A critical reading of the corpus on the four levers of growth (speed, language, idea, psychology): what in it is corroborated by data, what is rhetoric, and what of it transfers to the practice of VL Export and ARLI in the context of "China for the world".
The nine NFX lectures form not a set of tactics but a coherent doctrine with a single central claim: in a competitive environment, speed is the master lever, and the other advantages (idea, language, money, talent) are derivative of it, because fast iteration pulls them up and levels up their quality. The corpus insists that speed is not about the number of hours, but about removing barriers: mental, operational and structural. James Currier, the author, founded four venture-backed software companies and has made about 300 investments, and builds the doctrine on his own experience as an operator and an observer.
The corpus derives the other three levers from speed. The idea determines where effort is directed: going fast in the wrong direction is worse than not going at all. Language determines how effort is sold outward and how it organizes the team from within. Psychology determines who is even capable of holding the pace. The master lever is master precisely because the three layers subordinate to it work only when speed works, whereas speed itself works even with a mediocre idea (it simply hits a ceiling).
The corpus offers a strong, operationally applicable framework for iteration and for attracting resources into an early stage, but it presents as a law of nature something that is merely one mode of growth (high-risk venture), and therefore it applies to VL Export and ARLI only in part, with clearly drawn boundaries.
The practical point for the reader: the document below separates the verifiable grain (the learning loop, competitive intelligence, the economics of small teams, the psychology of fear) from the ideological shell (hostility to partnerships, the cult of the office, anti-PR) and shows where the second block contradicts the conditions of the "China for the world" business.
Before transferring anything, one must name the nature of the material. This is not a neutral textbook, but a venture firm's creed, spoken by a partner of the firm before founders to whom the firm gives money. Such a text has a built-in interest: the firm benefits from founders moving faster, because speed accelerates the exit and raises the returns of its portfolio. This does not make the advice false, but it means the advice is systematically biased toward actions that increase turnover rather than durability.
The second source of bias is survivorship. Almost every piece of evidence in the corpus is selected by outcome: the success stories of Google, Airbnb, Zynga, Uber, Dropbox, Meerkat, Houseparty. Selection by success does not show those who applied the same tactic and died, so attributing causality to the technique itself (for example, "start with a skateboard") does not withstand strict scrutiny. We see the result, but we do not see a comparable sample of failures.
The third source of bias is geography and culture. The corpus is built on the American (above all Californian) context: Delaware jurisdiction, Californian talent ecosystems, American legal risks of copying, an English-language language of positioning. Many of its stances (fast firing, improvisation in the legal structure, a bet on PR-free growth) assume a dense market of capital and talent that may not exist in the "China for the world" context.
| Bias | How it shows up in the corpus | How to handle it |
|---|---|---|
| Self-reporting | All figures are given from the founder's memory (revenue, valuations, user counts) without primary data | Treat metrics as illustrations of mechanics, not as proofs of a law |
| Survivorship | Almost exclusively winners are examined; unsuccessful applications of the tactics are not shown | Before adopting, seek counterexamples within your own industry |
| US-centrism | Delaware, Californian ecosystems, the English language, American law | Localize each tactic for the Chinese-export environment |
| VC interest | Emphasis on accelerating turnover and exit, not on durable margin | Filter advice by the criterion of business durability, not by the speed of an event |
| Dogmatic absolutism | "Never do partnerships", "never work remotely", "don't hire PMs" | Read as a default strength, not as a prohibition; look for the conditions of exceptions |
The corpus contains not a single falsification procedure. No piece of advice comes with a condition under which it would prove harmful. A doctrine that cannot be refuted by its own examples is closer to a religion than to a method, and requires external verification.
The corpus explicitly names four supports in one of the lectures: language, speed, the right idea and mindset (psychology). The value of this framework is that it turns the question "how to grow" into the question "which lever is currently in deficit". Diagnosis is simpler than total effort: if the product does not sell, the deficit is usually in language, not in the product; if the team is not moving, the deficit is in psychology, not in processes; if everything is fast but in vain, the deficit is in the idea.
The key claim about the interaction of the levers: speed multiplies the other three. Fast iteration accelerates the discovery of the right words, the early finding of new energy in the idea, and the learning of psychological resilience through repetition. Hence the corpus's conclusion that speed comes first not because it is the end goal, but because it accelerates the work of all the other levers at once.
The reverse dependency is also important and is implied in the corpus: without an acceptable idea, speed only accelerates the burning of resources, and language is what makes speed visible to the outside world (investors, talent, the press). Psychology is what makes speed durable over time, because the fear the corpus talks about acts as a brake on all three layers at once.
Speed sits at the center because it accelerates the other three nodes. Psychology is marked as the binding constraint: it does not create growth, it releases the brake on all the other levers. Capital is not a fifth lever, but a consequence of the evidence that the first four produce.
The most valuable substantive core of the corpus is that it redefines the unit of measurement of speed. The unit is not an hour and not a task, but a completed cycle: build, ship, learn, decide. A company is measured by the number of completed loops per unit of time, not by the volume of unfinished work. This is an engineering definition (reducing batch size, shortening waiting time), and it transfers almost without loss.
From the definition of the cycle follows the arithmetic of movement: if you make two moves for each move of a competitor, you win even against a larger and smarter rival. The logic is not that each of your moves is better, but that the number of moves per period multiplies quality: among twenty attempts a working one is more likely to be found than among one. Hence the corpus draws a strong normative conclusion: ship daily, not weekly, experiment by the dozens of attempts per month.
The corpus separately and repeatedly insists: speed is not the number of hours. It is the removal of barriers, not the piling up of effort. However (see the section on contradictions), the list of tactics also contains the directly opposite advice to work more hours, which makes this statement internally inconsistent and requires the reader to decide for themselves.
The economic justification of speed in the corpus is built on a bundle of three quantities that it calls the "holy trinity": speed, quality and cost. The thesis is that they are not mutually exclusive, and that at a high pace of iteration quality grows faster and cost falls, because a saved day at a large daily burn costs more than the premium paid for accelerating delivery. This lends itself to direct verification against your own P&L.
Defining speed through a completed cycle rather than through busyness is the most transferable element of the corpus. It is neutral to industry and jurisdiction: VL Export can measure itself by the number of completed cycles "factory check, onboarding to the platform, first order", rather than by the number of hours worked.
The corpus insists that the main brake on speed is not operational but psychological, and that fear is the base constraint. The diagnosis is simple: people slow down not because they cannot, but because they are afraid to judge the result. Hence the corpus's project is one of redefinition: the product is not the founder's identity, work is a series of experiments, not the building of a monument. Each of these redefinitions lowers the cost of failure and thereby releases pace.
The corpus builds a taxonomy of fears and ways around them: fear of judgment about the product, fear of uniqueness (cured by borrowing), fear of the final decision, procrastination until the money runs out. Self-sabotage and the psychological mechanism of "proving that I am not a failure" are named separately, which, by the author's observation, correlates with founders' slowness more strongly than other factors.
Here the corpus draws its strongest and most contested distinction: between skills (mental models that can be learned) and personality traits (which hardly change). It refers to the five-factor model (the Big Five) and claims that high conscientiousness, extraversion and openness plus low agreeableness and low neuroticism correlate with founder success. The metaphor of "commandos versus soldiers and police" is drawn directly from this: speed supposedly cannot be taught to those in whom it is not dispositionally given.
This is the key point for criticism, because it conflicts with the other premise of the corpus, that speed can be learned. The resolution of the contradiction is that the corpus conflates two claims: speed as a set of practices (teachable) and speed as a personality trait (not teachable). For practice, the first matters more.
| Layer | Element | Teachability per the corpus | Practical takeaway |
|---|---|---|---|
| Skill | Speed as practices (small batches, cycles, deadlines) | Yes, trainable | Implement processually, not by relying on a type of person |
| Skill | Judgment (through the number of decisions) | Yes, grows with experience | Give people more decisions, not more instructions |
| Skill | Tolerance of uncertainty | Partly | Screen at intake, train at low risk |
| Personality | Agreeableness (low is desirable) | Almost none | Do not try to remake; account for it in hiring |
| Personality | Neuroticism (low is desirable) | Almost none | Build an environment that lowers the cost of error |
| Motive | Drive / need to prove | Given, but manageable | Use as fuel, watch for self-sabotage |
The commando/soldier/police metaphor is useful as a diagnostic language, but dangerous as a hiring criterion: it turns a poorly agreed-upon personality trait into grounds for dismissal. The corpus itself admits that this is "very hard to change", and thereby turns the label into a sentence. For a Chinese team, where agreeableness may be culturally reinforced, the label is especially risky.
The idea layer of the corpus is built on the most counterintuitive claim of the whole series: big companies are rarely born of a revolution, more often of two or three obvious changes to something that already works. This is "new energy", the narrow zone between total novelty (the market does not understand it) and pure cloning (legal and reputational costs). Facebook, Airbnb, Google and Lyft are presented in the corpus not as inventors, but as fast followers who changed two or three parameters.
From this is derived the distinction between market risk and execution risk. The first means betting that demand exists at all, the second that you are able to serve it better than others. The corpus, relying on the track record of the venture market, recommends the second bet: the market is proven, you compete on execution. This directly contradicts the popular advice to seek a "blue ocean", and the corpus deliberately defends the red ocean: in it there are prices, channels and buyers that can be copied and verified.
The layer concludes with the "under the curve" model and the boundary layer: a business becomes real when the product of audience size and average ticket crosses a threshold. The corpus illustrates this with the gaming economy, where a large audience at a micro-ticket (a few cents per person per day) yields a real business, while a small audience at a high ticket in some cases yields nothing. This is a quantitative intuition, not a formula, but it transfers.
The preference for the red ocean rests on portfolio experience, not on a measurement of the probability of success of an individual company. The logic "in the red ocean there is something to copy" is true for speed of market entry, but may be false for long-term defense: the red ocean has no structure of its own if you do not bring new energy along with a network.
The language layer is the most operationally useful part of the series, because it gives concrete, verifiable techniques. The central claim: first language, then product, then engineering. Not at the end, as "copy for marketing", but at the beginning, as the source of the product. You find the words that describe the desired result for the user, and the product is built to fit those words, not the other way around.
The corpus proves the thesis not by logic but by rewriting one word in the interface: replacing "store photos" with "share photos" led to registrations growing into the millions within months. The technique transfers and requires no trust in the ideology: if the words at the entry point change behavior, this is measurable and reproducible in any industry.
The second layer is positioning through category. The corpus examines the hierarchical nature of human perception: if you cannot be first in an existing category, you must create a new category in one word (the classic example being "light beer"). The third layer is semantics at the level of the verb: the difference between the buttons "share" and "invite", between "recommended" and "people who bought this also bought". This is the engineering of trust through grammar.
An important conclusion of the corpus: language cannot be delegated. Founders are the only ones who can change the words on the home page and in the presentations, so language is a personal duty, not a marketing function.
| Level of language | What is decided | Transferable technique |
|---|---|---|
| Word of the result | What value the user sees | Rewrite the headline and measure the change in conversion |
| Category | Whether the comparison is with the leader | Create a new category in one word, rather than fight for first place |
| Interface verb | What exactly the user feels as their own action | A/B at the level of the button, not only the screen |
| Product persona | The attitude with which the user is addressed | Set the "character" of the product as consciously as a function |
| Team language | How employees think about priorities | Fix forbidden and mandatory words inside the company |
Language is at once the cheapest change and the closest to revenue. Unlike hiring and office tactics, language experiments require neither money nor structure, and therefore it is the best starting point for a business without venture capital.
The pricing layer begins with an ordered hierarchy of purchase motives. The best motive to make someone buy, per the corpus, is the promise of additional income: if you give the client more money than you ask for, refusal is irrational. Below that come conformity and safety, then ego and status, then urgency. The worst motives are saving money, "doing the work better" (offensive to the buyer) and saving time (declared, but not paid for).
The second link is the question "who pays". The corpus separately highlights indirect monetization, when the consumer does not pay: insurance companies, banks, third parties. The example where free HR software was monetized through insurers' commissions is a model where the one who pays is the one for whom the value is higher. A hierarchy of the sales addressee in B2B is also given: best of all the CEO, then the VP of sales, then the CFO, worst of all the HR department.
The third link is the construction of the price. The corpus contrasts a flat subscription with a subscription plus a usage fee. A subscription fixes the price and loses both those willing to pay less than the base rate and those who get far more value (the whales). A usage fee recovers part of that difference. For a B2B platform this is directly relevant, since it allows combining access at a low annual subscription with variable payment for volume.
The fourth link is behavioral economics as a tool for presenting price: anchoring, the left-digit effect, the decoy third option, social proof, the free effect. All the techniques are empirically verifiable and transferable, but the corpus itself warns that they may not combine with a premium brand.
A factory subscription plus a variable fee for inspections is exactly the "subscription plus usage" construction that the corpus describes as a way to recover bargaining power with large clients. The annual subscription fixes access, while the per-batch fee captures value from those who buy a lot.
The capital layer introduces a single axis for all external conversations: the ladder of proof. An investor, a journalist or a candidate asks one question: which rung are you on, and does that rung match my risk profile. The rungs run from team and idea through experts and product to customers, retention, revenue and, finally, rapid growth. Rapid growth is special: it displaces all other questions, because it implies answers to them.
From the model follows a non-obvious operational conclusion: fundraising is not the search for money, but the search for information and the building of hub connections. The first meetings should be used for learning, by asking questions about what you are underestimating, rather than for presenting. This lowers the fear of rejection and at the same time accelerates the refinement of your own position. The corpus describes a shift of stance from "pitch" to "conversation" or "consultation".
The second conclusion is the linking of nodes: an investor, a client, a journalist, a talent are nodes that bring their capital, status, traffic and trust into your network. The compensation of a node depends on the rung of the ladder: the earlier a node enters, the more risk it takes and the more it asks. This formalizes why a company's valuation rises as it moves up the ladder.
The practical mechanics that the corpus gives as operational consist of continuity: keep a ready presentation and a list of investors at all times, rather than starting fundraising at the moment of need. This lowers dependence on a single round and distributes risk over time.
The most transferable part of this layer is the redefinition of fundraising as information gathering. It works the same way with a venture fund, a bank, and a large buyer: the earlier a conversation becomes learning, the faster you find the rung you need.
The lecture on the market map gives the most underestimated, but the most mechanical tool of the series. The claim is simple: you cannot move fast across unfamiliar terrain. The market map is not an exercise for a presentation, but a tool of decision speed, because knowledge of the terrain turns every new event (a competitor's arrival, a price change, a client's departure) into a fast and precise conclusion, rather than into panic.
The corpus sets out four artefacts. The first: a competitor sheet maintained as a living document, with details down to the names of executives, investors, and the history of name and positioning changes. The second: several two-dimensional matrices with different axes (not one), so as to see the market from several angles. The third: additional visualizations for an external audience. The fourth: a precise definition of the customer, broken into a core, a secondary segment, and an explicit rejection of segments.
A separate element is the substitutes tab: not direct competitors, but alternatives that take up the same time and attention (the example: video versus games, margarine versus butter). The corpus insists that substitutes must be known just as thoroughly, because the user's budget competes between categories, not within them.
The finale of the lecture is a methodology for sizing the market. The corpus rejects the "top-down" approach (taking a share of a big number) as unconvincing and requires a "bottom-up" calculation: through the number of sellers, their productivity, the number of deals. This is boring arithmetic, and that is precisely why it works as decision infrastructure: it gives numbers to lean on in an argument and in planning.
Talking to former employees of competitors and reviewing their old pages in the web archive yields information about probabilities almost for free. For VL Export this means: interviews with former managers of Chinese trading platforms will yield more than any market report.
The newest and most speculative part of the corpus is the thesis about small teams. It is built on an observable fact: companies reach revenue in the tens of millions with tens of employees, and such cases are more common than ten years ago. The corpus explains this through allometric scaling: the smaller the system, the fewer supporting structures it needs, and the reduction compounds in both directions.
The mechanics of the thesis is that founders set up agentic systems for sales, marketing, service and legal tasks that run like "spinning plates", and the founders check on them from time to time. From this is derived a requirement for the team's composition: three roles. The visionary (product vision), the numbers person (metrics, finance, operations) and the words person (brand, message, taste). All three must be able to manage AI systems.
The corpus itself outlines the limit: AI does not produce flashes of insight and does not change direction. That place remains with the founders, roughly a tenth of the work. The corpus does not disclose the technical details of that share, and this makes the thesis more a direction than a method.
It is critically important that the thesis is presented as a prediction partly come true. The selection of examples is again by survivorship: companies that achieved the result are shown, but the share of small teams that achieved nothing is not shown. Therefore the thesis should be read as a description of a possibility, not as a guarantee.
The thesis does not account for the fact that a small number of people means a high concentration of risk in three personalities and a lack of redundancy. For a B2B platform selling verification and trust, where trust often requires presence and human contact, a small team may deliver speed and at the same time undermine reputation.
The corpus is not homogeneous, and its contradictions matter more than its agreements, because the reader has to choose a side. The first and main contradiction: speed as a teachable set of practices versus speed as an innate trait that "cannot be taught". The corpus holds both premises at once, and this is not carelessness, but a sign that it serves two audiences (founders and investors).
The second contradiction: "copy a lot" versus "start with a skateboard". Both tactics are named as sources of speed, but the first assumes borrowing ready-made solutions, while the second assumes building a small version of something that does not yet exist. Resolution is possible only situationally: copy the non-core, and build your own at the minimal scale.
The third contradiction: "never do partnerships" versus a business model built on a platform and nodes. A platform by definition is a system of partnership ties; the advice to forgo partnerships applies to old-fashioned business-development negotiations, but not to the construction of a platform.
The fourth contradiction: "hire only commandos" versus "hire those you have worked with for years". The first is based on personality type, the second on trust and on reducing time; they may require different people. The fifth: "no PR, obscurity is a blessing" versus "build in the open and attract resources". The sixth: "work in one room" versus the allometric model, where size and presence supposedly do not matter. The seventh: "speed is not about hours" versus the tactic "work more hours".
| Contradiction | Thesis A | Thesis B | How to resolve |
|---|---|---|---|
| Nature of speed | A teachable set of practices | An innate trait, not teachable | Implement the practices; account for type in hiring, but not as a sentence |
| Source of the idea | Copy a lot | Start with a skateboard | Copy the non-core, build your own minimally |
| Partnerships | Never do partnerships | Business as a network of nodes | Avoid biz-dev negotiations, but build a platform network |
| Hiring | Only commandos | Only those you have worked with for years | Trust as a filter, trait as a signal |
| Publicity | Obscurity is a blessing | Build in the open, attract nodes | Quiet at the start, loud when raising capital and talent |
| Team form | Work in one room | Few people, AI systems, location does not matter | Office until scale, distribution after a hundred |
| Hours | Speed is not about hours | Work more hours | Practices matter more than hours; overwork is a consequence, not a goal |
The contradictions do not devalue the corpus; they show that it is not a method, but a catalogue of techniques for different circumstances. The reader's task is not to pick one side of the doctrine, but to determine which side matches the current deficit of their business.
For the transfer to be honest, the corpus must be divided into three classes of claims: supported by data (at least observable and verifiable on your own business), based on a single example (a story), and pure assertions or rhetoric without support. The key theses are classified below.
| Thesis | Class of support | Strength | Comment |
|---|---|---|---|
| Small batches and less waiting time accelerate iteration | Observable engineering practice | Strong | Verified directly against your own process |
| One word change changes conversion | Examples + logic of the mechanics | Strong | An A/B test confirms it for anyone |
| Hierarchy of purchase motives | Behavioral economics + observation | Strong | Consistent with external literature |
| Competitive intelligence accelerates decisions | Observation, a cross-cutting motif | Strong | Low cost, clear effect |
| The red ocean is better than the blue | Portfolio experience, selection by survivorship | Medium | True for pace, contested for defense |
| New energy (2-3 changes) | Stories of winners | Medium | There is no reverse sample |
| Personality type (commando) predicts success | Self-report + the Big Five | Medium | Correlation without causality; risk of the label |
| The three-person unicorn | An observable trend + a prediction | Medium | Selection by survivorship, high concentration of risk |
| "Being a founder for everyone" is false, it is a gift | Firm rhetoric | Weak | Not verifiable; serves the firm's selection |
| Remote work halves speed | "Data that we see", without disclosure | Weak | A claim without disclosing the method |
| Copying everything is the norm (Google and others) | Anecdote + a bold generalization | Weak | Conflates idea and execution |
| Delaware, simple contracts, zero bureaucracy | Practical advice | Medium | Locally true in the US, requires revision outside the US |
Accept the strong class as is, test the medium class on your own business before transferring, discard the weak class. Most of the operational core falls into the strong class; almost the entire ideological shell falls into the weak.
Context: a B2B platform of verified Chinese factories moving to a factory subscription (roughly 500 dollars per year) plus package inspections, where the goods audit is sold to the buyer's side, and crypto payments are permissible only on the seller's side. Four layers transfer here from the corpus: price, market map, the ladder of proof and language.
The price transfer is direct. The construction "annual subscription plus a variable fee for inspections" is the "subscription plus usage" model described by the corpus. It solves a concrete problem: factories with low volume pay the base rate, factories with high volume return through inspections the value that a fixed rate loses. The corpus adds the warning not to lower the price out of fear and to build the price as part of strategy, not as a reaction to negotiations.
The transfer of the market map is realized as a competitor sheet in three planes: existing Asian trading platforms, inspection and audit services (substitutes by the function of trust) and clients' in-house procurement teams (substitutes by budget). The transfer of the ladder of proof shows up in the fact that the conversation with a factory and the conversation with a buyer are held at different rungs: with a factory you sell access and demand, with a buyer you sell an already-confirmed offer.
The language transfer is the cheapest and the most powerful. The category name and the words at the entry point determine whether VL Export is compared to a cheap trader or to a system of trust. The corpus insists that language is not the marketing department's job, but the founder's personal work, and on this platform it is precisely the wording "verified factory" that determines the price the market is ready to pay.
| Layer of the corpus | What transfers to VL Export | Order |
|---|---|---|
| Price | Annual subscription ($500) + fee for package inspections; the audit is sold to the buyer | 1, immediately |
| Language | The trust category and the entry words; A/B of headlines and calls to action | 2, immediately |
| Market map | Competitor sheet + substitutes tab (inspections, in-house procurement) | 3, first month |
| Ladder of proof | A different rung for the factory and the buyer; gathering information, not pitching | 4, first quarter |
| Cycle speed | Measure the cycle "factory onboarded, first order", not hours | 5, ongoing |
The corpus says nothing about crypto payments, but its logic of simplification suggests: if crypto complicates the buyer's journey, it must be kept on the seller's side (settlements with the factory), and the buyer should be given the simplest possible payment channel. This corresponds to the principle "remove everything that creates waiting".
Context: arliwork.com as an "agent laboratory in Shenzhen" and a separate line "we deploy agents for companies". Here the corpus is applicable above all through the thesis about three roles, and it must be applied honestly, not decoratively. The test is simple: does ARLI have three distinct functions of a visionary, a numbers person and a words person, and if not, which of them is missing and who is covering it.
The second transfer is the very question of whether deploying agents is a viable door for revenue. The corpus offers verification through the "hierarchy of motives" and "who pays": if deploying agents promises the client a saving of time, that is the worst motive, per the corpus, because the client declares concern but does not pay. If it promises additional income or a reduction of risk (including compliance risk), it is viable. It follows that the "agent deployment" door must be sold through money or safety, not through "we will make your work easier".
The third transfer is the role of the "agent laboratory" as a site of accelerated learning (an accelerator in the corpus's terms). The corpus says directly that the value of an accelerator is not in the learning, but in an external deadline and a mandate to focus. The Shenzhen laboratory can play this role for itself: create insurmountable external deadlines so that every product line moves as a separate experiment, not as unfinished construction.
ARLI: is there a separate holder of the numbers (metrics, unit economics, decision discipline) and a separate holder of the words (category, positioning, product language)? If both functions are in one person or are absent, the thesis about three roles points to a concrete deficit that must be closed by hiring or by explicit distribution.
Finally, the corpus warns against self-deception in selling agents: if the line grows on the enthusiasm of early clients, that is not proof of the "revenue" rung on the ladder of proof. Retention and a predictable channel are needed, otherwise ARLI will find itself in the medium class of evidence (a story), not in the strong.
Directness matters more than completeness here. A number of the corpus's stances are derived from the American venture context and, in the context of "China for the world", are either harmful or require adjustment. They must be named explicitly, so that they do not seep into practice under the guise of "principles of speed".
| Stance of the corpus | Why it does not transfer to the "China for the world" context |
|---|---|
| Zero partnerships and biz-dev | An export-sourcing business rests on relationships with factories, logistics and payment intermediaries; forgoing partnerships destroys the model itself |
| Everyone in one room is mandatory | A Chinese-export team is physically distributed between Shenzhen, buyers and customer markets; a rigid "one room" makes some functions impossible |
| Delaware incorporation by default | Jurisdiction, payments and crypto settlements are dictated by the requirements of Chinese counterparties and channels, not by the convenience of a US investor |
| Obscurity is a blessing, zero PR | For a platform of trust, visibility and reputation are the product; quiet growth undermines the function of verifiedness |
| The red ocean is preferable to the blue | In an overcrowded trading segment, copying entry means a price war; the trust category is precisely a blue zone within a red market |
| Quickly fire the "non-commandos" | The Chinese labor context and cultural norms make rigid selection by personality type risky and expensive |
| Never hire PMs | For platforms with a two-sided market, the product function is critical; the advice is drawn from particular cases |
The general criterion: what transfers is what describes mechanics (cycles, language, prices, proof), and what does not transfer is what describes the conditions of the American venture market (jurisdiction, office, publicity, hostility to ties).
Below is a sequence of concrete actions, flowing from the strong class of evidence in the corpus and bounded by the named limits. The order reflects the principle "first the cheapest and closest to revenue, then infrastructure, then scale".
| # | Action | Business | Owner | Timeline |
|---|---|---|---|---|
| 1 | Introduce measurement of the completed cycle instead of man-hours | Both | Founder | This week |
| 2 | A/B language wording for the category and calls to action; replace a prominent word of the result | VL Export | Founder (words) | 2 weeks |
| 3 | Rewrite the pricing grid as subscription + usage, test on five factories | VL Export | Founder (numbers) | 3 weeks |
| 4 | Assemble a competitor sheet and a substitutes tab; interview former employees | VL Export | Founder | 4 weeks |
| 5 | Test the agent-deployment offer through money or compliance, not through saving time | ARLI | Founder (numbers) | 4 weeks |
| 6 | Conduct an honest audit of the three roles; if there is a deficit, close the function by hiring | ARLI | Founder | 6 weeks |
| 7 | Start a daily metrics digest and a priority list on the wall | Both | Founder (numbers) | 6 weeks |
| 8 | Introduce external deadlines for each product line (a laboratory mandate) | ARLI | Founder | 2 months |
| 9 | Reframe fundraising and sales as information gathering, not pitching | Both | Founder | Ongoing |
Do not start with what requires money and people. Start with language and price, because these are the cheapest and the closest to revenue, and build the infrastructure (market map, metrics) in parallel as a habit, not as a project.
Since the corpus contains no falsification procedures of its own, they must be introduced from outside. Below is a set of refutable signals that would show that the corpus's doctrine is wrong or only partly right in your context. Each signal should lead to a revision of the corresponding tactic, not to blind adherence to it.
| Signal | What it refutes | Reaction |
|---|---|---|
| The number of completed cycles grows, but revenue and retention do not move for months | That speed is the master lever | Then the deficit is in the idea or the channel, not in the pace; stop the race |
| Language A/Bs yield no significant change over several iterations | That language is primary for your audience | Return priority to the product or the price, recheck the segment |
| The subscription grows, but buyers do not order package inspections | The "subscription + usage" model | Simplify to a single tier, test whether the buyer pays at all |
| Agent deployment sells, but client retention is low | That this is a viable revenue door | Change the sales motive or close the line |
| Forgoing partnerships slows procurement and payments | The dogma of "no partnerships" | Accept that the platform model requires a network of partners |
| Hiring by the "commando" type produces turnover in the Chinese team | The transferability of selection by personality | Replace the filter with trust and with teachability of practices |
| The market map does not change decision speed in practice | That intelligence always pays off | Reduce the artefacts to one that works |
If two or more signals hold for more than a quarter, the doctrine of speed in its pure form does not suit the context. Then move to the mode of "practices without ideology": keep the cycles, language, pricing and proof, discard the office, jurisdiction, hostility to ties and the cult of personality type.
Below, both businesses and the four levers are brought together into a single transfer table, so that the final decision is visible on one page. This is the answer to the question of what exactly to take from the corpus and what to leave behind.
| Lever of the corpus | VL Export | ARLI | Decision |
|---|---|---|---|
| Speed (cycles) | The "factory, order, retention" cycle metric | External deadlines for product lines | Transfer |
| Language | The trust category, the entry words, selling to the buyer | Positioning as an "agent laboratory" | Transfer |
| Idea | A blue category within a red market | The agent door through money/risk | Transfer with a caveat |
| Psychology | Remove the fear of judgment from the verification team | An honest audit of the three roles | Partly |
| Ideology (office, jurisdiction, anti-partnerships) | Do not apply | Do not apply | Discard |
The NFX corpus is valuable as an engineering vocabulary of speed and as a discipline of proof, and dangerous as a venture ideology in American packaging. For VL Export and ARLI, one should take from it the mechanics of the four levers and leave outside the boundary of transfer everything that describes the conditions of someone else's market of capital, labor and law.
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.