Fiverr CEO & Founder, Micha Kaufman: "If You’re Not Adapting to AI, F* You. You’re Done!"

Summary

• Micha Kaufman’s wake-up call to Fiverr’s own staff — the email and ensuing discussion — is that every employee should aim to automate 100% of their job, because doing so frees them for the non-automatable work (taste, strategy, nonlinear thinking). And the employer is not responsible for making you better on the way there: the last 10-15 years’ idea that your workplace must upskill you is “an aberration.” “If you’re not going to take the time to make yourself valuable… you’re doomed. I’m not going to help you” — he’ll help only those helping themselves.
• “Time to clone” is the metric that changed everything: ~1 year 25 years ago, 6-9 months when Fiverr launched 15 years ago, “10 days, whatever” now — “for simpler stuff it’s nothing.” Fiverr’s 9-month head start was precisely when its brand and community got seeded; when the clone window collapses to near zero, “you leave more to luck,” and incremental products get incremental results.
• Free technology confers no moat. OpenAI’s November 2022 go-to-market was “as democratized as possible… you can’t be freer than free” — it elevated the standard but “picked up everyone to the starting line.” So venture judgment reverts to the oldest rule: people. His best-ever investment was a seed company he’d written off (“so many reasons why this company needs to die”) that went public with multiples in the 50s.
• Maybe “999 companies out of a thousand” won’t be around in a year or two. Investors may trigger the cleanup by pulling back, while two further forces — regulation and “expensive tech” (AI so costly only large or well-backed players can afford it) — sort the market. Foundation players might be “reduced to the equivalent of paying for GPU cycles… it’s a cloud play essentially” — though the dot-com also gave birth to Amazon.
• “Copyright is dead. It’s a notion from 1710 and it’s dead overnight” — AI jumped over attribution entirely, and if creation gets eaten and churned without recognition or the chance to commercialize it, the motivation to create and share “is going to start slowing down and then decreasing.” He hedges it (“I don’t know if this would be the case, but I’m raising a flag”) but the kicker is categorical: “Harry, I have news for you. You are working for AI and so do I.”
• Governments may take control of frontier AI: what labs are doing is “like privatizing the Manhattan Project,” and if AI yields the ultimate weapon or cures the worst disease, “do you see government leaving this in private hands? No way, dude.” We’re “pre-expensive AI, pre-AGI, pre-superintelligence — probably not that far” — which is why he calls five-year business predictions “laughable.”
• Operationally: velocity over speed (“velocity is speed plus direction”), reduce the cost of failure (one person failing in 3 days beats 10 engineers failing in 3 months), and Fiverr will probably have more engineers in five years, not fewer. The most disrupted function isn’t coding — it’s marketing, where entry-level juniors “have been replaced the fastest, definitely faster than developers.” Every new hire must be an AI native: “I’m not going to teach you.”

Deep dive

1. The 250-person brain dump: automate 100% of your own job

• The “AI is coming for your jobs” post had been “boiling for a while.” Kaufman asked his chief of staff for a room holding 50 and offered to sit for three hours with his laptop for anyone wanting to talk — about 250 people crammed in, and the planned conversation became a brain dump.
• The core instruction: “my expectation of you is that each and every one of you… is going to replace 100% of what you do with automation.” The obvious objection — then you don’t need me — gets inverted: if you can automate 100%, you have 100% of your time free for what can’t be automated — nonlinear thinking, working with him on strategy. His larger frame: “maybe in a weird way, what’s happening with AI is forcing us to rediscover our humanity.”
• The blunt half: the past 10-15 years’ notion that your employer must make you better is “an aberration.” “Whose responsibility is it to make you a better spouse… a better dad? Why do you think it’s my responsibility to make you better as professionals?” He says, “I’m not going to help you.” His vow: “I will help those who want to go through that process… But if not, you’re done. You’re done. You’re not done at Fiverr. You’re done. Period.”
• Harry relays a bleak read from someone he identifies as likely Jason Lemkin: “No one wants to work today. Young people don’t want to work like they used to.”
• Kaufman refuses the frame — “the way people judge the present” does not have too many implications for the future — because “we’re in a very unstable world.” But on the individual level he’s merciless: “If you don’t want to work, the exit from the building is on ground floor… you’re either going to be poor or a burden on society. Grow up.”

2. Time to clone collapsed from a year to ten days

• His feed shows “40 new startups every day. There is no need for so many.” The metric he’s tracked since his first company: time to clone — ~1 year 25 years ago, 6-9 months when Fiverr launched 15 years ago, “10 days, whatever” now, “for simpler stuff it’s nothing.”
• His Everest analogy for what AI actually changed: builders now “get a ride to 300 feet from the peak,” but on summiting you discover you’re at ground zero on distribution — who will know about what you built? “My sister builds apps. Great… But what’s the true value? It’s very small to none.” And: “when you do incremental stuff, you get the incremental results.”
• Harry’s pushback — isn’t Fiverr’s value its brand, distribution, and supply-side trust, not the front end, making time-to-clone less important? Kaufman: “You’re right. But here’s the but.” At nine months old Fiverr had ~30 clones, later thousands; against the camp that wanted to sue them all, his camp chose speed — “try running by looking back… you are going to slow down.” The but: those nine months were when the brand and community got seeded. “If the time to clone is 10 seconds, you don’t have that time… so you leave more to luck.”

3. Free tech lifts everyone to the same starting line — so bet on people

• “Technology by itself doesn’t give advantage to anyone when it’s 100% democratized.” OpenAI’s November 2022 launch was the extreme case — “you can’t be freer than free.” Everybody felt smarter, “but so does everybody else… it is elevating the standard but it is just picking up everyone to the starting line.”
• His answer to Harry’s investor dilemma (20 companies in every category he invests in, not two or three): revert to “the oldest way of making judgments” — the people. Missionaries, not opportunists; first-principles thinkers; adaptable — “technology will evolve, ideas will change, companies will pivot.”
• The proof as told: a seed-stage company he wrote off — tough market, internal issues between the founders, “so many reasons why this company needs to die” — went public and became “probably the best investment I’ve ever made,” with multiples “like in the 50.” He cites an Etsy investor’s logic: “this company should have died at least seven times and it didn’t — this speaks for the strength of the team.”

4. Maybe 999 in 1,000 won’t last — regulation and “expensive tech” sort the market

• Will there be an AI bust? “Absolutely… 999 companies out of a thousand maybe” won’t be around for a year, maybe two. It’s going to be ruthless. But the dot-com “also gave birth to Amazon” — companies with true mission and patience survive.
• The cleanup trigger is investors: “at some point they’ll say, f* it, I’m out… so that’s going to dry up.” Every AI company has two fundamental problems — getting customers and retaining them — and “the overwhelming majority fail, rightfully so.”
• Two transformations haven’t arrived: regulation and expensive tech — “technology that is freaking expensive so that very few can actually allow themselves to have it,” creating differentiators for the large or well-backed. Foundation players (the Googles, the likely Anthropics, the OpenAIs) “might be reduced to the equivalent of paying for GPU cycles… it’s a cloud play essentially.”
• On Harry’s bank that laughed AI out of the room over compliance: “So they shouldn’t use it… which is going to be another reason why banks are going to be left behind.” Compliance will sort itself out — hosted on-location, whatever — “nothing is 100% safe. I don’t think these are the interesting problems.”

5. Copyright is dead — and “you are working for AI”

• The interesting problems: “are we okay with the fact that truth is dead?” — not knowing what’s real, what’s human, when we’re manipulated — and above all whether we accept losing our place “at the center of the universe.”
• His history of the engine at risk: the printing press bent humanity’s linear arc, and the 1710 UK act gave creators a moral right (attribution) and a commercial right — “this is how we advance,” compounding for 500 years. AI “jumped over the hoop of attributing anything to anyone… essentially copyright is dead. It’s a notion from 1710 and it’s dead overnight. Everybody calls it fair use. I’m fine, whatever — I’m not a regulator.”
• The flag he’s raising — hedged as a theory, and explicitly not as Fiverr’s CEO but while thinking about his kids and the future: if work gets “eaten and churned out without any recognition,” the motivation to create and share “is going to start slowing down and then decreasing.” “I don’t know if this would be the case, but I’m raising a flag.”
• The line that lands hardest: “Harry, I have news for you. You are working for AI, and so do I. Because we’re producing content and that content is going to be eaten by a machine and used to produce new stuff… we’re at the risk of losing our place in the center when we start working for something else.”

6. Governments may take control of frontier AI

• What foundation labs are doing “is like privatizing the Manhattan Project… imagine they do build the atomic bomb, and now they’re sitting and saying, maybe we’ll do it nonprofit, or we’ll license the atomic bomb — do you see a government allowing this?” If AI cures the worst disease, or is the ultimate weapon, or solves infinite food and energy: “No way, dude.” He speculates about nationalization; some functions might be prohibited and reserved for government branches — “I don’t have answers… these are equations with too many variables.”
• Harry’s counter: with likely DeepSeek and 12 Chinese open-source models, isn’t model knowledge commoditized and China-vs-US moot? Kaufman: “No, I don’t think so… we’re pre-expensive AI. We’re pre-AGI. We’re pre-superintelligence. We’re probably not that far, but we’re not there yet.” Companies “are allowed to play” only until governments get uncomfortable with that level of power.
• This is why he laughs at five-year questions. Being a CEO now is “like asking a captain of a ship in the middle of a storm… it’s wet, it’s dark, and you can’t see a mile ahead.” His father’s semiconductor world had Moore’s law — “you can extrapolate and say I know exactly what’s going to happen in 10 years” — and that era is gone. On Harry’s worry about value concentrating in a handful of multi-ten-trillion companies: “is it really different than AWS and Azure?” — two-and-a-half cloud providers have taken ~80% of that market for many years.

7. Velocity, not speed — and probably more engineers, not fewer

• Will Fiverr have more or fewer engineers in five years? “Probably more.” Same democratization logic: if you can squeeze 3x out of each person, “so does the next company” — square one again — and in an uncertain market “one of the biggest strengths of a company is speed.”
• But: “I never use the term speed. I always use the term velocity. Velocity is speed plus direction.” On HubSpot generating more code than it can ship: “it means you didn’t solve your infrastructure… your priority is incorrect. Solve your infrastructure first, then start building like crazy. Plus, you’re probably building stupid things — stop doing that as well.”
• The real payoff of AI-driven output is reducing the cost of failure: Amazon runs maybe 5,000 concurrent tests and most fail — who cares. “If building something took 10 engineers 3 months and you failed, that’s unpleasant. If it took one person 3 days and it failed, great.” And on leverage: does AI make 1x developers 10x? “Probably an easy no.” Does it make 10x developers super-super-superhuman? “An easy yes.”

8. Marketing is the most disrupted function — and juniors went first

• At Fiverr the function most impacted so far is customer support, but “the area going through the most fundamental disruption is actually marketing.” Coding copilots mostly automate what developers hated anyway — copy-pasting from open source, reading repo notes, “following blue links… it’s robotic.”
• Marketing has “no AI at the level that we have in programming,” and you’re still marketing to humans — so the past year brought “an awakening out of this illusion that this could be fully automated.” Yet “the entry-level juniors have been replaced the fastest, definitely faster than developers” — because with code you still must open it and understand it; not so in marketing. Harry agrees from the trenches: the bank’s “we’re thrilled to announce” social manager is gone, but “I would pay a million dollars for someone to do social like me.”
• The hiring bar follows: before adding anyone, maximize the existing team, and any new hire — legal, finance, support — “needs to be AI native… because again, I’m not going to teach you.” The cost he’d always cut if he could: marketing — but only “because you can grow without it” when organic growth is strong.

9. Quickfire: skip university, the OnlyFans that wasn’t, meaning over happiness

• Changed mind on AI: OpenAI’s nonprofit origin made him think this could be “a global endeavor that would give more power to more people” — reality reversed that (“you were open and now you’re closed”), and for a technology this fundamental, “I would probably rather have open than closed.”
• Advice to a graduate arrives a step earlier: don’t go to university — “in 90-plus cases of modern professions, you can study on your own in a fraction of the time.” The missed strategic opportunity: Fiverr could have created OnlyFans “many, many years before OnlyFans” — it was bubbling organically via code words on the platform — and founders and board declined on values.
• Is Fiverr his last company? “I don’t know. Right now, yes.” He runs an annual accounting — energy, contribution, curiosity, am I still the right person — with one hard rule: “if I wake up five days in a row saying ‘oh no,’ then I’m done. And it never happened.” On happiness: “it’s about meaning… you are going to have a mixture of misery and happiness by definition.” How he wants to be remembered: as a giver — “I hate takers… the world doesn’t owe you anything” — and “I’m probably an artist. I just express my art in the things I build.”