Uncapped with Jack AltmanTobi Lütke – Building Shopify and the Future of AI | Ep. 50
CHAPTERS
- 0:00 – 5:14
Why “beautiful problems” keep founders engaged for decades
Tobi explains that he stays energized by seeking hard, meaningful problems that force him to learn. He describes how motivation comes from directly experiencing a problem’s relevance—turning learning into a lifelong game.
- •Motivation comes from solving problems you can feel, not abstract learning
- •Karl Popper idea: a beautiful lifelong problem with “problem children” branching off
- •Success is simple: understand the cost (time, discomfort) and be willing to pay it
- •Curiosity about personal limits fuels sustained drive
- 5:14 – 7:21
Shedding CEO “barnacles” and returning to product craft
He describes getting trapped by external expectations of what a CEO should look like and how that made him miserable. His way out is focusing on creating joyful tools that make customers better—avoiding room-temperature mediocrity.
- •CEO aesthetics and performative duties can accumulate as “barnacles”
- •Guiding principle: build joyful products, not just acceptable ones
- •Cathy Sierra: “Don’t make better cameras, make better photographers”
- •Mediocre products feel like “room temperature”; great products require heat and care
- 7:21 – 10:13
Customer love helps, but meaning can come from craft anywhere
Tobi argues that you can build great software even without an inspiring customer mission, depending on how you approach the work. Still, Shopify benefits from exceptionally inspiring merchants, which attracts entrepreneurial employees.
- •Great work can come from focus and craftsmanship even on unglamorous projects
- •Shopify’s customers are unusually inspiring entrepreneurs
- •Employees often have founder experience or aspire to start companies
- •Freshness despite age: Shopify spans generations of employees
- 10:13 – 11:44
Originality as a competitive requirement (and why “failure” is discovery)
He lays out why being different is necessary to be meaningfully better—copying bounds outcomes. Experiments that don’t work are valuable because they clarify reality, so Shopify reframes failure as learning.
- •To be much better, you must be different—not slightly polished sameness
- •Convergence can validate first-principles understanding; divergence teaches faster
- •Null results are underrated; they refine your mental model
- •Shopify language shift: “successful discovery of something that didn’t work”
- 11:44 – 16:10
Conformity pressures in Silicon Valley and preserving eccentricity
Tobi and Jack discuss increasing herd mentality and professionalization in tech. Tobi argues distance from the Valley can reduce preloaded assumptions, and he criticizes cultural dynamics that suppress distinction and quirky individuality.
- •Environment installs priors; teams naturally converge on the safest path
- •Being outside the Valley can preserve independent thinking
- •Highlight reels distort how companies really operate internally
- •Companies should feel like an “island of misfit toys,” not enforced uniformity
- 16:10 – 19:31
Founder-led change: spending credibility to move fast on uncomfortable truths
Tobi explains how founders can accelerate change by using accumulated trust (“credibility tokens”). In AI adoption, he felt it would be unkind and unfair not to clearly state what was becoming true about leverage and impact.
- •Founder presence is an organizational advantage, not just an individual trait
- •Credibility accrues through onboarding stories; founders can “cash it in” for change
- •Organizations often prefer kind lies over hard truths; founders can break that
- •AI memo logic: if AI increases impact, it’s unfair not to tell people explicitly
- 19:31 – 23:28
Shopify’s AI adoption mechanics: tools, incentives, and token economics
They discuss how Shopify operationalized AI usage—encouraging tinkering, providing broad access, and tracking consumption without turning it into a gamed metric. Tobi notes token spend is significant but still worth it given the leverage.
- •Unlimited token policy to remove friction for experimentation
- •Leaderboards can create perverse incentives; Shopify moved away from them
- •Usage visibility (percentiles) exists partly due to budgeting and OpEx reality
- •Markets will find token clearing prices; near-term spend may rise, not fall
- 23:28 – 26:52
Small teams, faster cycles, and replacing six-week reviews with higher pace
Tobi argues AI enables smaller, more capable teams because everyone can be “7/10” across many skills with assistance. He emphasizes pace as a leadership responsibility and critiques quarterly/H1/H2 planning as a sign of dangerous slowdown.
- •AI reduces need for specialized headcount; small teams (3–5) become stronger
- •Customer insight can be routed and summarized automatically to teams
- •Parkinson’s Law: work expands to time allotted; leaders must compress windows
- •Six-week review cycle set a pace floor—now may be too slow in an AI era
- 26:52 – 31:05
What AI means for small businesses: more ambition, less friction, more formation
Tobi rejects the “permanent underclass” narrative based on Shopify’s merchant data. He argues AI reduces early hurdles to entrepreneurship, increasing business formation and employment—especially as setup friction approaches zero.
- •Shopify can’t reproduce doomer narratives in merchant reality
- •Merchants experience AI as finally making computers usable via conversation
- •Lowering setup hurdles reliably increases successful business creation
- •Small businesses employ the majority of workers in many economies
- 31:05 – 35:15
From Turing test to “prompt: build me a business”
They explore a practical benchmark for AI: autonomously creating and operating a real business that earns meaningful revenue. Tobi describes Shopify’s ambition to become the “vessel” that lets AI handle everything besides the core product idea.
- •A better test than Turing: can AI build a million-dollar business end-to-end?
- •Shopify already supports sourcing/manufacturing options and reseller models
- •Product thesis: AI should do “everything else” so entrepreneurs can focus on product
- •Digital products and on-demand manufacturing broaden what’s possible
- 35:15 – 35:40
AI, robotics, and the path to abundance (not joblessness)
The conversation shifts to AI’s impact on the physical world—manufacturing, robotics, housing, and standard of living. Tobi argues doom narratives underestimate human demand creation and the compounding tailwinds from automation and tools.
- •Physical-world impact (housing, transport, food, healthcare) is the real lever
- •Additive manufacturing and robotics make production increasingly tractable
- •Doomer job-loss framing ignores how humans constantly invent new work
- •AI as “Jarvis” enables ambitious tinkering and real-world projects
- 35:40 – 41:11
Why software progress looked invisible—and why the browser is a modern wonder
Tobi claims humanity didn’t stop building infrastructure; it shifted to digital systems. He explains why web browsers are extraordinarily complex and reliable, and why society undervalues digital infrastructure compared to physical megaprojects.
- •Digital infrastructure (Linux, browsers, the web) rivals or exceeds physical wonders
- •Browsers run untrusted code safely while reconfiguring computers on the fly
- •Reliability and openness of the web enabled massive economic creation (e.g., Shopify)
- •We’re near the end of the “opening chapters” of digital infrastructure buildout
- 41:11 – 48:13
Predicting AI trajectories: learn by doing, not just talking to labs
Tobi describes forecasting as collecting many data points and fitting trajectories. He argues that being embedded in real usage—watching how people apply new releases—often beats having proximity to research labs for understanding what’s next.
- •Forecasting method: gather data points, infer curves, connect trajectories
- •AI memo aimed to give Shopify a head start by acting on what was becoming true
- •Customers reveal problems; product teams must invent solutions (not just implement requests)
- •Using tools in the wild and observing patterns yields better future visibility
- 48:13 – 52:23
Talent in the AI era: interns as teachers, and why “good people are good”
Tobi explains how Shopify re-energizes its talent pipeline with a large intern program and learns from AI-native younger workers. He ultimately concludes the biggest determinant remains deep problem understanding and strong judgment, not tool novelty.
- •Keeping a company “young” requires deliberate renewal mechanisms
- •Intern programs help inject AI-native behaviors and reverse-mentorship
- •Programming is steering and problem comprehension—not just typing code
- •Adoption speed varies, but strong talent tends to stay strong after the shift
- 52:23 – 58:01
Public markets, legitimacy, and reading as a competitive ritual
Tobi shares why he values being public: broader participation, legitimacy with larger customers, and attracting certain talent. He closes with reading habits, book recommendations, and rituals that protect attention from internet-driven fragmentation.
- •Going public lets more people “vote” for a world via ownership and shared upside
- •Public-company discipline can be a feature (diligence, responsibility), not a bug
- •IPO helped Shopify’s legitimacy with larger customers and some hires
- •Reading: prefers short, distilled books; recommends dedicated rituals and a Kindle
