Chieh Huang

The through-line

Across a decade of talking about Boxed and now Pelgo, Chieh Huang keeps returning to the same question: what happens to the people on the receiving end of a technology wave, and who is responsible for them. At Boxed that showed up as an unwritten rule that fulfillment center team members were treated as well as anyone in the company [1]. At Pelgo it shows up as a business built on the premise that "there's so many companies that are well-funded that are producing technology that will eventually take people's jobs. But there's very few companies saying, okay, how do we how do we train these folks" [2]. He describes taking that on as trying to "fight fire with fire" [2].

The second through-line is a repeated bet on a platform shift while it still sounds absurd. He built one of the first mobile social game studios when the iPhone launched, then wagered that people would transact their daily lives on phones, at a time when customers wrote in to say they would never type a credit card number into a handset [1][8]. He is making the same kind of call on AI now, and the temperament has not changed: long-run optimism about job creation, blunt honesty that "in the short run, there's going to be a lot of pain" [3].

On purpose as the thing that isn't written on the wall

Huang separates goals, mission and purpose and argues most companies conflate them. Goals, especially as a public company, are "to make or meet or beat uh quarterly earnings kind of expectations"; the mission is something you want to do for the customer; purpose is different because it is "often times is actually what's not written on the wall. It's how everyone behaves when just it's you and one other person in a room making a big decision and there's no guide on how to make it" [1]. He is unusually careful not to claim he engineered it. Unlike founders who say the purpose was set on day zero, "That was through a very hard road of discovery that we found purpose in ourselves in our company" [1].

The proof he offers is a negative test. Treating frontline team members well was never a stated core value and was never posted on a wall, but if someone had walked into a room and proposed that "we can save a lot of money by cutting all the benefits at the fulfillment centers," they "would get laughed out of the room, even though it's never written anywhere as a as a cardinal rule" [1]. He frames the practice as doing right by the people packing boxes, unloading trucks and writing the handwritten notes that went into every order, and insists it was both the right thing and good business [1][8]. Over time it drew publicity, so new hires arrived knowing what they were joining and then found "it's not just a glossy brochure" [1]. His broader claim is that purpose is not optional: "it doesn't matter how rudderless a company is, they will have a purpose. It might not be a purpose they like" [1]. The same people-first orientation is what he has discussed elsewhere as a "we" over "me" culture, including his practice of personally interviewing every candidate before hire [10][11].

On betting on the smartphone, and now on AI

Consumer was what he knew. Games were the first wave on the iPhone and Android, and the follow-on thesis was that people would run their daily lives on those devices, which "sounds completely innocuous today, but rewinding back to 2012, 2013" produced customer service emails asking "Wait, you you don't have a website? You want me to put my credit card into my mobile phone? Capital ha, good luck." [1]. The thesis held and he rode it for ten years [1].

He sees the next consumer wave in the same place Boxed almost got to and could not finish: "the automation of your life for the more mundane tasks" [1]. Boxed used ordering data to predict replenishment, which worked well for businesses whose pantry stock gets consumed anyway, and badly for consumers, because unwanted shaving cream on a credit card is a problem [1]. What has changed is both capability and receptivity. Seven or eight years ago, telling an office manager you were mixing machine learning with their habits drew blank stares; now AI is "firmly in the mainstream" and the reaction is that it makes life easier [1]. His advice is direct: if that service does not exist, go build it [1]. He extends the same logic to work, arguing that AI is less likely to take jobs outright than to absorb "all the mundane things that they're just doing over and over and over" [1].

On AI, job destruction and job creation

Huang expects certain workflows to reach full automation and is unsentimental about the consequence: "there's going to be huge job destruction across the world. But I am also a firm believer that there's going to be huge job creation as well" [2]. His evidence for the optimistic half is historical and comes from his Pelgo co-founder Frank D'Souza, who co-founded Cognizant: look at the services tab on Cognizant's site today and rewind to 1994, and "virtually none of those jobs or none of those services existed" [3]. Cognizant began by helping companies place work in India because of the internet, and then grew because the internet spawned job categories nobody had imagined, content moderation for social media being the example he uses, where clients needed thousands of people yesterday for a function that did not exist when the firm was founded [2]. He expects AI to follow a similar trajectory, "maybe just way more compact than than 20 years" [2].

The near term is where he agrees with the pessimists. He points at his own early investment banking years and notes that essentially every task he was given then could now be done by a model, which is why the moment is so hard for recent graduates carrying student debt [3]. A survey of 5,000 unemployed Americans across geographies and demographics that he cites found only 15% would reapply to a four-year university knowing what they now know about AI, only 6% tried to upskill or reskill while unemployed, and more than half said AI is the new norm [3]. He also flags a strange asymmetry in the market: "you have entire industries cratering in real time and you have entire industries just on fire," with entry-level software engineering hard to land while forward deployed engineering roles go begging, and companies cutting staff finding they are not reputationally intact enough to attract the talent they suddenly need [3].

On career transitions becoming the norm

The founding insight behind Pelgo is a change in the shape of a working life. Forty or fifty years ago it was abnormal to change companies and unthinkable to change careers; now "transitions are going to be the norm," and someone with "one line on my LinkedIn" for an entire career will be "a true unicorn in the coming years" [2]. He puts the same point on a longer arc: over the last fifty years it became less and less common to stay at one company, and over the next fifty "it'll be very rare for someone to be in the same industry for 50 years," which makes reskilling, upskilling and being nimble critical [3]. The company's stated purpose is to lessen the blow of those transitions and to pipe people losing jobs into AI-first and AI-enabled roles [2].

He is deliberate about what realistic retraining looks like. If a fulfillment center becomes "lights out automated, meaning that you don't have lights because there's no humans in it," it would be absurd to propose retraining that worker as a medical doctor [2]. Data centers, by contrast, still employ hundreds of people on site and have genuinely distinctive requirements around temperature, security and electricity, which makes them a plausible destination for someone coming out of a blue-collar environment [2]. He also rejects the fatalistic alternative, saying he does not fundamentally believe the future is everyone "just sitting on a couch collecting UBI checks" [2]. Early demand signals surprised him: the most forward-thinking interest is coming from tech companies themselves [2].

On the missing first rung

The roles being cut first, in his early reading of the market, are disproportionately junior [2]. That worries him on two counts. Mentorship disappears when the mentors are gone, but the deeper problem is structural: "if that first rung isn't there to start your career, I think that poses a really big question and problem for society," and he asks openly "where's that first rung in the future?" [2]. He hears the same anxiety from both directions, saying it is the number one question from college students and from executives watching their own children study [3].

On the skills worth building

His answer is not more coding. Three or four years ago the smart move for any major was to pick up programming languages on the side; that advice has already inverted [3]. He reports a shift among senior leaders toward pushing children into the humanities, decision-making, and industries requiring judgment calls and constant human interaction [3]. The core competency he names is orientation toward the machine: being proficient with AI matters, but what matters more is operating at a level where you make decisions and direct the system rather than program it. "instead of fighting A.I. being able to direct it, I think it's a number one thing that folks have to learn how to do" [3]. The practical version, drawn from talking with students, is "being able to ask the good questions so that you get the right response" [3]. He also expects entirely unimaginable categories to appear, saying "there's going to be some gnarly jobs that we can't even imagine today that will be commonplace in in just a few short years," and points to forward deployed engineering, on-site work discovering a company's peculiarities and building quick solutions, as a live example of a role created by the same shift that removed other engineering jobs [2][3].

He is curious rather than settled about what AI does to software distribution. If every coffee shop can vibe code its own small modules, does the shop realize twenty other independent coffee shops need the same thing, and if so, what is the point of distribution, given "they're not going to have like a a go-to-market kind of muscle at all to to to flex" [2].

On the garage years and the shape of consumer traction

The garage was real, and less romantic than the retelling suggests. He was thirty, kids crossed the street rather than walk past, and the trucks eventually forced them to replace the garage door with mesh screen [1]. The version in his head had been orders arriving in week one; the reality was months with days where nobody ordered anything, and an early employee who had left a well-paid job asking how it was possible that out of 300 million Americans who all use these products, not one had bought [1]. "there's a lot of self-doubt that creeps in when again when you're sitting in a quiet garage just staring out into the ether" [1].

The break was a single television segment. An influencer regularly on the Today Show asked to plug them when the company was doing three to five orders a day; they assumed it was spam, dropped product at the studio by 5 p.m., watched live, and orders began flowing within the hour, with friends drafted to work nights at the house by the end of that week [1]. That is what makes consumer both attractive and volatile: a first-time founder can solve a problem they and their neighbors share, and "the first time you see someone in the wild using your product, it's a really cool experience" [1]. He remembers slamming on the brakes when he spotted a Boxed box outside a house in a nearby neighborhood [1]. The flip side is becoming customer service for millions of Americans and hearing about someone's last order while watching a game [1]. His designers ran limited edition printed box tops, and one co-founder put a caricature of Huang on the plate as a joke, went ahead over his objection, and that plate ended up the only one printed, on the largest and most popular box size, so his face travelled New York streets in variable condition [1]. The company competed with far larger retailers by carrying fewer items at competitive prices with a handwritten note in every box, shipped tens of millions of boxes, and reached a valuation over $600M [8]; he has separately discussed the SPAC route to going public [6] and how Boxed leaned on technology to absorb pandemic demand for bulk direct delivery [7].

On operational reality

Behind the consumer story is a low-margin, high-volume unit economics problem, and he has walked through the operational engineering required to make it work, his early adoption of autonomous guided vehicles in fulfillment centers, and the move from pure automation toward human-machine collaboration [5]. None of the founding team knew supply chain; an early captured email celebrates getting the UPS thermal printer working, which he uses to show the base of knowledge they started from [4]. His fix was to hire slightly ahead of need. Their VP of transportation, then a consultant, told them at his interview that they did not need him yet, and Huang's response was "we're going to need you next year though" [4]. The principle generalizes: some problems have never been solved and you have to invent your way through, while "some of the things like shipping transportation that's all been done before so there's no need for us to learn the hard way" [4].

From the Six Flags board seat he makes a related point about where money quietly goes. Peeling paint on a coaster is not evidence of cost-cutting on safety: "If there's a dollar and an hour Six Flags as well as any other company is spending it on safety" [2]. He also notes how thoroughly Fright Fest reshaped the theme park calendar, with Halloween rather than July or August as the industry's defining moment [2].

On making the leap and surviving it

His legal training was preparation more than detour. Reviewing deal documents after already brutal corporate hours, he researched the businesses on his own time and concluded "there wasn't anything special or magical about these folks," they got lucky, worked hard and made the right decisions [4]. What the law left him with was the ability to tell when something can go badly wrong and requires professional help versus when it does not matter, and he sums it up as "understanding the rules of the game um uh really helped me play the game" [4]. He started at the firm hours after Lehman collapsed a block away, watching thousands of people carry boxes out of a building across Broadway, which made quitting cold turkey harder [4].

The leap itself he describes through cliff diving in the Pacific: you climb an atoll, the descent is more dangerous than the jump, there are rocks below so you have to jump out rather than let yourself fall, and once you step back and go, "you don't look behind you." Walking out of the law firm felt identical, and afterwards "all my lifeboats are gone so i better just make sure i hit the water safely" [4]. He pairs this with two pieces of hard practicality. First, personal runway: founders obsess over company runway and neglect their own, and running out of money after 60 or 90 days "that's not a real run at your dream" [4]. Second, emotional management: "you have to treat it like a statistical curve and delete the outliers," because extreme highs and lows arrive on the same day and riding them is unhealthy for you and your team [4]. When motivation flags, his advice is to go back to the person who made the jump and trust their judgment, because something about that moment convinced them [4].

On motivating a team, and how that changes

For the first two or three years he ran an annual company meeting in a rented movie theater, let people play video games on the big screen, walked through the previous year's wins, and closed with photographs of the first twenty people at Google, Disney, Alibaba and Amazon [4]. Then he showed a photo he had taken secretly of everyone as they walked in that day, and told them that if they did this right, "one day there will be a movie theater of people just like you looking at a picture of you" [4]. He still gets goosebumps telling it, and the point was that a great job is always available while a shot like this one is not [4]. He is candid that the method does not scale: with hundreds of remote employees, nobody fits in the same picture and he has no magic bullet, only a commitment not to pretend things are easy when they are hard [4].

His own motivation changed shape too. It starts as making it financially or being your own boss, and then "it is no longer about what it is what it means for you personally anymore," it becomes responsibility for the people who came with you [4]. One regret sticks: an early employee wanted to film everything from day one in the garage onward, and Huang dismissed it as the dumbest idea he had ever heard, partly out of lawyer's caution about what would be on tape. He now thinks any journey from first day to outcome would have made a compelling story [4].

Takeaways

• Distinguish goals, mission and purpose: purpose is what governs behaviour when two people are alone in a room with a decision and no rulebook, and it is usually discovered over years rather than declared on day one [1].
• The strongest cultural rules are unwritten and enforced socially; at Boxed, proposing to cut fulfillment center benefits would have got you "laughed out of the room" without any core value saying so [1].
• Expect short-term job destruction and long-term job creation from AI; the Cognizant precedent is that almost none of today's service categories existed in 1994, and the AI version will compress that timeline well below 20 years [2][3].
• Career transitions are becoming the default, and a single-industry, single-line career will be "a true unicorn in the coming years" [2][3].
• Retraining has to be realistic about the starting point: a displaced fulfillment center worker moves plausibly into data center roles built around temperature, security and electrical needs, not into medicine [2].
• The scarce skill is directing AI rather than fighting or programming it, alongside judgment, human interaction and the ability to ask good questions [3].
• Only 6% of 5,000 surveyed unemployed Americans upskilled during unemployment, and just 15% would reapply to a four-year university knowing what they now know about AI [3].
• Fund your personal runway before you quit; running out of cash after 60 or 90 days is not a real attempt at the idea [4].
• Delete the outliers from your emotional record, since extreme highs and lows often land on the same day and riding them damages you and the team [4].
• Hire specialists slightly before you need them, and refuse to relearn solved problems like shipping and transportation the hard way [4].