $35 to $299 per month. Two provisional patents filed. The numbers were impressive by any standard — except the standard that mattered most. The metrics that would unlock the next round of funding remained just out of reach.
The hardest truth was one I had to confront with characteristic honesty: I remained the primary closer on more than eighty percent of deals. My network. My relationships. My personal credibility. These were the company's most effective sales tools. But founder-led sales, by definition, does not scale. And we had made two pivots, with an unsuccessful third, that complicated the cap table and exhausted the patience of a market that had turned brutally cold.
—
In April 2023, we found Product-Market-Fit amidst the launch of chatGPT. But it was too late. We were way behind our Series A Annual Recurring Revenue (ARR) target. Our cap table was over six years old. And all new capital was flooding into AI. We had to start making hard decisions.
The board ran a retrospective that forced a difficult conclusion: Whym could not secure a Series A on its current trajectory. I ran an almost-impossibly close M&A process with FAANGs + forty of the top technology companies in the market. We reached final-round diligence with two strategic acquirers and came tantalizingly close to a meaningful acquisition.
Nearly.
I restructured the team, making painful cuts while maintaining the morale and dignity that my leadership style demanded. I explored every alternative path. And when there were no paths left, I did the thing that required more courage than any product launch or fundraising round: I wound the company up with grace – leaving ‘no stone unturned’ in attempting return to our capital partners.
Years of building, pivoting, surviving two market downturns, and holding a founding team together through it all — over. Seven years. The longest chapter. The one that taught me everything about myself — as a leader, as a fundraiser, as a non-technical founder. Through all of it, Rhenee and Ryan and I held what my mentor Alexis would later call an "unbreakable co-foundership" — a feat she described as very rare. Reviewing my leadership during this period, Alexis also wrote something that has stayed with me:
“You've shown that you can make decisions quickly and have resilience, especially when it comes to making hard decisions, and I saw you get better at doing this over time. I think this is a skill that some better-funded CEOs don't have because they can lean on longer runways to make decisions, and in doing so, they can end up hurting the culture of the team.”
Startups can achieve some level of adoption, traction and success – yet still fail. Much of success is determined by velocity – something we didn’t have. What killed our ability to grow at the right speed was a combination of decisions, market forces and age of our capital table. Sometimes the inability to raise your next round is that you haven’t compiled metrics that are on-par with your raising class. Be bolder. Burn [out] faster. There are iterations of Whym’s products and features that are being rebuilt by new startups in the market today.
The data layer underneath Whym — the idea that a person's shopping context, identity, and intent could become a portable thing they owned — didn't die when we wound the company down. It became a seed.
The Other Side of the Table
In 2024, after winding down Whym, I crossed to the side of the table I had never sat on.
I took an Entrepreneur in Residence role at Bright Ventures and worked closely with partners at M13 and Alpaca VC. For the first time in my career, I wasn't pitching — I was reading pitches. I combed data rooms. I wrote sixty investment memos on companies across dating, fashion resale, event marketplaces, live video retail, personal shopping assistants, and communication tools. Each memo was a forensic accounting: what the founder claimed, what the data actually said, and where the blind spots were buried.
Since joining the Techstars Family in 2012, I receive warm intros to between fifty and a hundred founders each year. The analysis wasn’t new. The vantage point was. From the operator seat, you're the brand fighting for shelf space. From the investor seat, you're the buyer deciding what makes it onto the shelf to compete with other brands.
A pattern emerged almost immediately. Nearly every company was trying to use an AI angle to fundraise – building features on top of foundation models, betting that better prompts or better fine-tuning would carve out a moat.
Each one — regardless of vertical — faced the same cold start problem. The AI becomes valuable only after months of sustained use — your routines, your preferences, your taste, accumulated slowly through real interactions. But almost no one sustains months of use of an AI that isn't valuable yet. The trap closes before it opens. Their AI didn't know the user. Not really. Not in any way that compounded over time.
Investors kept asking me variations of the same question: what's the consumer AI thesis that actually holds? I started to realize that I was identifying the harder problems and gaps in the market.
Alongside the investor work, I took on what I came to call a founder/CEO co-pilot role — a handful of operator engagements with companies scaling their go-to-market from pre-Series A toward $100M ARR. I sat with boards, rebuilt funnels, recalibrated merchandising, restructured pricing. I took board seats and advisory roles where participation could make a real dent.
What looked, on the surface, like the wandering year of a founder who had just closed a company was something else entirely. It was the most rigorous market research study I had ever run. By the end of 2024, I had a calibrated view of consumer AI from both sides of the table — and I was getting impatient.
Learning how to talk to computers
Over the next year, I vibe-coded eighty personal applications. Taste testing each agentic solution against the next.
A health agent that integrated my 23andMe data, Apple Health metrics, meal photos, and peptide schedules using Letta's memGPT. A meal prep workflow spanning Memories.ai, Apple Notes, ChatGPT, OCR screenshots of my pantry, and two Instacart lists — one calibrated for my high-protein diet, one for my sixteen-year-old's gluten-free needs. A closet reconstruction project that scraped email order confirmations, remade product images through Google Whisk, and connected to a resale listing agent. A WhatsApp travel agent that aggregated group chat inspiration into shared Google Maps lists and babysat reservation openings. A runner's PR dashboard that correlated Garmin output, Apple Sleep, weather, terrain, and time of day to find the conditions where my son runs his fastest cross-country times. A co-parenting dashboard that filters school emails, tracks grades and homework, and scrapes the high school calendar so both households stay aligned.
I’ve treated my own life as a controlled test environment for the entire applied AI consumer landscape. Each one was a real attempt to solve a problem I actually had — feeding my teenager well, knowing when my son will run his fastest mile, making sure his school emails don't fall through the cracks of two households.
What I learned, in eighty different shapes, is that the failure mode is structural. Every project — regardless of vertical, model, or framework — hits the same walls.
Siloed, unstructured data. Growing context windows that forget more than they remember between sessions. Tools that assume I'm a developer with a local terminal when Claude Code itself can't run the command. Zero infrastructure for multi-modal, multi-persona workflows — videos, images, transcripts, family members, taste, time, mood. Personal life breaks every assumption that enterprise AI infrastructure was built on – structured context, APIs, explicit workflows, measurable outcomes. Personal life is mobile-first, siloed, API-less, profile-switching, accountable to no one. Build for one and you cannot serve the other.
Statistically, I am not alone. The same friction is keeping the majority of vibe-coded personal apps stuck at 80% completion rates.
What I documented across this project collective is what an investor would call competitive intelligence and what an operator would call a spec. The connective layer doesn't exist. The infrastructure that would let an agent travel with you across devices, sessions, apps, and the eventual robots arriving in our homes — none of it has been built. Every consumer AI company I evaluated was building a feature where infrastructure should be.
That is the problem worth committing the next ten to twenty years to.
Our new odd kin, embodied AI
I’m shaping a new thought experiment – project Oddkin.
Oddkin is the personal AI that doesn't just think — it does. It executes the tasks you would actually outsource if you could. The multi-step, multi-app, multi-context errands of daily life: the closet, the meal plan, the school calendar, the travel itinerary, the call to the insurance company, the gift for the friend whose taste you almost remember. It is the operating system for the part of your life that doesn't fit on a Notion page.
Underneath the product is a thesis I have been writing toward for twenty years without knowing it. The promise of personal AI cannot be delivered on top of foundation models alone. The bottleneck is not the model — models are already astonishing. The bottleneck is the persistence layer. The place where your taste, your routines, your relationships, your context, your idiosyncrasies are owned by you, travel with you, and become legible to any agent or device you choose to introduce them to. Today that layer does not exist. Every consumer AI app is forced to either ask you for everything every time, or pretend it remembers and then quietly forget. That is the cold start problem. It is an infrastructure problem masquerading as a model problem, and it is the reason the majority of vibe-coded personal apps die at the finish line.
I'm calling the long arc the "Do It — Like Me" gap. The story has three acts.
Phase One. Virtual learning. Years of personal data accreted through engaging environments where AI learns your aesthetics, routines, tolerances, idiosyncrasies — through play. The training set is your life, captured passively through a medium you enjoy using.
Phase Two. Physical calibration. Cameras and sensors capturing how you actually perform tasks — the specific way you fold a towel, organize a pantry, set a table.
Phase Three. Robot deployment. A machine that arrives in your home knowing both what you want and how you do it.
—
At Parrot, I shipped the first consumer AR drone — putting an AI-enabled flying camera in the hands of people who didn't know they wanted one. At Sphero, I shipped one million consumer robots to homes and delivered 40,000 to schools. At Snap, I shipped a wearable camera that made content creation a first-person act and convinced 100,000 humans to put a computer on their face. At Whym, I built the data layer that lets a shopper's intent travel across channels. Each chapter was the same problem at a different layer of the stack — humans, software, hardware, data. The connective tissue between digital and physical AI is the next layer, and there are very few operators alive who have shipped at consumer scale across all four. I am one of them.
Robots are coming. They will arrive without any understanding of how you, specifically, do the things they were designed to help with. I’ve made a career out of bringing embodied technology to life – helping humans feel something — flight, friendship, identity, connection. The next one is about helping AI do it like we do.
This is work I was built for.