A complete breakdown of the nine episodes of the NFX masterclass by James Currier: from founder psychology and magic words to the ladder of proof, pricing and 42 tactics. Quotes, timestamps, numbers.
This is an exhaustive summary of NFX's ninth speed season: nine separate episodes recorded by James Currier (co-founder of NFX, previously founded four venture software companies; the first company was a psychological testing company with 150 million registered users). Together the nine episodes form a single coherent course: speed as the startup's operating system.
Each episode is broken down in its own block of several sections. Everything stated in the source is present here directly: numbers, names, examples, methodology. The format quotes: where the author's wording matters, a verbatim quote is given with a timestamp in the format [mm:ss].
The document is deliberately heavy: it is meant to replace watching roughly four and a half hours of video. If a topic is only mentioned in passing in the original, it is still named here so that nothing is lost.
The first episode of the course opens programmatically. The speaker explains that he has been polishing this text for ten years, that it is the most-watched video among NFX founders, and that many companies show it to every new employee on their first day. For the first time it is being published externally.
The only criterion by which NFX judges founders is stated in one line: speed. Not market size, not the TAM slide, not eloquence, but whether the person feels like they are moving.
The central metaphor of the course: you are in a startup, you are in an X-Wing fighter, and coming at you are incumbents who are not going anywhere. The speaker stresses that this is not 1994, when there were only two incumbents on the whole internet.
Luke says two words that set the tone for the whole course: full throttle. And Wedge, his wingman, does not ask questions.
From this follows the arithmetic of speed: if you make two moves for every move your competitor makes, you can beat anyone who is bigger or smarter. Not because you are better, but because you are more frequent.
Speed does five things at once: the product gets better faster, the team has more fun, the press and investors take more interest, the best employees are drawn in, and the valuation rises. This is not altruistic advice, but the mechanics of resource accumulation.
Before giving tactics, the speaker takes the blame off the founder. He insists: a low speed bar is not your fault. It was set by two institutions you succeeded in, school and the corporation, and both are slow by nature.
The reason successful people find it harder to speed up is paradoxical: they have something to lose, and they feel they know how the world works.
Then the whole episode is built as a series of proofs that the bar can be reset and that it will not hurt. He introduces the notion of a "founder of speed" and moves on to the stories.
James calls Mike Cassidy the "father of speed" and his own Obi-Wan in the late nineties. The standard startup cycle of that time was 24 months: six months to research the idea, three for money, three to hire the core team, 13 for the product and three for launch. Cassidy compressed this into four months.
Hiring the core team followed the same principle: a ready contract was already on the table by the end of the interview.
The achievements of Cassidy with which the speaker reinforces the thesis: Direct Hit was built in three months and sold for 500 million dollars after 500 days; Xfire was built in three months and sold to MTV for roughly 110 million after 12 months.
The speaker's own story. In the gaming business, the hired game specialists claimed that a good game would take 12 or 14 months. The company had 35 days. Then a web developer who had never worked in gaming said he would do it in 21 days, asking for only one thing: no meetings.
After 21 days the game worked. Three weeks after that it had three million users. The key phrase: he did not know that it could not be done.
The second example, now about Dragons of Atlantis. The stakes were high: 600 thousand dollars remained, the company was nearly bankrupt. Experts estimated a serious game at 4 million dollars and 12 months. The team made it in five months and 450 thousand.
The scale: 45 million registered users, up to three million concurrent players spending two to eight hours a day. And above all, the release rhythm.
The speaker adds that in the accelerator weeks were spent persuading founders to ship releases daily, and when that happened, "the air opened up, acceleration arrived".
A software company should run 20 to 100 experiments a month. At Dragons of Atlantis the team ran 20 to 100 experiments a week. The speaker's wording: «if you can't do 20 to 100 a month, then you're not in the game» [12:00]. For hardware there is a caveat: it is harder, which is why «hardware is hard».
In the same block there are three quick examples: James Hong built Hot or Not in eight days, naming the company "Eight Days"; Jiff's companies built a product from scratch in 30 days and presented it, raising 7.5 million with 30 days of capital left; Meerkat and Clubhouse were 30-day projects.
The most emotional example of the episode: the speaker's son, who could not sail, decided to cross the Atlantic alone. The industry standard he broke was this.
The reaction of the professional community, which had spent 350 thousand, was negative, and the speaker explains this precisely by the critics' low speed bar, not by real danger.
There is one main barrier on the path to speed, and the speaker names it directly.
| # | Way | Wording |
|---|---|---|
| 1 | The product is not you | Put it out; it is just pixels and a feature |
| 2 | Don't chase correctness | Incremental development instead of perfection |
| 3 | A skateboard, not a convertible | Start with something that looks ridiculous |
| 4 | Launch behind the scenes | A different URL, a small group of people |
| 5 | Experiment | Not building a business, but a series of experiments |
| 6 | The studio model | Five ideas, one launch every 30 days |
| 7 | Learn | What did you learn today |
| 8 | Borrow mentality | Copy what already works |
| 9 | The "fuck it" mentality | Reach it not in the last 30 days |
| 10 | A chip on the shoulder and self-sabotage | Recognize it in yourself and in co-founders |
| 11 | Stop being a wimp | Make a decision, not half measures |
| 12 | Do it for people | If not for yourself, then for the team |
The VP of Product, hired to bring the product to perfection, comes in for particular criticism. The speaker says he does not hire a VP of Product at all, but keeps a trio: engineer, designer, CEO. And he gives a proportion in which the product is not the main thing.
The diagram founders return to for years. You cannot build an imaginary car in parts: first the wheel, then the chassis, then the body. You have to start with a skateboard that has nothing to do with the car, fall in love with it and make it excellent.
The fourth way: if your company is called NFX, don't launch the product on nfx.com, launch it on another domain. The point is to remove shame from the team.
The sixth way is revealed through the story of a company the speaker calls Yevo (also Yevvo in the transcript), brought over from Israel. 14 months of product, then another eight, all misses. On a small whiteboard, in 90 minutes, five ideas were born: a color for each, a name for each, one or two features each, and a launch every 30 days.
Each cycle ran longer than planned: 14 months, 8, then 60 days for Meerkat instead of 30. But it is still compression. Meerkat took off, Twitter shut it down, and the next product (blue) was assembled in 30 days: it was Houseparty. The company was sold, and the green, red and pink ones remained unrealized, but ready. Conclusion: think of yourself as a studio.
The seventh way: the founder's job is experiments and learning. His question to founders: what did you learn today. The eighth: borrow from others, don't invent. The Bebo story illustrates that even identical code gives different results.
That founder sold Bebo for 850 million dollars. The thesis about copying is radical: Google copied everything.
Then come the examples: Facebook copied Friendster; Airbnb became like VRBO, taking 10% instead of 300 dollars per listing; Uber copied Lyft within 30 days; Lyft copied Sidecar, which invented the model but was too slow and died.
The ninth way is the "fuck it" mentality. The speaker is honest that the best decisions of his career were made under pressure, when the money was running out and he was not sleeping. His advice: don't wait for the last 30 or 60 days of capital, start from that state.
The tenth way begins with psychology. Many found companies to prove something. The speaker recounts a conversation with venture capitalist George Zachary about a young Elon Musk.
From this follows a warning about self-sabotage. The link between speed and self-sabotage in founders the speaker calls almost direct.
The eleventh way: stop being a wimp and just make a decision. Half measures (keeping the old product and starting a new one in parallel) the speaker calls self-sabotage.
The twelfth way: do it for people. If you cannot speed up for yourself, speed up for the employees, partners and investors who are counting on you.
After fear comes the second big block, simplification. It opens with legal basics: don't innovate in the standard.
East Coast investors will come with warrants, milestones and contingencies, with a right of first participation. The speaker responds radically.
He tells how he was five minutes late to a lecture because he was dealing with a company where a bad person had joined the board, who through contract clauses was trying to take control from the founders. It is a vivid illustration of the price of complexity.
Partnerships and biz dev come in for particular criticism. The story of the first investor, Warren Adams, who wrote a check for 125 thousand dollars and asked for one promise: not to do partnerships.
Next there is a long list of simplifying decisions, each of which removes friction:
| Simplification | The speaker's reasoning |
|---|---|
| Keep a board of three people | Founders should control the board; his once had more than three, and it was chaos |
| Sometimes no PR | «Obscurity is bliss», fewer obligations to a story you have already voiced |
| Zero marketing budget | The product should grow on its own, find Reddit and organic channels |
| No budget process until 30 people | Just do it fast and cheap |
| Make less software | Sometimes renaming a button is enough, not writing code |
| No QA, usability tests, focus groups | Ship it and watch the metrics; people don't know what they want |
The story about the button is telling. The VP of Engineering asked: if we rename a button on the first page, can we skip building a whole section. The answer was yes, and a week of work turned into a language change.
A psychological self-testing company: 43 venture firms said no. He replied that women's magazines have tests at the checkout, and it works. He shipped the product, and traffic came. This is the same thesis that later unfolds in the episode about language: wording matters more than technology [36:00].
The principle: don't build something no one needs, and don't reinvent what has already been tried before you. Research compresses years into months.
A concrete tactic the speaker repeats to founders two or three times a week: find on LinkedIn a VP of Sales who worked at a competitor, and invite him to lunch with questions. Founders usually object that he won't tell them anything, and return from six such meetings feeling that life has been turned upside down.
Then the river metaphor. Staying where the going is "easy", and sleeping well, is the trap of slow growth.
An example of how one person aggregated resources faster than anyone on the Facebook platform. Within two weeks he bought Texas HoldEm Poker, an already working product, that is, again he borrowed rather than invented.
The speaker is honest about the quality of the hiring: they were not the best people, but those who could not find work elsewhere, yet they were available. In the moment, money, press, traffic, remarketing and network effects mattered more. The company at one point was valued at 11 billion dollars. Conclusion: attract nodes to your company's network as fast as possible.
Recruiting is offered as an art to be elevated: the offer is ready before the candidate leaves, and a maximum of three weeks to fill a position. The speaker rejects the typical six-month timeline.
The team's mental setup: you are here to kick ass and to have your ass kicked, by the market, investors, co-founders. Competitiveness is welcomed, it is a gladiator sport, not a career path.
An example is the founder of Twist Bioscience, Emily LeProust, who looks for founders she calls "sauvage", which means "savage", wild, fierce. It is not about anger, but about decisiveness and aggressiveness in decisions.
On team size, the speaker explains the psychology of the growing headcount: each new person adds communication load and human needs that hinder speed. The key thought about the quality of hiring.
The speaker recounts that he fired his first three VPs of Product, and the last one said the classic phrase to him.
His counterexample is the company Stoke (Seattle), which makes metal rockets for spaceflight and has turned speed into a religion. The math is simple: burning 80 thousand dollars a day.
Hence the holy trinity: speed, quality and cost go together, rather than contradicting each other. Faster means more experiments, faster attainment of quality, and on average, over the long term, higher quality.
An important caveat the speaker makes separately: speed is not about overtime and burnout, but about removing mental and operational barriers.
A rubric the speaker honestly calls borrowed. Three types of thinking determine who speeds a company up and who slows it down.
The hiring economics of the types: a Commando costs 150%, but delivers 10x; a Soldier costs 110% and delivers 3x; a Police costs 100% and delivers negative productivity. The speaker's rule is uncompromising.
A nod to Steve Jobs: never hire a B-player, because he brings in C-players. The same applies to speed: a Soldier brings in Police to look like a hero. And the answer to the frequent question about underperforming employees: they are fired in order to focus the team on speed.
The speaker recounts that in the early 2010s boards paid him about 1.5% of the company to help with growth. But with the CEO he never talked about growth.
A practical tool, the priority dashboard: a projector on the wall, a list of priorities that is changed in front of everyone, and each person can write on it.
A separate thread of the episode is getting information without building the product. The first technique: the street. The team would go out with a laptop and interview 8 to 10 people in half an hour, and after lunch they fixed whatever people did not understand.
The second technique: an email group. Gmail did this, where the speaker for 14 months belonged to a group of about 80 people before launch. Then come paper mocks (the Outdoorsy story, where nine out of ten said yes on paper, and the company grew to 2 billion), landing pages with a fake checkout, and explainer video.
The Zappos story closes the block: for the first two thousand orders the company fulfilled them by buying shoes in a store, boxing and shipping them, with no warehouse.
Bill Gates's tactic of announcing products he had not built, in order to scare investors away from competitors and preserve option value, is also mentioned separately.
The principle of the external deadline: people believe it more than they believe you. The speaker explains why Y-Combinator works.
The deeper reason for the accelerator's effectiveness: it gives everyone around you a clear excuse to go into focus.
Then: outsource everything, minimize processes (except trust and safety, payments), and make the metrics transparent to everyone, otherwise the team cannot act without you.
The finale of the episode is a list of the things that personally slowed people down for him. Remote work comes first.
| Brake | The speaker's wording |
|---|---|
| Remote work | For companies under 100 people, being in one room is essential |
| Acquisition talks | Flatters the ego, but takes time |
| An incomplete team | If hiring takes more than three weeks, you are doing it wrong |
| Wanting to be right | Most of your ideas are wrong |
| The wrong team | Always a problem |
| Lack of focus on one client | Two or three years with only one key client |
| Supporting two products | Happens constantly |
| Not changing the company name | Change it if you need to; good URLs are plentiful |
| Problems with the cap table | 16 angels in the first company demanded updates and favors |
| Too little capital | 12/18/24 months of runway, but also don't over-capitalize |
| Complex deal terms | A bad investor on the board |
| A slow "no" on BD | A constant source of delay |
| The wrong team / product | A general brake, encountered everywhere |
In his first company the speaker took on 16 angel investors. Each wanted updates, helped his cousin, asked for a reference check on behalf of another founder. Conclusion: too many people and too much complexity on the cap table.
The second episode begins with a thesis that turns the usual frame upside down. The biggest waste in the startup ecosystem is not bad founders, but good founders on bad ideas.
Evidence from his own FAST program at Stanford GSB: 75 teams applied for checks of 100 to 250 thousand dollars. Of the 75 ideas, 68 were bad.
The problem is that students work for months on bad ideas before applying. The goal of the episode is to move founders along the spectrum from bad soil to fertile soil.
The map of the idea landscape has three zones. On the right, fertile soil, where a team of two or three people with a good pitch will sell the company for at least 10 million in two years. On the left, mediocre soil, which will consume an entire life. In the middle, uncertainty.
An example from the middle: Airbnb. The speaker honestly admits that he did not know whether people would sleep on an air mattress in a stranger's room, and it turned out that they would, contrary to his expectation. His job is to move the founder to the right, the founder's job is to execute.
The first law of big business: companies that change the world tend to have network effects at their core. NFX statistics across 30 years of technology.
You can get a billion from enterprise SaaS, but it is hard to get ten. That is precisely why the firm is called NFX. Then comes the rule of the three S's.
A completely new idea generates enthusiasm, but no one buys it, because it is not needed. A full pixel-for-pixel clone leads to legal problems and condemnation. The right zone is between them: two or three obvious changes to something that already works.
| Original | Copy with changes | What was changed |
|---|---|---|
| Prosper | Lending Club | Linked the loans, talked to the FTC in advance |
| InfoSeek | Better people, an algorithm about 10% better | |
| VRBO | Airbnb | A couple of changes to a very similar product |
| Friendster | The same blue, the same five features, a better team |
The Prosper and Lending Club story is telling: two changes led to a valuation of about 7 billion, while Prosper was shut down by the FTC, then came back and copied Lending Club's approach, and now they are comparable in size.
A distinction the speaker calls key, and which he himself did not understand in his first four companies: market risk and execution risk. The first time is changed by x.com, which became PayPal, and afterwards he says he will never again take market risk, because he knows that people drive cars, pay to launch cargo into space and install solar panels.
The first time, founders take both risks, but they have no execution experience, so the probability is low. The second time, founders take only execution risk, because they know they can execute.
The second dimension of risk is the story of Rick Frisbie of Battery Ventures. The speaker persuaded him to invest internationally, and within a year they lost 100% of the money. Frisbie formulated the conclusion.
The lesson cost 8 million. The conclusion for a founder: list all the risks of your idea and compare them with the alternatives. An example of the high price of risk is medical software with a deep track record of losses over 15 years.
The generally accepted advice to avoid the red ocean the speaker turns around. By his observation, those who go into the red ocean end up doing better, because the market is already proven.
He cites a company that failed at two blue oceans and was heading for a third, and advises it to find a business it admires and build something very similar, because the market, price and go-to-market are already visible.
Timing is determined by shifts, and the speaker distinguishes three types. The most important is a rapid technology platform shift, not an individual technology.
An example: generative text and graphic AI, whose price fell by a factor of 100 over eight months, creating a Cambrian explosion. Cultural shifts, by contrast, take decades, so everyone sees them and there is no advantage. Regulation is the third: the right moment is to be after the law changes, not before it.
Separately, a warning about "technology in search of a market": an invention out of a laboratory rarely becomes a business, except when it is a platform with multiple attempts, which is why in biotech NFX invests only in platforms.
A big TAM today is not required. You need a fast-moving market, even if it is small, like home stays instead of hotels. And you should not try to digitize the existing transactions of a B2B marketplace, but create new transactions in a sub-segment.
One of the densest concepts of the episode. The company Wonderhill made games, and the first ones did not gather enough users, frequency of use and revenue per player; they remained inside a certain boundary layer that prevented the business from materializing. As a benchmark they took Farmville: about four cents a day from 100 million people. Wonderhill had 30 million on one of its games, and it was still not enough.
The practical conclusion: calculate not only the size of the market, but the product of ASP times the number of users, the area under the curve. An example of a ceiling: 3D modeling for game developers, where there are perhaps 50 thousand buyers, and all such companies over 10 years were sold for 50 to 100 million.
A benchmark for games: Mafia Wars, which drove Zynga, had about 80 million users and made about 2.70 dollars a day per user. And the joke that the working game themes are visible in the Lego department: pirates, Atlantis, city builder.
When do you change course? The speaker's answer: "passionate beliefs loosely held", passionate convictions that are easily let go when data appears. Of his four companies, three pivoted, and two changed their name after product-market fit.
The skateboard appears again: don't fall in love with the sports car, build the skateboard and treat it with all your soul. Facebook started as college photo sharing, Twitter as "what you ate for lunch in 140 characters", Airbnb as an air mattress.
Mental setups: commando versus soldier and police (VC firms "sniff" for this type, because it is lethal); play to win, not to avoid losing (the speaker himself admits he played not to lose, which is why none of his four companies became a billion-dollar one); an ambitious vision as a magnet for the best people and capital. The story of the team in Toronto, where they were asked to move the headquarters there, and where the founders lectured four times a week at a Holiday Inn.
The same message as in the Speed Talk: Facebook had a hundred competitors with the same five functions, but the best team. Microsoft had the operating system network effect and the best people. Google copied InfoSeek and had an algorithm only about 10% better, but the best people.
A video illustration of the power of the power law: «These Regular Guys Challenging NBA Player and Instantly Regretted It», where a player who for three years was 363rd out of 365 in the NBA, 12 years later and 30 pounds heavier, destroys the best of the G League with a score of about 13:0.
Sources of big companies: Facebook, PayPal, Stripe, Square, Salesforce, PeopleSoft and 8200. By geography, the Bay Area is about seven times better than any other ecosystem. Articles to read: «Your Life is Driven by Network Effects» and «Where To Build Your Startup». Sharing KPIs decides the business: someone else's click-through rate of 7% versus 2% once doubled their company.
On returning to the office the speaker is direct: creative and competitive industries need physical presence, and the back to office trend will strengthen over four years.
The most controversial section. The idea of interviewing customers about what they want is criticized harshly, because customers do not see the future.
A metaphor from the docks of the 19th century: if you ask the stevedores hauling sacks of grain onto a ship what they want, they will answer "better shoes", and no one will say "a block with a pulley". Breakthroughs come from non-users, from side transactions, not from the main ones.
The final principle: start from a transaction, from a concrete action by a concrete person. The most inspiring example, the Scots who bought used GSM networks in Europe and deployed them across Africa as purely commercial businesses, which transformed economies. And the conclusion: today you need to be not a missionary founder, but a speedy founder, a savage founder, a competitive founder, and you can have your mission later.
The third episode analyzed is entirely about language. The frame is radical: language is not an add-on at the end, but the center from which the product is born.
The key question to ask constantly: what is this for the user and where does it fit into their life. People are busy, they have health, work, romance, sports, taxes, vacations and dogs, and you live in a tiny sliver of their day.
A quote is given about how, if you cannot explain it simply, you do not understand it. The founder is a "language discoverer", a semiotician, because all of reality is structured by language.
A programmatic example: a cartoon from the New Yorker. Lemonade at 20 cents, no one wants it, but a lemonade detox at 20 dollars draws a line. The only difference is the words.
Then the thesis of a power law in words: the word "database" gives one result, "vector database" another, "CRM" a third. The word "free" is so strong that books have been written about it; one cent is already far from free. A FitMob example: out of thousands of phrases, the best click came from the anecdotal «Weapons of Ass Reduction», and the company grew for eight months on that alone.
A story from the dawn of digital photography. The home page said "store your photos", 2 to 3 thousand people a day came, and there was no growth. The decision to make the product multiplayer expressed itself in changing one word in the HTML.
A week later, sharing and viewing features appeared. Over the next six months, 47 million people registered for free. This changes the TAM, growth, the customer and the team's thinking: an engineer hired for storage objects that now they have to share.
Two examples of the same idea, diverging because of the wording. The speaker described his product as "a community of people voting on news" and put a photo and the author's name at the center. Six months later Digg appeared, then Reddit copied Digg, Digg undermined its own community, and today Reddit is worth tens of billions.
The second example: a dating site. The slogan "find yourself a partner" did not work, because everyone said that. A product manager suggested "help someone find a partner", and then married and taken people could also participate and invite friends.
The company bought up ads on Facebook and LinkedIn with variants of "Wars in Egypt", "Wars in Xanadu", "Wars in Paris". Having spent 2000 dollars, they saw that Atlantis was more popular than all of them. A second wave of tests: "Amazons of Atlantis", "Dragons of Atlantis", "Realms of Atlantis", "Wars of Atlantis". The dragons won.
A competitor came to the office and admitted that they were paying four times more per install, so they were going to launch a game about dragons and buy up all the dragon traffic. They had to be merged, and the business was sold for 800 million dollars.
Snapchat: according to Evan Spiegel, before «Picaboo» there were 34 projects and numerous name changes. «Peekaboo» launched in July 2011, was renamed Snapchat in September, and then it took off.
The Israel company Yevvo: "stream live video to any other user", 18 months with no result. Then the product Air: "stream live video only to your friends", better, but seven functions were missing. And finally Meerkat: "stream live video to Twitter followers in a single click". The key is that the distribution channel is built right into the tagline.
The technology, the people and the mission were the same, only the words changed. Three hours after Meerkat launched it was clear that it would be a hit.
The 1982 book the speaker recommends: «Positioning» by Jack Trout and Al Ries. The main ideas: the hierarchy and the physiology of the human mind. People are tribal and think about status all day long.
The Avis story: the slogan «We try harder» worked from the number two position. When Avis became number one and bought the ad «We're number one», people emotionally rejected it, and the company fell to third place, then returned to second, and the world calmed down. The same with beer: Budweiser first, Miller second for 25 years, so Miller invented a new class, "light beer", became number one in it, and remained so for 29 years simply because it changed one word.
In B2B the status quo is even stronger, because organizations are risk-averse and full of "policemen" who care that nothing breaks. That is why the phrase "no one gets fired for IBM" wins.
A consequence, in B2B the one who got out ahead, as a rule, stays there: he attracts talent, capital, marketing, big booths at conferences. This is called preferred attachment, and this is precisely why speed is critical.
Language is the foundation of growth, because it lies under advertising, sales, PR and virality. The speaker analyzes verbs: a verb should stand at the start of the phrase, and replacing the verb changes everything.
A measured example: the button «Invite your friends» versus «See who's in here!» produced a 22% increase. And the main example about Amazon: the wording «Items we recommend for you» was irritating, because it was pushy and looked like an attempt to foist junk on people.
Emotional hooks: «Create a Profile» is worse than «Start the Action!». And the principle: language should be about the user, not about you.
Slack looked like any other messenger, while HipChat had existed 6 to 8 years earlier with the same product. The difference was in the language and the presentation.
The second example is Anduril: the words «software-first», «speed», «venture-backed», «AI», «autonomous» and the slogan «Rebooting the arsenal of democracy». A name from «The Lord of the Rings», a self-definition as a «merchant supplier» versus «prime contractors», an antithesis to the language of Raytheon, and they sell rockets while doing it. Today it is the hottest defense contractor.
On the importance of language within the team, two witnesses: Jeff Wilke (COO of Amazon), who in every room and on every slide said «this word needs to change», and Sequoia, where they do not use the word «deal» but say «company» and «partners». And Simon Sinek's model: first why, then how, then what, whereas most sell from the what.
The final thesis: only founders can change the words on the home page. An agency will not help. If you are not strong at language yourself, find someone who is strong and give them co-founder status.
Language is at the center, because it touches hiring, fundraising, sales, marketing and Saturday decisions all at once. A power law in words means that one right word (share instead of store) changes the TAM, growth and the team's thinking. Positioning against the competition requires words: be number one or invent a new class (light beer). Always ask: what is this for the user.
The episode about the market map opens with a simple thought: founders under-invest in understanding the landscape, and it is precisely knowing the territory that gives speed. Many investors say they would pick the market before the founders.
The main provocation of the episode: a competitor is outsourced R&D. Don't bury your head in the sand, but study, copy the best and don't repeat mistakes. A founder should be Indiana Jones, a jungle guide for employees and investors.
The key figure that makes the market map important: an investor looks at a deck for an average of 3 minutes 44 seconds, and almost two minutes of that goes to the slide with the competitive map.
The reason: the investor is checking whether you are in a dead end, whether you are walking into a brick wall, whether he knows of eight similar companies that failed and why. That is precisely why the competition slide is valuable for the founder too.
A separate warning: don't consider a competitor stupid. He signals to the market with the same words as you, which covers the whole area, not just the little cell you placed him in.
The market map consists of four parts: a table of competitors, which you look at weekly; 2x2 matrices, in the plural; other charts for yourself, employees, journalists and investors; and an understanding of the real customer, leading to SAM and TAM.
| Data source | What is taken from it |
|---|---|
| ChatGPT, Claude, Gemini, Google | Quick gathering of facts and history |
| Crunchbase, DealBook, CB Insights | Databases of competitors and round histories |
| Industry articles | Context and dynamics |
| Investors | Other companies you remind them of |
| Customers and competitors / ex-employees | What was tried, what worked, what did not |
| Industry trade shows | Positioning and presence |
| archive.org (Wayback Machine) | Previous language, onboarding, positioning |
The table should be extremely detailed, because through a written breakdown the map enters the head and makes the founder fast. The required fields are listed.
The second tab, substitutes: not direct competitors, but replacements.
2x2 matrices: you always put yourself in the upper right corner, but the cuts of the market differ, and the speaker has seen companies with six such matrices. The point is fast communication: show a diagram and say, here we are, there they are, and the debate ends.
The fourth part of the map is the customer and the size of the market. You need to be able to point to the core of the segment, to the secondary segment, and to those you are definitely not going after in the first and second phase. And calculate bottom-up.
Bottom-up means specifics: how many salespeople I have, how many sales each one makes per week. Three levels: the initial target market (beta, alpha, the first months), the serviceable addressable market (accounting for the competition to be beaten), and the total addressable market as the terminal value after defeating the competitors.
Not an obsession with competition, but detailed knowledge. Knowing the territory reduces decision time when new data arrives, and increases persuasiveness in conversation with investors, customers, journalists and candidates.
The episode about revenue begins with the fact that once revenue appears, the only metric of speed is the rate at which it grows. But not all revenue is equal: high margin, recurrence, fast growth and defensibility (patents, network effects, embedding) are more highly valued.
The central model, the hierarchy of reasons to buy. At the top "make me money", then compliance/security, then ego and brand, then urgency, then saving money, then doing the job better, and at the very bottom saving time.
The trap of saving money and doing the job better is analyzed separately: the customer thinks "maybe", gets no compensation for it and does not buy, while the proposition "do the job better" sounds like an insult. Saving time matters to almost no one, even though everyone says the opposite.
The rule: usually the side of the marketplace that is easier to get pays. A consumer pays for himself, the state pays for protecting citizens, and sometimes there are indirect third-party payers, as in medicine, where the insurance company pays, not the doctor and not the patient.
The Zenefits example, at one time the fastest-growing SaaS company in history: the software was given away for free, and the health insurance companies paid 30 dollars a month per person, because Zenefits was in effect a broker and received a commission.
The order of priority of payers in B2B: CEO, then VP Sales, then CFO, and at the very bottom HR, the worst place to sell. The key question: who owns the budget, not who suffers from the problem.
A subscription sets a fixed price, and then customers with a low willingness to pay leave, while those who extract enormous value pay little. The solution: add usage payments, in order to reach the "whales".
A list of types of usage payments: per seat, per API call, for a "superpower", for storage volume, per transaction, for processed revenue, for location, for acceleration. On acceleration, a joke: «otherwise known as graft in third world countries».
The whole idea is to approximate, with a combination of subscription and usage payments, the individual willingness-to-pay curve and capture more of the area under the curve.
Gross margin is price minus variable and marginal delivery costs, including raw materials, shipping heavy parts, salespeople, marketing, a booth in Vegas, Facebook ads, the cost of financing and variable labor. The speaker gives benchmarks by type of business.
| Margin | Type of business (the speaker's examples) |
|---|---|
| 90% | Software or transaction fees (Visa); this is precisely what created the venture industry |
| 80% | Software with some inference cost, that is, AI |
| 65% | Software plus services |
| 45% | Expensive products, hard to manufacture |
| 40% | High-end services (McKinsey) |
| 22% | Cheap products |
| 10% | Brokerage and commodity services |
If you considered yourself a software business, and the margin came out at 40%, then you are not a software business, and investors will not value you as software. Margin is not reporting, it is a verdict on positioning.
Five methods. The first and simplest is benchmarking against existing prices: start at the price of a direct or substitute competitor. Pricier, and it is unclear why it does not work; cheaper, and perceived value falls and vital margin is killed.
The second method, cost benchmarking: calculate the minimum price from your costs. The third, value-based pricing: the price is a share of the economic value, usually 5 to 20%, typically 10% (if I save you 500, I take 100). The fourth, willingness-to-pay interviews: show a paper mock to four to ten customers and ask at what price this is (1) suspiciously cheap, (2) a good buy without hesitation, (3) expensive, (4) prohibitively expensive, and build a curve with the point of maximum revenue. A method from the book «Monetizing Innovation».
The fifth, tiering: set levels and see what people choose, then remove the ones that do not work and point out the "most popular". And channel-driven pricing: if you need resellers, retail, brokers, stores, partners or salespeople, room must remain in the price for their margin, CAC payback and commissions.
The densest block of the episode. A list of techniques with numbers.
| Technique | Effect |
|---|---|
| Anchor: was 142, now 92 | Conversion rises |
| Charm pricing, left digits (39 instead of 40) | Conversion rises at the same price |
| Decoy / third option (Economist 59 / 125 / 125) | The share choosing 125 rises from 68% to 84% |
| Popcorn: three sizes with a decoy | +70% revenue |
| Wine: second most expensive or second cheapest | People take neither the most expensive nor the cheapest |
| Boundary test: a $499 tier for rubies | ~40% of revenue from one option, up to $10,000/month per player |
| Social proof | Conversion growth, "most popular" |
| Paradox of Choice: 6 versus 24 kinds of jam | Optimum of 3 to 4 options |
| Perceived value / placebo | More expensive = perceived as more effective |
| Framing "only 10% fat" versus "90% fat-free" | A positive frame sells better |
| Free shipping framing | +80% choose the third option with a free product |
| Scarcity: rare and ultra rare | Allows charging more |
The Economist example: 59 dollars for the magazine or 125 for print and web, and 68% took the first; after adding a decoy in the form of a separate paper subscription, 84% began paying 125. The beach beer example: the same beer for 3 dollars at the fridge and for 15 dollars at a beach bar. The high fashion example: boutiques cluster, and a 2500 scarf makes an 800 belt an obvious bargain.
Analyzed separately: Google's page with a simple, calm price without a trailing nine; a warning that charm pricing may not suit a brand (Nordstrom, where it is simply 1700), and a reference to «Predictably Irrational» by Dan Ariely and «Influence» by Cialdini. And the king of all techniques is Booking.com, where all the tricks are used at once.
The word free triggers dopamine and an irrational reaction. Buy one, get one free gives 30 to 50% more conversion than a 50% discount, precisely because of the word.
An experiment with 14 cents: on a supermarket table 73% take one option and 27% the other, and if you make one of them free (the difference still being 14 cents), the preferences flip. But there is a warning.
A recommendation from the book «Freemium» by Dave Boyce: a free trial or freemium is better than completely free, because it does not attract a crowd of freebie hunters.
Many businesses fail because there is no natural point of transition from free to paid. The speaker gives a list of proven points.
And the F2 example from New York: after a 14-day trial the company charges for a whole year upfront, about a million dollars per customer, and because of this it needs less venture capital and gives up less equity.
The final slide of the presentation is a call to raise prices and not lower them out of fear. Seek a monopoly position, a product people will pay anything for, and a strange willingness to pay.
The strangeness of willingness to pay: people keep junk in storage lockers for 85 and 200 dollars a month for years; in games they pay for storing shields and spears; coffee has risen from 86 cents to five, seven, nine dollars; and Texas HoldEm Poker, the foundation of Zynga, sold game chips for real dollars with no right of withdrawal, only for status, and on this a company worth 11 billion grew.
The episode about fundraising begins with the fact that fundraising is a permanent function of the founder, not a pause before building. And it introduces the central model: the ladder of proof, by which an investor asks where you are and whether that fits his risk profile.
The rungs of the ladder, from bottom to top: team (LinkedIn and a reference), idea, market, advisors and expertise, ability to build the product, customers, retention, payments, revenue, and at the very top rapid growth.
| Rung | The investor's question |
|---|---|
| Team | Were you able to recruit a great team |
| Idea | What are you doing, is it timely |
| Market | Did you find a fast-growing market |
| Advisors | Who helps, where the expertise comes from |
| Build | Did you prove the team can build |
| Customers | Are there users of the product |
| Retention | Do people stay |
| Payments | Is anyone paying |
| Revenue | Is there revenue |
| Rapid growth | Is revenue growing fast |
Rapid growth is a special rung, because it zeroes out the others: if you are growing very fast, then you have a good enough team, an already built product, customers, retention and a found channel.
A tool: signal.nfx.com, where investors themselves raised their hand by stage (pre-seed, seed, series A, B) and by category (women founders, Israel, LATAM and so on).
A startup is a learning machine, and metrics show the team and investors that you are learning and climbing the ladder of proof. Without data there is no learning.
Requirements for metrics: timeliness and accuracy. The investor will ask where the number comes from and how you got it. You need to think about how the team sees the metrics and how decisions are made on their basis, and spend more engineering time on measurement and reporting systems than you would think.
Practices: from day one, a daily email with statistics to everyone on the team (at first two or three metrics, gradually more accurate and more numerous), metrics on the wall in real time, a weekly gathering with a projection on the wall.
Two principles. First: only the metrics you measure and show move. Second: you need to measure the painful metrics, the ones that change your behavior; vanity metrics give you a peaceful sleep, but do not change the business.
The most common metrics are the sales funnel with conversion at each stage, and retention curves. The speaker calls himself a "triangle chart reader".
A single metric is needed to focus attention and resources. Requirements: measure customer value, be actionable and directly correlate with revenue.
Examples: monthly revenue, NPS, DAU, retention rates, marketplace transactions. For a marketplace, the single north star metric is liquidity: if a transaction took place, then both the buyer and the seller are satisfied.
Milestones are the same rungs of the ladder of proof, reducing risk. They should be written out, made actionable and measurable, and each time you should ask how to halve their number and halve the time. An example of simplification is SpaceX, which makes rockets at 1/100 of NASA's cost precisely because it removed steps and milestones.
But for the metrics to move, you first move the rocks: gather the team together physically, get the story and the name right, set your head straight (determination, flexibility), find the customer, define the ICP and the distribution channel. And only then iterate like crazy.
A startup is a network, and the task is to bond investors, customers, journalists, influencers and talent to it. Each node brings its network, attention, status, traffic, capital, skills and reputation.
Compensating a node: equity to an investor, a dream to an employee, titles and status. Stronger nodes (Sequoia, Benchmark) demand more, because they give their reputation. And early nodes demand more, because they take more risk, since the ladder of proof is still low.
The speaker does not like the word pitch. He proposes the frame of a conversation or a review. The first five conversations with investors are needed to learn, not to sell.
The method of a founder who has been working for three years and is raising a 40 million Series B: eight minutes of pitch, then immediately a question he calls a can opener.
The question itself goes like this:
Another 36 similar questions are listed in the article «The Winning Psychology of Fundraising for Founders» on nfx.com. A caveat: do not burn all 30 minutes on the pitch, otherwise there will be no time left for an honest conversation. And context: in the 1980s and 90s pitching was necessary, because there were only about 150 venture capitalists, and now there are around 35,000, so you can meet three a day.
The shortest episode of the course and one of the most provocative. A caveat is made right away: this is not a call to build only three-person companies, but a way to talk about the principles of speed. Big teams go slow, small ones go fast.
The benefits of a small team are listed in theses: less drama, less politics, fewer salaries, less dilution, and you hire only A-players. But you cannot just sit three people in a room: they have to be able to run AI systems, wear many roles and work at a very high level.
| Company | ARR | People |
|---|---|---|
| Cursor | ~$100M | 30 |
| ElevenLabs | ~$90M | 50 |
| Mercor | ~$50M | 30 |
| Lovable | $10M in 2 months | 15 |
The speaker stresses that this is not the future, but the present, and that previously only narrow types of businesses could grow like this, like Instagram, WhatsApp and Groupon, and now many do. The mechanics: the founder sets up AI systems for sales, marketing, service, HR and legal tasks, they are like plates at the circus, you only need to periodically check that they are not falling.
The first role is the founder-visionary, who sees the opportunity before others and leads the product, like Steve Jobs. The second role is the numbers person, the «Mentat» from «Dune», who loves spreadsheets, metrics, finance, operations and details.
The third role is the words person, Don Draper, who turns AI output into a human message, who gives the taste, the feel and the brand. The formula: vision, numbers, words.
What AI cannot do is flashes of insight, and that is the founder's advantage. About 90% of the company will be AI agents, but 10% will not, and that is where the founders work.
A new forecast: hyper unicorns, companies that become unicorns within six months of launching.
And the ironic finale of the episode: the speaker recalls that the script was written by AI, the video was edited by AI, and the only thing really on camera is him in the room and the camera, and adds: «or am I avatar? Hard to say».
If you can break your thinking work into three functions (vision, numbers, words) and hand execution to AI systems, team size stops determining the scale of the company. There is one limitation: the founders' ability to catch the moments of a change in direction.
The final episode is an encyclopedia of speed. The speaker breaks the tactics into three categories: planning and thinking, team and culture, operations. He calls the collection an "encyclopedia of speed" and asserts that if you internalize them, you will be able to go fast.
| # | Tactic | Rationale |
|---|---|---|
| 1 | Make sure the idea is good | A pivot costs 6 to 12 months; moving in the wrong direction is not fast |
| 2 | Find fast-moving water | Companies become slow when it "works"; move yourself into the fast part of the market |
| 3 | Formulate a mission and vision | Speed compounds, you need long focus; novelty fades |
| 4 | Find the white-hot center in your customer base | Whoever is the most desperate buyer is the point of acceleration |
| 5 | Move to a startup city | You need employees, capital, customers, press and ambition that understand speed |
| 6 | Say "no" | No to customers, revenue, geography, acquisitions that distract and slow you down |
| # | Tactic | Rationale |
|---|---|---|
| 7 | Work with people you have already worked with for years | Firing a co-founder costs 6 to 12 months |
| 8 | A culture built around speed | In job postings, on the About page, in meeting rules, in celebrating the right failures |
| 9 | Speed as the main hiring criterion | A Soldier and a Police do not learn speed; you need Commandos |
| 10 | Ship on day one | An engineer ships code to production on the first day, a salesperson makes calls, support talks to a customer |
| 11 | Minimize company size | Less communication, hiring, onboarding, culture, firing, dilution |
| 12 | Minimize project team size | No more than six people, a pizza-box team, less handover and communication |
| 13 | A board of three people | Two founders and one investor, you can go up to Series C; the rest are observers |
| 14 | The contract ready by the end of the interview | Give 24 hours to sign; people value confidence in them |
| 15 | Recruit continuously | A live pipeline of exceptional people instead of a 3 to 6 month delay |
| 16 | Fundraise continuously | Keep the deck and the investor table; capital accelerates |
| 17 | Advisors and mentors, 10 to 25 people | Build slowly, decide quickly; CEO friends especially |
| 18 | Externally imposed deadlines | Customers, investors, conferences, competitive pressure instead of personal urgency |
On advisors the speaker weighs honestly: building a network of advisors is slow, but it makes the company fast. And external deadlines are needed, because relying on your own urgency is exhausting.
| # | Tactic | Rationale |
|---|---|---|
| 19 | Your own office, where you can close the door | «Every great company started with its own door»; coworking spaces rarely produce great companies |
| 20 | Ship product daily, not weekly | Reducing batch size speeds up everything; ship fair and good |
| 21 | Remove everything that makes you wait | The unit of speed is completed cycles; build, ship, learn, decide |
| 22 | Pay for speed | Saving three days for 600 dollars, with a burn of 40 to 100 thousand a day, is unconditionally worth it |
| 23 | Parallelize cycles | No one waits for another; do several things at once |
| 24 | Find and remove the main bottleneck | The single constraint on speed; attack it with the whole team (Elon Musk, military operations) |
| 25 | Talk in person or by phone | A five-minute call replaces 30 minutes of messaging or hours of waiting |
| 26 | Cheap, fast feedback | Tests with target users (usertesting.com), watching the video with the whole team on Friday |
| 27 | Automatic daily metric emails | People at all levels make decisions faster; management stays sharp on the choice of metrics |
| 28 | Wall panels with real data | A sense of a live business for employees, visitors and investors |
| 29 | The company priority table | Half an hour a week of public debate about priorities; it prunes pet projects |
| 30 | Create competition | Investors, customers, partners, candidates, journalists: urgency, scarcity, rivalry |
| 31 | Remove deadwood | Products, features, hires, meetings and strategies survive through inattention; notice and kill them |
| 32 | Prototype before building | A mock, a fake, manual delivery, Wizard of Oz before a scalable product |
| 33 | Reduce the team's fear | A different URL, a small market (New Zealand, Canada), a rollout to 1% or 0.1% of users |
| 34 | No remote work under 100 people | Remote work kills speed by 50% or more |
| 35 | Set the time, not the result | Not "how long will it take", but "what can we get done in 48 hours" |
| 36 | Remove distractions | An excuse that everyone around you understands, for example an accelerator with an immovable deadline |
| 37 | Work more hours | Early cultures slept in the offices; in China Saturday is normal; 70 to 100 hours is the bar |
| 38 | Remove, reduce, simplify | Remove features, processes, requirements, approvals; the best process is no process |
| 39 | Don't reinvent the wheel | The core is yours, but non-core features can be shamelessly copied from Google, Facebook, Microsoft, OpenAI, Anthropic, Apple |
| 40 | One accountable owner | One owner, not a committee; accountability reduces coordination time |
Despite the «40 tactics» title, in the finale the speaker says a different number and adds two bonuses.
Bonus one: distinguish Type 1 decisions (irreversible, one-way doors) from Type 2 (reversible). About 95% of decisions are reversible, so they should be made, tested and, if necessary, reversed. A decision is made at 70% of the information.
Bonus two: say "fuck it" often. The speaker calls this the ethos that changed his life over 20 years of running four venture companies.
The episode title speaks of 40 tactics, but in the finale the speaker says «42» and adds two bonus tactics (fast decisions and «say fuck it»). Below we give this number as it sounds in the source, not as in the title. The earlier wording «the best 40 tactics» is also heard at the start of the episode.
The last episode of the course is devoted to the mental models that speed a founder up, and those that kill him. The frame: after four startups and ten years of venture capital it becomes clear that everything is decided by the psychology, mindset and personality of the founders, along with language, speed and the right idea.
The speaker argues against the idea of Steve Blank and Eric Ries that founding is for everyone. In his view, it is not for everyone, because people have different mindsets, personalities and psychologies. VC firms publicly name the traits they seek: NFX says speed, Sequoia says disagreeableness, Founders Fund says heretical and contrarian thinking, Hummingbird says neurodivergence, and one venture fund, whose name he does not disclose, devotes the second interview to your relationship with your parents and guesses remarkably well at pre-seed.
NFX has a model of 48 founder attributes, part about mindset, part about personality, part about psychology. The speaker's first company was a psychological testing company: 150 million registered users, 25 billion questions answered, five PhD psychologists on staff.
1. Speed. It can be learned, like the other models. 2. Drive. Why you are doing this: to prove your father wrong (like Musk), to prove your worth, anger, revenge (a boss fired you, a co-founder left), attention, love, the desire to be in Wikipedia, power, victory over enemies, money, freedom.
3. Managing a psychologically violent environment. The speaker says he quit founding after four companies because he got PTSD, and that no one before him had spoken about it publicly. Firing a co-founder, fighting a bad board member, losing a big deal, this is psychological violence.
4. Suffering. At the fifth HBS alumni reunion he stood before the class and said that he had built and sold a company, but suffered terribly, and that school had not prepared him for how many sleepless nights and how much NyQuil it would take to get three or four hours of sleep.
5. Commitment. «At the moment of commitment, the universe conspires to assist you». «Burn the boats», leave yourself no path of retreat. The speaker warns that talented and educated people are precisely masters of optionality, and this harms them. An example from the film «300»: Leonidas refuses to flee, and this is called glory.
6. Judgment. It develops through a large number of decisions made. In a startup there will be far more decisions than at McKinsey or Goldman, so judgment grows faster.
7. Tolerance for ambiguity. The two biggest predictors of entrepreneurship in a Harvard study: a high tolerance for ambiguity and the fact that your parents were founders.
8. Extreme ownership. The book by former SEALs that is handed out at Stoke and other strongest portfolio companies: no excuses, no complaints, remove blame and ego and fix it. If this happens six times with a person, he is fired.
9. Be reality based. Honest information is hard to get, so you have to ask for the negative and ask for a «no». A fresh example: the board of a company that rose to a 500 million valuation and dreams of 1.5 billion ran on «la-la, everything is great for us», without the necessary level of negativity.
He asks you not to be Doug Burgum, the 12th GOP candidate in 2024, who put in 13 million of his own money, raised another 4 and dropped out of the race, having spent 16.9 million, because everyone around him kept telling him «I support you».
10. Strong opinions loosely held. In the 1990s they said that the entire contribution of a VC to the board is «focus, focus, focus». 12. Be contrarian: read those who think independently, and train heretical thinking.
10X. An improvement of 4% or 10% means nothing while you are small. Only 10X in every area matters in the market. The exception is Microsoft, where even 4% will be eaten up.
14. Culture. Quotes: «Culture eats strategy for breakfast» (Drucker), «Obsessed over who sits next to who» (Jobs), and «How we talk to each other is core» (Jeff Wilke of Amazon, who in every meeting would stop and say how to think about a thing).
15. There is no corporate transformation without a personal one. If you want a quiet life, get a degree and go to McKinsey or Goldman. If you want a personal journey, then you are in the business of personal transformation, and every failure is an experiment on yourself. In three or four years of a startup you will mature as if in 30 years.
16. The risk of a mediocre life. Failure should be feared less than a mediocre life. 17. Thirteen ways to do something (from Dennis Highwater): eleven of them will work, just pick one and go.
18. Fundraising. Don't fixate on the investor's money, take his information. Ask for a «no». By the ninth conversation another investor will realize that you know everything, and will offer to invest himself. And remember that 90% of them will not invest, even if the pitch is perfect.
19. Paranoia. «Only the paranoid survive». Do not become complacent, do not give room to vanity, fame and self-congratulation. An example is the company that rose from 500 million and targets 1.5 billion and on the board is «letting itself be dragged into this», which the speaker calls rot capable of killing the company, which is why he is going to fly to them and sit with them.
20. Commando thinking. A high speed bar, simplification, going where the action is, finding resources. 21. The goldilocks zone. In every attribute of a business there is a scale from arrogance to timidity, and the right confidence lies somewhere on that scale. But the goldilocks zone is not the middle: the advice to «just find the middle» the speaker calls nonsense.
22. Grit and determination. Resilience, endurance, the ability to cope with adversity, in the spirit of Rocky: what matters is not how hard you hit, but how hard you get hit while you keep moving forward.
Separate from mindset, personality is harder to change. The best personality test in the world is the five-factor personality test, the «big five», invented in 1972: openness, conscientiousness, extroversion, agreeableness, neuroticism.
The desirable profile: openness, conscientiousness and extroversion high; agreeableness and neuroticism low. Disagreeable people do not need to be liked, and therefore are free to think uniquely and heretically, pass more easily through conflicts and violent events, more often get what they want, and along with them the company gets it too.
Other unexpected correlations: childhood trauma, neurodivergence, Asperger's syndrome and ADD.
The title promises 22 mental models, but the spoken numbering slips: number 11 is skipped (between strong opinions loosely held and be contrarian), the 10X model is spoken without a number, and the commando model likewise without a number. In addition, in the middle of the list the speaker says «there's 21 on this list», although the title and the finale speak of 22. We keep his wording as it is and do not credit him with greater precision of numbering than the recording had.
The user has two assets: (a) VL Export / vlxport.com, a B2B platform of verified Chinese factories, which is pivoting to factory subscriptions (about 500 dollars a year) plus bundled inspections and the sale of commodity audits to buyers, with crypto only on the seller side; (b) arliwork.com, «agentic lab in Shenzhen».
Let us analyze the pivot honestly, through the hierarchy of «why people buy». At the top of the hierarchy stands «make me money», followed by compliance and security. Inspection is not about saving money, but precisely about compliance and security: the buyer pays so as not to receive defective goods, not to lose a contract and not to risk his reputation. This is good news, because compliance holds the price. The bad news is that you can slide into the part of the hierarchy where the buyer says «maybe», if you sell inspection as «saving on defects». Saving money and saving time almost never sell.
Practical takeaway: a commodity audit should be presented as protection against risk (off-take, quality, legal cleanliness, reputation with the end buyer), and not as a cheaper purchase. And the factory subscription should be sold on the logic of «make me money»: the factory pays 500 dollars a year for access to a flow of Western orders and verification, that is, for incoming revenue, not for an abstract service.
On «who pays». In the VL Export model, the side that is easier to get and that receives the money pays: factories subscribe because they need a flow of orders, and buyers pay for the audit because they need the safety of the deal. Crypto only on the seller side falls into the same logic: for the factory it is a tool for receiving money, and for the Western buyer it is an unwanted complication (compliance), which is why it is right to keep it out of the buyer's path.
Before fixing 500 dollars a year, it is worth conducting willingness-to-pay interviews: show 4 to 10 factories a mock and ask at what price it is suspiciously cheap, a good buy without hesitation, expensive, prohibitively expensive. This gives a curve with the point of maximum revenue and protects against a lower bound imposed by fear.
The principle «language first, product second, engineering third» for positioning an agentic lab means that you must first find the phrase a Western client will latch onto, and only then build agents for it. Not «we make AI agents» (a category where everyone is), but a verb plus a result for the user: not «stream video», but «stream to followers with one click». In a Chinese-Western context the working formula is built around reducing risk and speed of execution, not around the technology itself.
The Anduril story is directly relevant: they took an antithesis to the language of the incumbents (merchant supplier versus prime contractor, «Rebooting the arsenal of democracy») and sold rockets, and the difference was precisely in the words and positioning. For ARLI this means choosing language that separates the lab from «yet another outsourcing agency». The thesis about B2B risk aversion («you can't get fired for buying IBM») also works: a Western B2B buyer will choose a comprehensible supplier, so the wording should remove risk, not emphasize exoticism.
Let us test the three-roles hypothesis on ARLI. Is there a product visionary? Is there a numbers person (Mentat) who loves metrics, finance and operations? Is there a words person (Don Draper) who turns opportunities into human messages and a brand? If any of the three blocks is empty, that is the bottleneck of speed according to allometric scaling: heavy structures require supporting weight. In a context where 90% of the work can be handed to AI agents and 10% kept by the founders, the words person is especially undervalued, because it is precisely this role that is responsible for Western demand for the lab's brand.
The ladder of proof is applicable to the user's conversations on both sides. With factories you stand on the high rungs (market, expertise, ability to build), but for the Western buyer the lower rungs (retention, payments, revenue) must be closed quickly, otherwise an investor or partner will not believe you. The rapid growth rung zeroes out the rest: if VL Export shows rapid revenue from subscriptions and audits, the team, product and retention are automatically considered proven.
The market map for the commodity-audit terrain is built in four parts: a detailed table of players (SGS, Bureau Veritas, Intertek, local Chinese inspectors, AI quality-control platforms), a substitutes tab (manual photo reports, trust in an intermediary, a personal visit to the factory), 2x2 matrices along the axes of «price/coverage», «report speed», «depth of verification» and a bottom-up TAM from the number of Western procurement buyers, not «1% of 200 billion». Knowing this landscape is precisely what makes decisions cheaper and faster.
The course is written for Silicon Valley venture software, and some of the advice transfers poorly to the «China for the world» context. Below is an honest list of what should not be copied literally.
The NFX course is a strong system for one type of business and one geography. Everything concerning founder psychology (fear, commitment, extreme ownership, the hierarchy of buying, language first, the ladder of proof) transfers almost unchanged. Everything concerning structure, jurisdiction, PR and partnerships should be adapted to the transactional Chinese-Western reality, not copied literally.
The course can be compressed into a single contour: take away fear and the low speed bar, find the right idea on fertile soil, name it with the right words, know the terrain better than everyone, raise the price and sell into the right part of the hierarchy of buying, climb the ladder of proof toward rapid growth, keep the team small and combine people with AI, and let all of this be held together by a set of mindsets and personality traits.
| Episode | Core | Key figure |
|---|---|---|
| The Speed Talk | Fear, simplification, psychology | Releasing 4 times a day |
| Hidden Patterns | The idea matters more than effort | 68 of 75 ideas are bad |
| Magic Words | Language is primary | 47M in 6 months after changing a word |
| Market Map | Knowing the terrain | A VC looks at the competition slide ~2 min out of 3:44 |
| Speed to Revenue | Hierarchy and behavioral economics | Decoy 68% → 84% |
| Ladder of Proof | Metrics and rungs | Retention is the answer 80% of the time |
| 3-Person Unicorn | Allometric scaling and AI | Cursor $100M / 30 people |
| 42 Tactics | A tactical set | 42 (in the finale) versus 40 in the title |
| 22 Mindsets | Psychology and personality | Big Five: agreeableness and neuroticism low |
Speed is not about how hard you work, but about how many cycles you manage to complete. The unit of speed is a completed cycle: built, shipped, learned, decided. Everything else in the nine episodes is ways to remove what prevents cycles from completing.
ARLI membership: the notes from the workshop
One site of yours, measured, with the price for the work in the same document
ARLI, Inc., a Delaware corporation. Questions: hq@arliwork.com.