a16z releases reputation network Cosign: How to uncover top talent and social capital flow ahead of the market

a16z Cosign: What is most scarce in the AI era is no longer resumes, but who is willing to bet on you. If you only understand a16z's newly launched Cosign as "another LinkedIn," you will miss what it truly aims to do. It is not simply trying to redesign the professional homepage. Nor is it just providing entrepreneurs with another place to find jobs. What it truly targets is a type of core asset that has always existed in Silicon Valley but has long been unstructured: Reputation Capital. On September 25, 2026, a16z officially launched Cosign and defined it as a curated professional network aimed at the entrepreneurial ecosystem. The official vision includes: directories of startups, investors, and practitioners; reputation relationships such as "who shaped your career," "who has truly fought alongside you," "who is willing to work with you again," and "who is worth keeping an eye on"; as well as private matching for investment, recruitment, and collaboration intentions. a16z even describes it as "a child of Wikipedia, LinkedIn, and early AngelList." But what is truly worth studying is not the product features. Rather, it is a major trend behind it: AI is making "self-description" increasingly cheap, while making "others willing to stake their credibility on you" increasingly expensive. In the past, a person could prove themselves to the market through: Education; Resumes; Experience at big companies; Beautiful portfolios. In the AI era, many of these signals are rapidly depreciating. Thus, the job market is transitioning from: Credential Economy to: Reputation Economy. This is the position that Cosign truly wants to occupy. ──────────────── 1. The real problem in the talent market has never been a lack of information, but a lack of credible information. The most absurd thing about today's recruitment market is: Information is increasing, but judgment is becoming increasingly difficult. A candidate can have: LinkedIn; GitHub; X; personal website; Substack; portfolio; AI-optimized resume; AI-optimized cover letter. In theory, the information available to recruiters is dozens of times richer than it was ten years ago. But what is the reality? After a job is posted, it may instantly receive: hundreds; thousands; or even tens of thousands of applications in the future. a16z partner Olivia Moore gave an example in a Cosign discussion, stating that when she posted a job on LinkedIn, she received about 2,000 responses; they expect that as Browser Agents and automated job applications continue to develop, this scale will further expand. This signifies a very important economic change: Information abundance creates trust scarcity. The richer the information, the scarcer the trust. ──────────────── 2. The biggest side effect of AI in recruitment is driving the "cost of applying" close to zero. In the past, applying for a job required: reading the JD; modifying the resume; writing a cover letter; preparing materials; filling out forms. So even if these processes were inefficient, they at least created a friction. This friction would automatically filter out some people. After the emergence of AI Agents: one machine can read 10,000 job postings; assess the match; automatically modify resumes; automatically submit applications. Thus: Cost of Applying → 0. When the cost approaches zero, the number of applications theoretically will expand infinitely. This is a very typical phenomenon in economics: Spam Problem. Why did spam emails emerge later? Because: sending an email costs almost nothing. Job applications are entering a similar phase. ──────────────── 3. When "expressing interest" becomes free, the interest itself loses value. In the past, someone would write a three-page cover letter for your company, this was a signal. Because: they invested time. In the future, AI can generate it in three seconds. The signal is gone. In the past, a founder would write a highly customized fundraising email to an investor, indicating they had done their research. Now AI can automatically generate emails for 500 funds at once. The signal is gone again. So the increasingly important question for the future is: What can AI not cheaply counterfeit? One answer is: another real person staking their credibility on you. This is the underlying logic of Cosign. ──────────────── 4. Why might an endorsement from a top figure be worth more than a Stanford degree? The Cosign team posed a deliberately sharp question in the program: If a young person: graduates from Stanford; and another has no similar label, but has long been followed, interacted with, and recognized by industry figures like Patrick Collison, Elon Musk, and Marc Andreessen, which signal is more worthy of study? They clearly lean towards the latter. This is not to say: that universities are useless. Rather, it is saying: The nature of the two types of credentials is completely different. Stanford is: Institutional Credential. Certification given by an institution. Willingness of top peers to interact with you over the long term is closer to: Peer Credential. Peer certification. ──────────────── 5. Why are Peer Credentials particularly valuable in high-uncertainty industries? Assume recruitment for: Accountants. There are: CPA; licenses; standardized exams. Institutional Credentials are very effective. But if recruiting for: AI Researchers; Founding Engineers; Startup Product Leads; Early Investors. What standardized exam do these positions have? None. Moreover, the best people often: are doing things that have no standard answers yet. At this point, the explanatory power of traditional Credentials naturally declines. What you really want to know is: Do truly outstanding people think this person is outstanding? This is: Reputational Transfer. ──────────────── 6. Why can reputation be transferred? Assume you do not know a young engineer at all. But Patrick Collison publicly says: "This is one of the best engineers I have ever worked with." What will happen to your judgment? You have not obtained: his code; his complete performance data. But you know: Patrick Collison himself has a high reputation for judgment. So you will update your probability. In Bayesian terms: Prior Probability is updated. This is the most core yet most invisible operating system in Silicon Valley. a16z even summarized in the article announcing Cosign: "Silicon Valley knows people by watching other people know them." Silicon Valley understands people by observing "who knows whom, and who trusts whom." ──────────────── 7. This is actually the core mechanism of the investment world: Signal Extraction. Venture capitalists do one thing every day: Extracting signals from a lot of noise. 10,000 entrepreneurs: Who is truly impressive? 1,000 AI startups: Which one is worth investing in? 500 young engineers: Who might become future founders? The most valuable information is never: this person saying they are impressive. But rather: who was willing to take on the reputational risk for them when they were not yet obvious. This is called: Costly Signal. Signals with costs. ──────────────── 8. The true value of an endorsement lies in the fact that the endorser may lose credibility. If I casually praise a stranger: the cost is close to zero. The signal value is also close to zero. But if Marc Andreessen publicly says: "This young founder is one of the best I have ever seen." He is actually staking: his judgment. If he endorses 100 scammers in a row, his endorsements will become worthless. Therefore: Endorsement Value ∝ Endorser Reputation × Scarcity of Endorsement. The higher the endorser's reputation; the more restrained the endorsement; the stronger the signal. This is very similar to Credit Rating. ──────────────── 9. Therefore, what Cosign should truly build is not a Social Graph, but a Weighted Trust Graph. The biggest problem with LinkedIn is not: a lack of connections. But rather: Connections have almost no strength. You can: work with someone for 10 years. Or: have met them once in a meeting. The page shows: 1st-degree connections. This is equivalent in graph theory to: all edge weights being almost the same. But the real world is not like this. There exist: weak relationships; collaborative relationships; mentor relationships; bonds of camaraderie; high trust; willingness to stake money. These edges should have different weights. What Cosign aims to do is essentially: Weighted Reputation Graph. ──────────────── 10. LinkedIn focuses on "who you know," while Cosign aims to determine "how much you truly trust someone." These are two completely different types of data. LinkedIn: A connected to B. Cosign: A would work with B anywhere. Or: A thinks B is a person to watch. Or: A would fund B if B starts a company. The latter has a much higher information density. Because it expresses not just: Relationship. But also expresses: Conviction. The most valuable data in Silicon Valley is not the network itself. Instead: Who has strong conviction in whom. Olivia Moore directly describes this as: "The most valuable data in all of Silicon Valley is who has conviction in whom." ──────────────── 11. If this judgment is correct, what Cosign is actually building is a kind of "human rating system." Why do financial markets need: Moody’s; S&P; Fitch? Because there are too many assets. Investors do not have the capacity to conduct complete investigations one by one. Thus, there is a need for: credit signals. The talent market is the same. In the future, the number of candidates brought by AI may be so large that: no company can investigate them one by one. Thus, the market naturally needs: Reputation Layer. This layer does not necessarily have to be called Cosign. But it will almost certainly appear. Because this is an inevitable product of an AI-rich world. ──────────────── 12. What is truly scarce in the AI era is not Intelligence, but Verification. This rule actually applies not only to recruitment. AI makes: Article generation cheap; Code generation cheap; Image generation cheap; Company demos cheap; Job applications cheap. Thus: Creation Cost ↓ brings: Verification Cost ↑ People are increasingly less concerned about: "What can you generate?" and more concerned about: "Is this thing real?" "Have you really done this?" "Who can prove it?" This is why one of the largest industries in the future may be: Verification Economy. ──────────────── 13. Why did universities have such strong social value in the past? Universities certainly provide: Education; Research; Community. But elite universities have another often overlooked function: Filtering. Stanford tells the market: This person has gone through a highly competitive screening process. Harvard does the same. Thus, recruitment companies do not have to screen individuals from all over the world one by one. Universities do part of the work: Filtering. This is called: Credentialing Function. ──────────────── 14. The AI era is weakening the monopoly of "static credentials." The biggest feature of academic qualifications is: Obtained once. It may follow you for decades. But the modern tech industry changes too quickly. A person: Enters Stanford at 18, does not fully indicate their: Technical ability; Judgment; Creativity at 35. On the other hand, A person without a top degree, may continuously accumulate public signals through: GitHub; X; Open source projects; Entrepreneurship; Writing. Thus, professional reputation begins to shift from: Static Credential to: Dynamic Reputation. Academic qualifications tell you: They passed a screening at some point in the past. Dynamic reputation tells you: What this person is currently doing. ──────────────── 15. This is actually a shift in professional identity from "certificates" to "real-time balance sheets." Traditional resumes: Periodically updated. The ideal Reputation Graph envisioned by Cosign: Continuously updated. Today: Someone publicly recommends you. Tomorrow: Join an important project. Three months later: Product launch. A year later: Colleagues are willing to start a business with you again. These continuously form: Reputation Ledger. So future professional identity may increasingly resemble: Corporate financial statements. Continuously having new data coming in. Rather than: A certificate issued ten years ago. ──────────────── 16. Why does X have a large number of high-value talent signals but has never fully transformed into a professional network? Because X is: Feed-first. Not: Profile-first. Today Patrick Collison praises a young engineer. At that moment it is very valuable. Two days later: It is drowned in 10,000 new pieces of content. A year later: Almost no one can find it. This is called: Reputation Evaporation. Cosign aims to: Turn these momentary signals into: Durable Reputation. a16z also clearly stated when introducing the product that excellent endorsements happen daily on X but disappear in the feed, while Cosign hopes to permanently attach these signals to people and companies. ──────────────── 17. The value of this may be greater than imagined, because the job market is essentially a "historical memory market." Why are top VCs willing to quickly write checks for a certain founder? Often because: "I worked with him ten years ago." Why can a top engineer quickly recruit former colleagues when starting a business? Because: Past credibility has not disappeared. The real problem is: This credibility is today largely stored in: People's heads; Private WeChat groups; Signal; WhatsApp; Text messages; X; Private emails. This is a very inefficient: Unstructured Database. Cosign wants to structure it. ──────────────── 18. If Cosign really succeeds, its core asset is not users, but "reputation edges." Traditional KPIs for social platforms: MAU; DAU; Time Spent. What Cosign should really look at is: How many high-quality Endorsement Edges. For example: 1 million users but each person only: casually likes. The value is very low. 100,000 users but with a large number of: Founder → Engineer; Investor → Founder; Mentor → Operator strong signals, the value may actually be very high. So what it is really managing is not: Attention Graph. But: Trust Graph. ──────────────── 19. This is fundamentally different from Facebook's early Social Graph. Facebook: Who knows whom? LinkedIn: Who worked where? X: Who follows whom? Cosign: Who believes in whom? If it can truly answer the fourth question, the data economy value may be very high. Because the first three describe: Relationships. The fourth describes: Expectations. And the essence of investment and recruitment is precisely: Making expectations about the future. ──────────────── 20. Why is "Person to Watch" one of the most interesting features of Cosign? Because it is essentially creating: Talent Futures Market. Assuming a person is now: 22 years old; An ordinary Tesla engineer; No one knows him. You publicly label him today: Person to Watch. Seven years later: He becomes the president of Cognition. At this point, you gain an intangible asset: Prediction Track Record. Cosign mentioned the case of Russell Kaplan in the program: seven years ago, he was still a 22-year-old Tesla engineer, already chosen by some insiders as someone worth watching; he later became the president of Cognition. ──────────────── 21. This means Cosign is not only scoring talent but also scoring the "talent scouts." This is a very important layer. If a user: Discovers A in 2018; Discovers B in 2020; Discovers C in 2022. Later: All three become top founders. Then this person's: Talent Picking Ability is validated. This is completely similar to VC Track Record. What matters most for investors is not just: How much they manage today. But: Who they discovered in the past when no one understood. Thus, Cosign theoretically establishes: Talent Reputation and: Scout Reputation. ──────────────── 22. This may give rise to a very new profession: Reputation Arbitrageur. What do financial investors do? Discover: The market undervalues assets. Then buy in. There is also: Mispricing in the talent market. An engineer's current annual salary: $180K. The actual creation ability may be: $1M. It’s just that the market does not know yet. If you can discover early: Hire him. This is: Human Capital Arbitrage. The same applies to venture capital. The founder has not yet started a business. You recognize him first. After starting a business: Invest at the first opportunity. Thus, the ability to discover top talent is essentially also: Alpha. ──────────────── 23. One of the most profitable Alphas in Silicon Valley has never been stocks, but "knowing future people in advance." Think about YC. What is the real advantage? It’s not the office. But: Knowing the founder before he becomes a founder celebrity. Sequoia is the same. Top headhunters are the same. Top university labs are the same. They are all doing: Early Talent Capture. So: Getting to know outstanding talent early on, is essentially an option. If the other party ends up being ordinary: The loss is very small. If the other party becomes: The next generation Zuckerberg; Altman; Collison, The value of the relationship may be huge. This is called: Asymmetric Social Optionality. ──────────────── 24. Why does sharing this kind of Alpha publicly seem counterintuitive? Assuming I discover: A certain AI engineer is very impressive. The most intuitive action should be: Keep it confidential. Because once it’s public: Others will rush to hire. Other VCs will rush to invest. So why tell the whole world on Cosign: "Person to Watch"? This is one of the smartest game designs of Cosign. It believes: Early public endorsement of others will generate future returns. ──────────────── Twenty-five, this is called "Reputation Investing" You publicly endorse someone unknown today. If this person succeeds later: Everyone will find out: You were an early believer. Thus, your judgment reputation increases. Moreover, the person themselves will likely remember: When the world had not yet recognized them, Who recognized them first. This creates: First Believer Advantage. The first believer advantage. a16z's explanation of Cosign also clearly emphasizes that those who discover and help a talent early often have better opportunities to recruit, collaborate, or invest first in the future. ──────────────── Twenty-six, this is completely isomorphic to VC investment Why do VCs publicly say: "We invested in this company at Seed"? On one hand, it’s certainly for publicity. On the other hand, it’s also to establish: Judgment Track Record. If a fund: Consistently enters before great companies’ Series A ten times, Founders will actively come to seek it out. Thus: Prediction Accuracy → Reputation → Better Deal Flow → Better Prediction Opportunities. Forming a flywheel. Cosign wants to replicate this flywheel from: Company investment to: Talent. ──────────────── Twenty-seven, thus Social Capital will also start to generate compound interest Suppose you help a young Founder. They succeed. Later they will help you: Meet clients; Introduce talents; Bring investment opportunities. Then help the next generation of Founders. Thus your network starts to: Compound. This is why truly top-tier networks are not: Knowing 10,000 people. But rather: Having 100 increasingly strong people, Willing to collaborate with you long-term. The value of these two is completely different. ──────────────── Twenty-eight, this is also why "number of connections" is a very low-level social metric LinkedIn: 5,000 Connections. It’s meaningless. What really matters is: How many people: Are willing to take your call; Are willing to introduce you; Are willing to vouch for you; Are willing to give you money; Are willing to start a business with you again. This is called: Effective Network Capital. The value of future career networks is increasingly likely to shift from: Connection Count to: Conviction-weighted relationships. ──────────────── Twenty-nine, why "Would Work With Anywhere" is a very strong signal? Because working together is the most brutal due diligence in the real world. In social settings: Anyone can perform well. Working for 3 years: You cannot fake it. You will see: How they react under pressure; What they do in the face of failure; Whether they have a sense of responsibility; The truth of their technical abilities; Whether they take credit; Whether they are willing to take on responsibilities. So when a former colleague says: I would unconditionally work with this person again. The information density is extremely high. It’s several orders of magnitude stronger than: "He is my LinkedIn Connection." This is the significance of Cosign’s "In the Trenches With / Would Work With Anywhere" design. ──────────────── Thirty, this is actually a kind of "Career NPS" In the consumer field, there is: Net Promoter Score. The question is: Would you recommend this product to a friend? The job market can have a similar metric: Would you bet your next five years of career on this person? If the answer is: Yes. This is a very strong: Human NPS. If future recruitment platforms can accurately obtain this data, The value is immense. ──────────────── Thirty-one, "Who Shaped My Career" solves another problem: invisible mentorship relationships Career growth is often not: Nurtured by the company. But rather: Nurtured by a specific person. A manager; A professor; A founder; A colleague can change career paths. Traditional resumes do not record these relationships. But these relationships can tell the market: This person’s: Thought system; Career culture; Ability inheritance comes from where. This can be understood as: Professional Lineage. Career lineage. ──────────────── Thirty-two, Silicon Valley has actually always had a very strong "apprenticeship" structure, but no one has drawn it out Fairchild Semiconductor: Went out to form: Intel; AMD; A large number of chip companies. PayPal: Formed the famous: PayPal Mafia. Facebook: Went out to form a large number of founders. OpenAI: Is also forming a new talent network in the future. Companies not only create: Products. But also create: Human Networks. What truly impacts the industry for decades, May very well be the latter. If Cosign can structure these relationships, It will start to have a: Innovation Genealogy. Innovation genealogy. ──────────────── Thirty-three, this diagram is of great value to investors Suppose you discover: A new founder from: A very strong team. All three of their former colleagues are willing to work with them again. Their former boss thinks: Person to Watch. Industry experts are also willing to: Would Fund. These signals together, Are more valuable than a: Pretty Pitch Deck. This is: Multi-source Reputation Underwriting. Underwriting talent through multi-source reputation. VCs are essentially: Underwriting Future Humans. ──────────────── Thirty-four, Cosign’s Private Intent feature may be more valuable than public endorsements The public layer solves: Reputation. The private layer solves: Transaction intent. For example: I am willing to invest in this person. I am willing to hire this person. I am willing to work with this person. In the future even: Willing to acquire this company. If this information is public: Many people will not express it. Because there are: Embarrassment; Political risks; Current employer relationships; Risk of rejection. Thus Cosign designs: Private Signaling. Only triggers connections when both parties match. ──────────────── Thirty-five, this is actually solving the most classic problem in the job market: Double Coincidence of Wants In early monetary theory, there is a concept: Double Coincidence of Wants. Two people must exactly have: You have what I want; I have what you want. The job market is the same. A founder really likes a certain engineer. But doesn’t know: If the engineer wants to leave. The engineer really likes a certain company. But is afraid that reaching out will seem: Too desperate. Thus both sides: Do nothing. Market transactions do not occur. Cosign’s Private Intent is equivalent to: Reducing Matching Friction. ──────────────── Thirty-six, if this system scales, it could even form a "Talent Dark Pool" Wall Street has: Dark Pools. Trading parties do not have to make all orders public. The job market may also see: Talent Dark Pool. Top employees do not publicly say: I want to leave. Top VCs do not publicly say: I want to invest in you. But the platform knows: Both parties have Interest. When a match occurs: Automatic introduction. This is especially important for high-end talent. Because the more outstanding a person is: The less willing they are to publicly announce: "I am looking for a job." ──────────────── Thirty-seven, this is also why LinkedIn Jobs is the strongest, yet not necessarily suitable for top talent The mass recruitment market needs: A lot of public supply and demand. The high-end talent market is exactly the opposite. Truly top-tier people: Usually do not lack offers. What they need is: Curated Matching. Not: More positions. But rather: Which position is worth wasting three years of life on? So the top talent market may ultimately resemble: Private Banking. Rather than: Supermarkets. ──────────────── Thirty-eight, Cosign’s AI cold start strategy is actually very worth learning for entrepreneurs The biggest difficulty for any new social network is: Cold Start. No one: No content. No content: No one comes. This is the classic chicken-and-egg problem. Cosign did something very clever with AI: Automatically collects public: Funding; Hiring; Personnel changes; Product launches; Public endorsements to build company and talent pages. a16z clearly states that AI can now automatically fill in company data, conduct online research, filter spam, and generate industry watch feeds. The core product team even consists of only three people. This means: AI has significantly reduced the Cold Start Cost of Social Networks for the first time. ──────────────── Thirty-nine, this may be the most underestimated impact of AI on social entrepreneurship In the past, to create social products: You had to first persuade users to fill out their profiles. Users are lazy. Thus the page is empty. Today AI can: Help you generate profiles first. Then ask: Is this you? Confirm it. This is called: Pre-populated Network. Users do not create from scratch. But rather: Claim + Edit. This is a completely different growth model. ──────────────── Forty, why can Wikipedia grow large? Because pages can exist before the person themselves. Elon Musk does not need to create a Wikipedia page himself. Others create it. Wikipedia can therefore cover: All important figures. LinkedIn is different. In theory, users must: Register themselves; Fill it out themselves. If Cosign combines the two: AI automatically generates public career profiles; The person claims it later, It may solve a significant cold start problem. This is also why a16z compares it to: Wikipedia + LinkedIn + early AngelList. ──────────────── Forty-one, but here also hides one of Cosign's biggest risks: who has the right to define your professional reputation? If the platform automatically generates a profile for someone: What if the information is wrong? What if something is missing? Which endorsements does the algorithm consider important? Who has the right to delete? This starts to involve: Reputation Governance. Reputation is not ordinary data. It directly affects: Funding; Jobs; Career opportunities. So once Cosign grows large, It will actually have a very large: Career Allocation Power. This needs to be handled with great caution. ──────────────── Forty-two, "only allowing positive signals" is a very clever yet very dangerous design. Cosign currently clearly emphasizes: It will not establish a negative evaluation mechanism, Mainly focusing on: Endorsement. Why? Because if allowed: "This person is bad." The platform immediately enters: Defamation; Retaliation; Workplace politics; Anonymous attacks quagmire. So: Radical Positivity Helps reduce the psychological cost of participation. But the problem is: Positive-only Systems will create Inflation. Everyone praises everyone. In the end, the signals may depreciate. ──────────────── Forty-three, the real solution to "endorsement inflation" is not negative evaluations, but making endorsements scarce. If users can: Watch 10,000 People. This label has no value. A truly high-quality system should limit: How many: Top Cosigns a person can have. Or allow sorting. Cosign has already mentioned in discussions that it will conduct rank ordering. This is very important. Because: Scarcity creates signal. If you can only recommend: The 5 most promising people of the year. You will take it seriously. This is similar to VC investing: Capital is limited, So every investment itself is a signal. ──────────────── Forty-four, the most powerful product of Cosign in the future may not be personal homepages, but rather a "Reputation Score." Of course, it doesn't necessarily show a: Score of 728. But the system backend can easily form some kind of: Implicit ranking. For example: Who endorsed you; The accuracy of the endorser; Depth of relationship; Time of endorsement; Subsequent results; Time worked together. Ultimately forming: Graph-based Reputation Ranking. This is similar to: Google PageRank. ──────────────── Forty-five, PageRank solved the question of "which webpage is trustworthy" back in the day. Google's early greatest insight: A webpage saying it is the most important is meaningless. What really matters is: Who linked to it. High-quality websites linking to you: Indicates importance. Moreover: Links from high-quality websites, Carry more weight. This is PageRank. Cosign actually has a very similar possibility behind it: A person saying they are great: Is meaningless. The key is: Who cosigned you. High-reputation people cosigning: Carries more weight. If further considering: Whether endorsements are later validated by facts, It could even form: Human PageRank. ──────────────── Forty-six, if Human PageRank emerges, its economic value could be extremely huge. Because hiring, investing, and collaboration all require ranking. Which company is the most worth joining? Which engineer is the most worth poaching? Which founder is worth investing in? Which angel has the best judgment? These questions today mainly rely on: Human networks. In the future, if the Reputation Graph is complete enough, It can be algorithmized. This will form a very large: Decision Infrastructure. ──────────────── Forty-seven, but Human PageRank also carries a huge risk of elite entrenchment. We must remain clear-headed here. If: Only those with high-reputation attention Can gain reputation. Then those already in the core circle: Will find it increasingly easy to gain opportunities. Outsiders: Will find it increasingly difficult to enter. This is called: Preferential Attachment. A classic in network science. Those with connections: Gain more connections. In the end: The rich get richer. ──────────────── Forty-eight, so whether Cosign ultimately has value depends on whether it can continuously discover "strong outsiders." If the platform ultimately only: Recognizes people endorsed by Marc Andreessen; Recognizes each other among YC founders; Recognizes each other among Stanford alumni, Then it is merely: Digitizing the existing Elite Network. The value is limited. A truly great version should achieve: Allowing someone without Stanford, without YC, without San Francisco connections, to be discovered by credible people solely because their work is strong enough. a16z itself emphasizes in its discussion of Cosign that this game only works when "newcomers can still win." This is actually the key to success or failure. ──────────────── Forty-nine, a truly excellent talent network must do two contradictory things at the same time. First: Utilize existing reputations to enhance Signal. Second: Continuously introduce new talents without existing reputations. If only the first is done: It becomes a club. If only the second is done: The noise is too high. So the real difficulty for Cosign is: Exploration vs Exploitation. Utilizing known networks. While exploring unknown talents. This is very similar to portfolio theory. ──────────────── Fifty, this point is the essence of VC. Good VCs, if they only invest in: Former Google employees; Stanford; Former YC founders, The short-term risk is low. But the greatest Alpha usually comes from: People who the market has not fully recognized yet. So the core ability of venture capital has always been: Non-consensus Insight. Non-consensus judgment. If everyone knows: This founder is great, The valuation is already high. Real wealth is generated: Before consensus forms. Cosign's "Person to Watch" is actually: Making non-consensus judgments public. ──────────────── Fifty-one, this can produce a very interesting "reputation yield curve." Early endorsements: High risk. But after success: High reputation returns. It is already when Elon Musk is mentioned: "Elon is great." Reputation yield: Close to zero. But in 1995, saying: "This young person is worth watching." If they succeed later: The signal is extremely strong. So: Reputation Return ∝ Earliness × Accuracy. The earlier; The more accurate; The more valuable. This is "Bole Economics." ──────────────── Fifty-two, the true ability of a Bole is not to recognize successful people, but to recognize successful people before they succeed. Many people say: "I know many billionaires." It is meaningless. If the recognition occurs: After the other party has become famous. What is truly valuable is: Relationship Before Status. Why? Because after success: Everyone wants to know. Before success: Only those who truly understand are willing to invest time. So early relationships have: Deeper: Trust; Loyalty; Returns. This is: Pre-status Relationship Premium. ──────────────── Fifty-three, why is this logic particularly suitable for VC? Because venture capital is essentially: Buying the future. Not buying the present. The public market can buy: Already excellent companies. VC must find: Future excellent people who are not yet priced. So Cosign is not an ordinary media product for a16z. If it succeeds, It may become: Proprietary Deal Flow Infrastructure. This is also the business motivation we must see when analyzing it. ──────────────── Fifty-four, why does a16z do this itself instead of letting startups do it? This is a very thought-provoking question. a16z says: Cosign is a community, not a business. But even if it does not directly make money, It can still bring huge value to a16z. Because it may help discover: Founders; Engineers; Angels; Products; Talent flow. This means: Information Edge. For VCs, Information advantage itself is an asset. ──────────────── Fifty-five, a16z is gradually evolving from a "fund" to a "startup ecosystem infrastructure company." In the past, VCs: Raised money; Invested; Board of Directors. Today a16z: Media; Recruitment; Talent Network; GTM; Policy; Content; Events; Products. It is building: Founder Operating System. a16z emphasized in its 2026 summary of New Media that one of the purposes of its network, software, and AI is to help Portfolio Founders find talent and key connections more efficiently. This is already very different from traditional VCs. ──────────────── 56. Why do VCs increasingly need to build platforms? Because capital is becoming more of a commodity. Today, excellent Founders can get money from: Sequoia; a16z; Lightspeed; Accel; General Catalyst; sovereign funds; Family Offices. The differences in capital itself are becoming smaller. Thus, VCs must provide: Brand; Clients; Talent; Media; Recruitment; Policy. This is: Venture Capital Productization. Venture capital is beginning to be productized. Cosign belongs to this logic. ──────────────── 57. The real moat for future VCs may not be AUM, but Network Liquidity. A fund has: $100B. That does not mean it is the best. If a fund can: Find a top VP Sales in 48 hours; Introduce 10 Fortune 500 clients; Find the next round of investors; Help with media dissemination, it is more valuable to Founders. So in the future, funds will truly compete on: Network Liquidity. How quickly can people and resources in the network be mobilized? Cosign can be understood as: Gradually software-izing a part of a16z's private network. ──────────────── 58. This is actually a very big trend: service organizations are being software-ized. In the past, headhunters: Remembered in their minds: Who is capable. VCs: Partners remembered in their minds: Who is worth meeting. PR: Journalist relationships existed in employee directories. After the emergence of AI + Database: These invisible networks began to be: Structured; Searchable; Automated. This is called: Institutional Memory Digitization. Whoever completes this first, turns the ability that relied on a few old employees into an organizational asset. ──────────────── 59. What impact might Cosign have on the recruitment industry? Traditional recruitment model: JD → Applications → Recruiter Screening → Interview. In the future, top positions may become: Reputation Graph → Shortlist → Mutual Intent → Conversation. In other words: Search replaces Applications. Truly excellent people no longer apply. Companies discover them proactively. This has already happened in the high-end talent market. AI will only accelerate this further. ──────────────── 60. In the future, "submitting a resume" may gradually become a low-end recruitment behavior. This statement sounds extreme. But the high-end talent market is already like this. Top CEOs do not submit resumes. Top AI Researchers rarely mass apply. Top Founders also do not apply for: "Entrepreneurial positions." Real senior opportunities come from: Warm Intros; Reputation; Inbound. If products like Cosign succeed, it simply software-izes a model that originally served only 1% of people. ──────────────── 61. This means that the key to future career competition will shift from "application ability" to "discovery ability." In the past, everyone learned: How to write a Resume. In the future, what may be more important is: How to establish: Proof of Work; Public works; Peer reputation; Real relationships. Because AI can help everyone write resumes to a score of 95. When all resumes look good: Resumes lose their distinctiveness. This is: Credential Inflation. ──────────────── 62. AI will create a new "resume inflation." After universities expanded enrollment: Degrees have become more numerous. Thus, master's degrees are becoming increasingly important. This is called: Degree Inflation. After AI: Everyone's resume: Professional; Perfect; Customized. Thus: Beautiful resumes will no longer be valuable. The market must seek rarer signals. Thus: Real works; Peer endorsements; Verifiable results will increase in value. This is: Signal Migration. ──────────────── 63. One of the most important personal assets in the future will be Proof of Work. Do not just say: "I am good at product management." Show: What you have done. Do not say: "I understand AI." Show: Models; Code; Papers; Products. Do not say: "I am an excellent Founder." Show: Users; Growth; Team. The strongest professional identity in the AI era is not: Self-description. But: Evidence. Then: Endorsed by those who truly understand the evidence. This is how: Reputation is formed. ──────────────── 64. This is also why young people should build in public in the future. In the past: First enter prestigious schools; Then enter large companies; Then accumulate credentials. Today: The internet provides another path. Writing; Open source; Products; Videos; Research; Entrepreneurship. A 19-year-old can: Directly showcase their abilities to the world. This is equivalent to: Bypassing traditional credential gatekeepers. a16z also believes when introducing Cosign that in the past, people gained identity through these "rites of passage" like universities and prestigious companies, but now more and more people are accumulating professional status through continuous public output. ──────────────── 65. But the next stage of building in public is not "producing more content," but "being seen by credible people." The first stage of the internet: Everyone can publish. Thus: Content Explosion. The real problem in the second stage: Who sees it? Therefore: Distribution. The third stage: Who believes it? So: Reputation. This is why: Follower Count is ultimately not enough. 1 million random fans may not be as valuable as: 10 truly top peers publicly recognizing you. ──────────────── 66. Influence and reputation are two completely different forms of capital. Influence: How many people know you. Reputation: How knowledgeable people evaluate you. A TikTok Creator: 10 million fans. May have: Huge influence. A top chip designer: Only has 3,000 fans. But: Top engineers from NVIDIA, AMD, and Apple respect him. The professional value may be completely different. Cosign attempts to capture the second type: Domain-specific Reputation. ──────────────── 67. This is also why there should not be a unified "social credit score" in the future. Reputation must be: Contextual. A person may be: A world-class programmer. But not: An excellent manager. An excellent founder. But not: An excellent investor. So a truly advanced Reputation Graph must know: In which matters you are trustworthy? Not simply: This person scores 97. Otherwise, it will be severely distorted. ──────────────── 68. In the future, career networks will increasingly resemble financial markets: different dimensions will have different ratings. Bonds have: Credit Ratings. Stocks have: Analyst Ratings. Talent may also form: Engineering Reputation; Founder Reputation; Manager Reputation; Investor Reputation; Designer Reputation. Each dimension: Different networks provide judgments. This is much more informative than traditional: Job Title. ──────────────── 69. Cosign's biggest long-term opportunity may not be recruitment, but the "career capital market." Once the platform has: People; Reputation; Intent; Companies; Funding; Jobs; Investors, it begins to approach: Career Capital Market. Talent: Is an asset. Companies: Are capital demanders. VCs: Are financial capital. Offers: Are prices. Cosign can connect: Human capital with financial capital. This is much larger than recruitment websites. ──────────────── 70. In the future, there may even be a "Human Capital Index." For example: In the past five years, the most noteworthy: AI Engineers; Robotics Founders; Biotech Scientists. Who discovered them first? Who endorsed them most accurately? Which companies continue to produce top talent? This kind of data can ultimately form: Industry Intelligence. For instance: Which companies are experiencing accelerated talent loss? Which teams are seeing net talent inflow? Which labs are becoming new talent centers? This is extremely valuable for investors. ──────────────── 71. Talent mobility itself is one of the leading industry indicators. Where capital flows: Indicates where investors are optimistic. Where top talent flows Often earlier. If the best AI Researchers: A large number leave Big Tech; Go to certain types of Robotics Startups, This may mean: Industry opportunities are forming there. So Talent Flow is: A Leading Indicator. If investors can see in advance: The talent migration map, They may discover trends earlier than financial reports. ──────────────── 72. The truly top VCs have been secretly observing "who follows whom" for a long time. When a Founder leaves to start a business. Who is willing to follow him? Very important. If: 6 out of the 10 best people from the previous team, Follow him, This is a strong Signal. Why? Because these people possess: Insider information. They know the Founder’s true abilities. Willing to give up their salary to follow, Is equivalent to: Insider Capital Commitment. This may be more reliable than the judgment of investors. If Cosign can capture this kind of data, The value is immense. ──────────────── 73. Therefore, "who is willing to work with you again" is likely more important than performance evaluations. Performance evaluations: Written by the boss. May include: Politics; Systems; HR. But colleagues are willing to: Work with you again, Is betting on their own future careers. This is: Revealed Preference. Economics believes: Looking at what a person says, Is not as good as looking at: What they actually do. If they are willing to start a business with you again, This is the most authentic evaluation. ──────────────── 74. The biggest barrier for Cosign may ultimately be Network Effects. If there are only 1,000 people: It has no value. If the entire core ecosystem of Silicon Valley is present: Founders; Engineers; VCs; Operators. Then: With each new user, The value to others increases. This is called: Network Effect. And this is an extremely difficult network effect to break. Because once professional reputation accumulates over the years: The cost of migration is very high. ──────────────── 75. This is also why LinkedIn has been nearly impossible to truly challenge for twenty years. Everyone complains about LinkedIn. But they are still there. Because: Your colleagues; Recruiters; Companies; Resumes Are all there. The network effect is too strong. a16z itself acknowledged in its interview with Cosign that LinkedIn is one of the most difficult iconic professional products to challenge in the past twenty years. So Cosign wants to directly defeat LinkedIn: Very difficult. A more realistic route is: To first occupy high-value sub-markets that LinkedIn is not good at. That is: Startup Power Users. ──────────────── 76. This is also the most classic way to break through in entrepreneurship: do not fight in the main battlefield of giants. LinkedIn serves: Global white-collar workers. If Cosign starts by serving: 2 billion professional users, It is doomed to fail. But if it only serves: Founders; VCs; Early employees; AI Builders; Startup Operators. The user base is small. But: The density is high; The transaction value is high; The relationships are strong. This is: Narrow Wedge. Entering through an extremely narrow opening. If it becomes the default identity layer for this group, Then expand outward. ──────────────── 77. AngelList once proved that this route is possible. AngelList was not initially: A financial platform for everyone. It first became: A Startup + Investor Graph. a16z team also mentioned that early on, AngelList had a very important investor network, but later the strategic focus changed. Cosign is clearly trying to reclaim: This entrepreneurial relationship map. Only: Expanding from Money Graph To: Reputation Graph. ──────────────── 78. What should entrepreneurs really learn from Cosign? First: Do not treat your resume as the core of your career assets anymore. The real asset is: The evaluations of others after collaborating with you. ──────────────── Second: Proactively accumulate Proof of Work. Let others see what you: Have really done. ──────────────── Third: Seek truly high-quality collaborations, rather than blindly expanding your network. Completing difficult projects with 5 top talents May be more valuable than attending 100 Networking Events. ──────────────── Fourth: Help outstanding people as early as possible. The people you help today, May be the future: Clients; Investors; Partners. ──────────────── Fifth: Build a "portable reputation." Do not let all your value only exist in: The Title of a certain company. When you leave the company, Does the market still know who you are? This is key. ──────────────── 79. What should young people really learn? Do not ask: "How to meet big shots?" This is the most basic question. You should ask: How can I be strong enough for truly knowledgeable people to willingly endorse me? The best Networking: Is not exchanging business cards. But: Do Great Work. When the work is strong enough: Relationships naturally emerge. This is the most sustainable networking strategy. ──────────────── 80. What should investors really learn? Do not just look at: Founder Pitch. Look at: Who is willing to follow him. Do not just look at: Educational background. Look at: Whether truly capable people respect him. Do not just look at: Today's Title. Look at: The speed of talent growth. Do not just study: Consensus Stars. Look for: Pre-consensus Talent. Because the greatest investment returns always occur: Before the market re-prices. ──────────────── 81. What should companies really learn? The future recruitment system should not just optimize: Applicant Tracking System. Because Applications themselves may become increasingly worthless. Companies should build: Talent Intelligence System. Including: Past collaboration relationships; Talent networks; Recommendations from outstanding employees; Potential candidates; Private intentions; Future recruitable talents. Recruitment should upgrade from: One-time job openings To: Long-term talent asset management. ──────────────── 82. The biggest change in HR in the AI era may be: recruitment shifting from Transaction to Portfolio Management. Today: There are job openings. Start looking for people. In the future: Companies should maintain long-term: 100 potential recruits. Update quarterly: Who is doing what; Who might leave; Who’s abilities are improving. Manage: Talent Pipeline Like VCs manage: Pipeline. This is: Talent Portfolio. Cosign's Private Intent is precisely in this direction. ──────────────── 83. The real competition for high-level talent is not "who has the highest salary," but who establishes relationships earlier. When a top engineer publicly announces: Their departure. The competition is already too late. OpenAI; Anthropic; Meta; Google Will all rush to recruit. The truly smart teams: Knew him two years ago. Provided help. Built trust. When he is ready to move: Naturally the first call. This is: Pre-positioning. The capital market is like this. The talent market is the same. ──────────────── 84. The greatest theoretical value of Cosign can be condensed into a formula. In the future, a person's career value is increasingly likely to come from: Capability × Proof × Reputation × Network × Timing Ability. Evidence. Reputation. Network. Timing. Educational background is just one weak variable. Resumes are merely a display layer. What truly determines career compounding is: How these variables accumulate over the long term. ──────────────── 85. This is actually a new form of "human capital compounding." Assuming you: At 23, do an excellent project. Are seen by two impressive people. They recommend you. You thus enter a better team. And work with even more outstanding people. The next time you start a business: It’s easier to raise funds. It’s easier to recruit. Thus: Good Work → Reputation → Opportunity → Better Work → More Reputation. This is: Career Flywheel. Cosign aims to make this flywheel: More visible; Faster. ──────────────── 86. But what really needs to be vigilant about is: the reputation network may also become a new machine for class solidification. Any Credential System has two sides. Prestigious schools: Filter talent. Also may solidify resources. VC Networks: Increase efficiency. Also may form cliques. Cosign: Discovers talent. Also may amplify the judgments of existing celebrities. If the algorithm continues to reward: Those who already have many strong endorsements, Ultimately: Newcomers become increasingly difficult to see. So the platform must design: Discovery Mechanisms. Specifically to dig: People who do not yet have networks. Otherwise: Reputation Graph Ultimately becomes: Status Graph. ──────────────── 87. Reputation and status are not the same thing. Status: Others know you have status. Reputation: Others know you are reliable in a certain matter. A person can: Status high. But actual ability declines. Another person: No one knows. But extremely strong. A truly great talent network must constantly accomplish: Status Correction. Realigning reputation with actual ability. If this cannot be done: The whole system will eventually become: Celebrities praising each other. Then Cosign becomes meaningless. ──────────────── Eighty-eight, therefore the most important product metric for Cosign may not be user growth, but "Prediction Accuracy" Assuming today: 1,000 Persons to Watch. Five years later: How many actually become: Top Builders; Founders; Leaders? What is the hit rate of different recommenders? If the platform can quantify: Reputation Prediction Accuracy It will have extremely powerful exclusive data. This data may even be worth dozens of times more than: Follower Count. ──────────────── Eighty-nine, a brand new "Talent Alpha" may emerge in the future Financial Alpha: Outperforming the market. Talent Alpha: Finding undervalued people in the market. A company's competitive advantage largely comes from: Whether it can hire someone worth: $200K for an actual value of: $1M. A VC's competitive advantage comes from: Whether it can invest in a future: $100B Founder at a: $10M valuation. Both are essentially the same: Find Mispriced Human Capital. ──────────────── Ninety, this is also why "talent identification" may be one of the most valuable human abilities in the AI era AI will increasingly be able to: Write code; Analyze data; Write reports; Design. But: Who is worth trusting? Who has true originality? Who can lead a team? Who can continue to grow over ten years? These judgments are very difficult to quantify completely. So human judgment may not disappear. Instead, it will be amplified. AI provides: Unlimited Candidate Supply. Humans provide: Conviction. This may become a new division of labor in the future labor market. ──────────────── Ninety-one, the real big proposition behind Cosign: as abilities become cheaper, reputation will become more expensive Industrial Revolution: Machines made physical labor cheap. Internet: Made information distribution cheap. AI: Made knowledge work and content generation cheap. Every time something becomes cheap, another scarce resource becomes expensive. After AI, the most scarce may become: Judgment; Taste; Trust; Reputation. Because: Machines can generate ten thousand answers. But: Who is worth trusting will not be automatically resolved. ──────────────── Ninety-two, this is also why AI does not necessarily weaken interpersonal relationships, but may instead increase the value of high-quality relationships Many people think: AI → Everyone relies on Agents. Interpersonal relationships are no longer important. It may be exactly the opposite. When the internet is filled with: AI emails; AI content; AI applications; AI reviews, a genuine introduction from: a trustworthy real person will increase in value. So: Digital Noise ↑ → Trusted Human Network Premium ↑ This is a very strong counteraction. ──────────────── Ninety-three, the most valuable message in the future may not be "AI recommendations," but rather someone you truly respect saying: meet this person This has been a mechanism in Silicon Valley. Founder: "You must meet this engineer." Investor: "You should know this young Founder right now." Professor: "This is the best student I have seen in ten years." Such a statement: can change a person's career. Cosign is just trying to turn these originally: Private; Instant; Non-searchable things into: Persistent Infrastructure. ──────────────── Ninety-four, if it ultimately succeeds, what might Cosign really achieve? Not: LinkedIn 2.0. But rather: Silicon Valley Reputation Layer. At the bottom are: Companies; Talents; Career histories. Above that are: Relationships. Further up: Conviction. Then: Intent. Ultimately occurring: Hiring; Funding; Co-founding; Acquisition. This is a complete: Trust-to-Transaction Funnel. From trust to transaction. ──────────────── Ninety-five, this is also why this matter is ultimately very close to financial markets What do financial markets do? Price Discovery. What Cosign may do is: Talent Discovery. Financial markets: Capital flows to high expected return assets. Cosign: Opportunities flow to high expected talent. Price: Stock valuation. The "price" in the career world is: Salary; Equity; Funding; Title; Opportunity. So in a real sense: Cosign aims to improve: Human Capital Market Efficiency. ──────────────── Ninety-six, but improving market efficiency will bring another result: truly outstanding individuals may become increasingly expensive If in the past a hidden engineer: was unknown. The company could: hire cheaply. Cosign makes everyone aware: He is extremely strong. As a result: Offers increase. Salaries rise. Startup financing becomes easier. This means: The platform enhances: Talent price discovery efficiency. For talent: Very good. For employers: Not necessarily cheap. But from the perspective of the entire economy: Resources flow faster to high productivity individuals. ──────────────── Ninety-seven, this may accelerate the "superstar economy" The AI era may inherently amplify top talent. Plus a more efficient reputation network, the very top: Engineers; Founders; Researchers may increasingly find it easier to obtain: Capital; Talent; Opportunities. Thus: Winners rise even faster. This will improve: Innovation efficiency. It may also increase: Income and wealth concentration. This is the real deep political economy issue of systems like Cosign in the future. ──────────────── Ninety-eight, the most important inspiration ordinary people should get from Cosign is not to register immediately What really needs to change is: Career thinking. In the past: What is my company? What is my title? In the future, more importantly: If I leave the company tomorrow, who is willing to publicly vouch for my abilities? This is an extremely harsh but valuable question. If a person works for ten years, with a high title, but no excellent former colleagues are willing to: work with him again. Then his: Career Capital may be far lower than it appears. Conversely: A common title. But the best people are willing to work with him again. This may be an undervalued talent. ──────────────── Ninety-nine, the real career asset is not the title, but "what others are willing to risk for you" Someone is willing to: Introduce clients to you. Someone is willing to: leave their job to start a business with you. Someone is willing to: invest in you. Someone is willing to: hire you for key business. Someone is willing to: recommend you with their own reputation. These are the true: Reputation-backed Capital. Titles can be inflated. This kind of capital is hard to counterfeit. ──────────────── One hundred, the most worthy long-term observation of Cosign is not the number of users, but whether it can become a "non-consensus talent machine" If it ultimately only: organizes already famous people, its value is limited. The truly huge version is: Today discovering: an unknown engineer. Two years later: Founder. Five years later: Unicorn. Then historical records show: Who discovered first; Who believed first; Who helped first. At that time: Cosign records not just: Talent. But: How innovation networks are formed. ──────────────── The most memorable sentence a16z doing Cosign superficially is redesigning: Career social networks. But the real change is much greater than that. The first stage of the internet solved: Who are you? LinkedIn answered with: Education, company, title. The social media era solved: What do you think? X answered with: Content, followers, interactions. And the increasingly important question in the AI era will become: Who actually believes in you — and how much are they willing to stake on that belief? Who truly believes in you? How strong is that belief? Are they willing to: Work with you again; Hire you; Invest in you; Stake their own reputation on you? This is: Conviction Graph. When AI can infinitely produce: Beautiful resumes; Perfect pitches; Customized emails; Code; Articles; Works, "What I say I am" will become increasingly cheap. What will become truly scarce is: After others have seen your real work, they are still willing to say — this person, I am willing to bet on again. This is reputation capital. And from the perspective of capital markets, the most worthy aspect of Cosign to focus on is not whether it can defeat LinkedIn. But whether it can establish a market that has never truly been digitized before: Price discovery for people themselves. If it can do this, The most important outcome is not just an additional career social app. But rather a core kind of Alpha in Silicon Valley — Discovering the most valuable people of the future before everyone else. The first time it may transform from the private contact lists, WeChat groups, DM, and memories of a few people, into a set of: searchable, accumulable, verifiable, and capable of compounding digital infrastructure. That is where Cosign truly deserves to be studied.