a16zHow to Spot Exceptional Talent Before Everyone Else
a16z
The Problem Cosign Solves 0:00
The most important skill in this industry, and perhaps anywhere, is the ability to spot things before the broader market does. Whether you are an investor, a founder, or a hiring manager, the challenge is the same: identifying people, companies, or products of value before everyone else catches on. Silicon Valley runs on this race to discover talent early, since the next great person might come from an unknown school or be only eighteen years old. Andreessen Horowitz built a product called Cosign, meant to be the most comprehensive directory of startup people and companies, designed around the idea of professional reputation. The idea traces back to a story about David Prell saying his Twitter account was more valuable to his career than his college education, since college bundles education, network, credential, and social life, and much of that network and credential effect can now happen publicly online through peer endorsement.
Three Kinds of Endorsement 4:00
Cosign asks every profile to carry at least three signals. The first is who shaped your career, the mentor who spent real time on you, which tells others you are worth their attention too. The second is who you would work with anywhere, the people you have truly been in the trenches with, a much more specific bond than simply following someone online. The third is a person to watch, someone you believe will rise, a call made before the world notices. David mentioned running this kind of experiment seven years earlier, emailing a few hundred people asking for their person to watch, and one name that came back was Russell Kaplan, then a twenty-two-year-old engineer at Tesla, now president at Cognition. The same logic extends to companies, letting people express conviction the way an investor would with money, even if they are not investors themselves, a kind of fantasy investing or social capital investing.
A Newsletter and a Living Directory 7:30
David's twin sister Justine, who works on Andreessen Horowitz's infrastructure team, co-ran a newsletter with him for six and a half or seven years, aggregating startup job postings after they ran a startup incubator at Stanford and needed a way to tell legitimate founders from the rest. That newsletter built a real community and a network of campus scouts, some of whom later became founders the firm invested in. Cosign aims to scale that same instinct, citing Pace McCormack's list of thirty-seven companies in deep tech as an example of someone lending his reputation to point others toward good bets. Unlike paid databases such as Crunchbase or PitchBook, Cosign is meant to be free, entirely positive in tone, a place for people to display accomplishment with real, granular signal rather than vague LinkedIn connections nobody can actually vouch for.
The SV Angel Conference Example 13:01
You hear about a standout event, the SV Angel conference from around 2018, where organizers sent out a list of 150 notable attendees beforehand and let people swipe on who they wanted to meet. If both people swiped right, they got an automatic introduction with full context. That single event created enough shared context to turn strangers into long-term friends and collaborators. The larger point is that meeting the right people is mostly an information problem, not a desire problem, since so many people who want to meet each other simply lack a low-friction way to find that out. The product described here goes further than simple interest, letting people signal more specific professional intent, such as I would fund this person or I would hire them if they ever left their job, without the awkwardness or risk of a cold DM.
Codifying Trusted Recommendations 15:00
You already do a version of this informally, remembering a great designer or PR person and matching them to someone who needs help months later, but that memory is unreliable and slow. The idea is to codify these judgments, letting people mark who they consider truly great in a given field, so that trust and reputation can be looked up rather than recalled from memory. This includes both a private list, kept for personal use, and a public list that signals endorsement to others. It also solves a timing problem, since someone might quietly be open to a job change or an acquisition, but directly asking them is risky or awkward, whereas a passive, two-sided signal lets timing align naturally.
Feeds, Announcements, and Durable Reputation 18:30
The feed design centers on questions and announcements rather than free-form posts, aiming to make reputation durable instead of ephemeral the way social media praise usually is. When someone changes jobs or a company raises funding, people who worked with them often surface warm, specific endorsements that would otherwise vanish after a day online. The team pulls in genuinely positive replies and quote-tweets from platforms like X and attaches them permanently to a person's profile. The stated goal is celebration rather than critique, and extending recognition beyond founders to the designers, engineers, and marketers who also build companies, recalling the Rise Awards from around 2015 to 2016, which honored those often-overlooked contributors.
Making Reputation Visible and Liquid 25:31
The conversation turns to how many people who quietly made a company successful, like an early engineer who turns out to be crucial, never get public credit because their contribution is not something that can be easily publicized. The goal described is to make the reputation market more efficient and more liquid, meaning that people with genuinely great reputations become known for them, and that awareness then translates into real opportunities to be hired or funded. The idea extends to letting people who ask for job or hiring advice tap into a trusted list of companies or candidates, so a personal recommendation can reach many companies at once instead of staying locked in one person's head.
Frustrations With X and LinkedIn 29:01
Both speakers describe wanting tools that current platforms do not offer. On X, one speaker wishes they could see who liked a post about joining their company, or get clearer signals about who wants to work for or invest in them rather than just follow them. On LinkedIn, despite it being called an iconic, hugely valuable product, there is a shared sense that people use it constantly while still being frustrated by it, wishing for more than just a feed or a flat profile.
LinkedIn's Original Vision and Its Limits 30:00
Drawing on time at LinkedIn in 2004 and 2005, the speaker recalls founder Reed Hoffman's three-stage plan of growth, usage, and revenue, and early efforts to build a social resume that showed not just who a candidate was but who in your network could vouch for them, plus putting the hiring manager's profile directly on job listings. Over time, though, connections on LinkedIn stopped meaning strong trusted relationships and became loose, collected contacts, so a request for an introduction very often gets an unsure response about whether the connection is even real. The hope for cosign is to capture that missing granularity, the true context of how people know each other, at a moment when AI-driven outreach is producing thousands of applicants per post and human endorsement is becoming the scarce, valuable signal.
Why LinkedIn is hard to displace 38:31
You might love the right opportunity to find you, and that is what made LinkedIn's network effect so powerful, since it now reaches over a billion people, many of them what the speakers call mildly happily employed, meaning content but still open to something better. Attempts to build a LinkedIn killer struggle because switching costs are high, LinkedIn is very good at bot detection so it cannot easily be scraped, and careers depend on being where everyone else already is. Even so, there is room for something new where high quality endorsements, high conviction opinions, or niche needs are not being served, the way GitHub already works as a kind of better LinkedIn for engineers.
Cosign's inspiration and approach 43:00
Cosign is described as pulling together information that is already happening quietly across the tech industry, rather than trying to simply out-LinkedIn LinkedIn. The team says they are inspired by AngelList, which once served as a public social network for founders, investors, and talent before drifting away from that role, and by Wikipedia, both because people find profiles aspirational and because Cosign builds profiles for people using public information gathered with AI, avoiding the cold start problem of asking people to sign up first. Everything pulled into a profile comes from public sources, such as funding announcements, and the product is presented as unfinished, an invitation for people to edit, endorse, or add feedback to their own or others' profiles.
How cosigns build careers 46:00
Much of the industry runs on cosigns, the example given is Paul Graham naming Sam Altman among the five best entrepreneurs around 2009, before Altman had a major exit, which helped propel him toward running YC and later OpenAI. Other examples include Ryan Hoover bringing Eric into Product Hunt, and young figures like Dwarkesh, Michael Trull, and Aschenbrenner rising through early endorsement despite their age. One critique raised is that cosigning someone feels like giving away your own advantage, but the counterargument is that people remember who believed in them first, as when Brian Chesky publicly credited YC and Paul Graham when Airbnb went public, so being the earliest public believer tends to pay off over the long term.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
