|
In this issue: - The Economics of Logos—When you visit the homepage of a SaaS company or the portfolio page of a VC's website, you're looking at a reputational credit market complete with booms, busts, and sharp operators.
- Elsewhere in Reputation Markets—Referral bonuses make little sense when a company is small and everyone has material equity, a lot of sense as it grows, and then stop working once referrals get easier to exploit.
- Knocking Down Data Walls—The growth of AI raises the value of unfinished drafts.
- Closing the Arbitrage—Putting some of a public company's treasury into crypto tends to help their stock price. But anyone can do it.
- Cost Avoidance—AI hasn't directly induced many layoffs, but it has eliminated countless counterfactual jobs.
- State AI Investment—The comparatively low cost of being the best place to do AI research that doesn't directly lead to a profitable business.
This issue of The Diff is brought to you by our sponsor, Intercom. The Economics of Logos
One of the recurring questions in The Diff is: what is money? We talked last week about credentials and moneyness, and a while earlier we explored why companies issue alternative currencies, whether they're in-game coins or loyalty points and how being able to issue currency sets some countries apart from the rest. But there's another quasi-currency that's also worth exploring: logos. Visit the homepage of a B2B company, and you'll often see a parade of logos. Visit a VC's site, and you'll see logos, too.
What you're seeing is the surface of a unique form of intracorporate lending, where companies let other companies borrow a bit of their brand in exchange for some exposure (or, in some cases, in exchange for knocking down the price of a product a bit) It's often a truncated selection, since some of the most prestigious companies are big enough to have an explicit policy that makes them less accepting of random startups asking to use their brand name. And some of the most interesting companies out there stay interesting by being secretive—the fewer reminders there are that they exist, the fewer competitors they have to worry about.
Valuable logos have extreme returns to scale because they have overlapping feedback loops:
- Sales pages are written on the assumption that the reader has a short attention span and the most valuable information needs to be upfront. If a company has two case studies, one of which involves doing something really amazing for an unknown company or a stodgy Fortune 500, and the other involves something trivial that they did for Databricks or Anduril or Stripe, the higher-profile logo wins over the better story. (They can always tell that story in the first Zoom meeting with a prospect.)
- A company's brand gets more recognizable if it shows up in more places, so the more times a given company's logo is out there on suppliers' sites, the more name recognition and thus value that logo has.
- And, of course, this social proof helps companies win more details, expanding the scope of products they need and growing the set of suppliers whose homepage could host a logo.
Meanwhile, it wouldn't be a quasi-financial system without more purely-financial intermediaries. One of the rituals around fundraising is that when a company raises money from a fund, each one has a chance to put the other's logo on their site. This, too, has the usual returns to scale: Sequoia, Founders Fund, and a16z are very likely to get mentioned—if there's a next-generation fund in the round, even if the cognoscenti would rather have an LP stake in that fund than in Sequoia's vehicle, it doesn't have the cachet to make it onto a short list. So the system of logos tends to reinforce whoever's most powerful, in much the same way that bilateral trade between two countries tends to be denominated in the currency of whichever one has a more widely-accepted currency, or in the dollar.
The profusion of logos is a form of reputational credit expansion, with all the usual upsides and downsides thereof. If tiny company X does a deal with established company Y, X bragging about it is actually a way to make the market more efficient—it's a valuable signal if they've convinced a reputable company that their product is a) a better deal than established providers, and b) better to buy than to build. So a section of the economy that doesn't have some way to make this kind of reputation-enhancing win visible is one that doesn't have enough circulation of reputational credit.
But it's possible to get Minsky Moments here, too, where a company's ubiquity becomes a bigger part of its growth. Any time there's a way to create credit, there will be people who engineer ways to create too much of it. Something similar happened in a more legible way with the link graph: when Google launched, they could use hyperlinks as an incredibly convenient tool for measuring reputation—if lots of people link to irs.gov, and those links tend to use words like "Tax," you know that irs.gov is a great place for information about taxes. But that created an ecosystem where other companies were also incentivized to earn those links, or even to buy them. In that case, Google had more or less privatized the link graph, and could set up policies that mitigated some of the most annoying abuses while still leaving room for some hustle. But nobody really owns the concept of startup reputation (Y Combinator might come the closest, and they certainly have some visibility into how logos end up in the places they do).
In the meantime, we're in an era of reputational free banking: anyone who can spend a few minutes in Photoshop, a few dollars on Upwork, or a few seconds with DALL-E can enter the reputational banking business and bootstrap from there. But as with other financial systems, this is great when it accelerates productive investment but vulnerable to overshooting.
 In 2025, every company is going to recognize the need for a customer service AI agent. Why? Because AI agents provide customers with better answers and faster resolutions, all at a lower cost to you. Fin by Intercom is the world’s leading AI agent for customer service. Fin can resolve up to 80% of questions in an instant—which customers love. And you can lower your cost to serve by 96% and finally get off the service rep hiring treadmill. The best part is that Fin is now available on every customer service helpdesk. That means you get all of Fin’s benefits without having to change platforms. See why more than 3,700 businesses have hired Fin for AI-first customer service. Elsewhere
Elsewhere in Reputation Markets
A decent model of a growth company is that everything it does matters only insofar as it lets the company get the right talent. Most of the value created in percentage terms happens early on, but in cumulative dollar terms it's mostly the result of the effort and judgment of people who were hired at some point after launch. (Some early career paths don't seem to make much economic sense for the company unless you see them as a way to identify and retain a small cohort of top talents.) So companies naturally care a great deal about hiring, and try to use various hacks to get it right, one of which is that they value referrals more highly than cold inbounds. Referrals combine some broadly useful correlations, like people meeting and hanging out with peers who share similar interests, and referrals also involve existing employees betting a bit of their reputation on new hires. For people who don't value that reputation especially highly, there are online marketplaces where people swap referrals in exchange for kickbacks. Part of how companies grow is that they establish norms when they're small, scrappy, and everyone knows everyone else. And, over time, they scale to the point where those norms just don't make sense any more.
Knocking Down Data Walls
Text and code are great media for training AI models because they're both ways to think out loud. It's hard to be sure you understand a topic well until you can write someone an email explaining some aspect of it, and codifying a process in terms a computer can understand is also a great way to ensure that you yourself understand it. In other media, there's a sharper distinction between process and output—which means that more of the data AIs train on shows the end result disconnected from how it happened. Which is one reason AI companies are paying streamers for outtakes and early cuts.
If you're worried about AI taking your job, one way to reframe that is that the output of your work contains training data, and 1) being able to produce unique training data is a good idea, and 2) that specific job is going to go away eventually. Getting paid a bit more to make it happen a little faster is a bargain that someone will probably take, and at least in this case the streamers seem willing to take it.
Closing the Arbitrage
The NYT has a good piece about public companies putting some of their treasury into crypto, sometimes with the explicit aim of attracting investors who want crypto exposure but have a mandate that doesn't allow it. Some fraction of the financial industry is always devoted to finding customers who say "I would like to do X but am only allowed to do Y; can you sell me a Y that is functionally X?" and the industry is too happy to oblige. Does someone need to own equities but want a bond-like payoff? Writing covered call options gets you part of the way there, with steadier-but-capped upside. If you want to own equities in a jurisdiction that limits local ownership, you may end up owning swaps instead (and depending on how much demand there is, you may be paying a premium to get access at all). In this case, it doesn't require any special expertise to switch from being a cash-rich public company to being a crypto play. But that also means there's more competition for the attention of the same set of investors, and that these investors are seeing a rising supply that may make the premium they paid regrettable.
Cost Avoidance
Layoffs are a surprisingly expensive choice, and one reason they're so severe and so clustered is that companies have incentives to delay them. The net effect of a round of layoffs is 1) of course, the direct effect of getting rid of some headcount and associated costs, 2) the loss of other people who quit because they want more job security, think the firm laid off the wrong people, etc., and 3) the ongoing productivity hit from reorganizing processes around the remaining employees and the inevitable morale hit. So one effect of labor-saving technologies like AI is that it leads to job loss through omission—companies are creating fewer new roles rather than eliminating existing ones ($, WSJ). This is, at last, a case where the economics of AI are more favorable for smaller companies than bigger ones: a smaller business is adding proportionately more headcount, so it's getting bigger savings. A large company might counterfactually want to employ one third as many people, but the only way to get there is to grow the scope of their activities 50% by holding headcount constant.
State AI Investment
The UK plans to invest in domestic AI capacity in response to a report advocating a state-owned 100k-GPU cluster ($, FT). From the article: "The new capacity, which would represent a 20-fold increase in the UK’s sovereign computing power, will be separate to privately owned AI data centres and will be deployed by the government primarily for AI applications in academia and public services." It's very hard to compete with the US-based labs and larger companies on private sector capacity, but it is feasible for a government of the UK's size to have the most government-owned compute. So what they can actually do is capture some of the AI-adjacent academic talent when they're deciding where to do a PhD, and that gives them a few years of lead time to find policy moves that lead to local jobs once those PhDs finish up.
Diff JobsCompanies in the Diff network are actively looking for talent. See a sampling of current open roles below: - A premier proprietary trading firm is looking for smart generalists to join their investor relations team, working with external investors, rating agencies, and the internal finance team. Investment banking and/or investor relations experience preferred. Quantitative background and technical aptitude a plus. (NYC)
- A growing pod at a multi-manager platform is looking for new quantitative researchers, no prior finance experience necessary, 250k+ (NYC)
- A well-funded startup that’s building the universal electronic cash system by taking stablecoin adoption from edge cases to the mainstream is looking for a senior full-stack engineer. Experience with information dense front-ends is a strong plus. (Singapore)
- A Google Ventures backed startup founded by SpaceX engineers that’s building data infrastructure and tooling for hardware companies is looking for a staff product manager with 5+ years experience, ideally with AI and data intensive products. (LA, Hybrid)
- YC-backed AI company that’s turning body cam footage into complete police reports is looking for a tech lead/CTO who can build scalable backend systems and maintain best practices for the engineering org. (SF)
Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up. If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.
|
