Disclosure: I am an investor in Friend.
People are upset about Friend. (The Friend is a $129, no-subscription pendant that listens attentively, and that then converses with you about your day.) People really hate AI, and they hate “surveillance capitalism”, whatever that means, and so they really, really hate the AI device that surveils. I can’t really argue with this; popular beliefs like a resolute hatred of AI or a selective disdain for “surveillance” is more akin to a religious dogma than any kind of empirically defensible belief. There’s no use arguing with these people, any more than you can dissuade a Christian of Jesus’ divinity.
But really, people are upset about the Friend founder Avi Schiffman’s deliberately provocative marketing campaign. Apparently, it’s the largest NYC subway campaign ever. It’s a massive print campaign spanning several cities. Some people in NYC are already defacing the posters in an attempt to make some kind of banksyesque statement about AI and surveillance and so on.
Avi is no stranger to provocative marketing. The launch video which depicted people going about their lives while they conversed with their Friend devices garnered 25 million views, much of it due to people performatively professing their disdain for the concept. The purchase of the presumably very expensive Friend.com domain also caused a stir on tech twitter. Avi has developed an eager reputation as an iconoclast when it comes to startup conventional wisdom.
I think AI wearables are fascinating. I have already written at length about my conviction that they will be a significant consumer hardware category in the coming years. People who fear or hate AI wearables will be condemned to the pessimist’s archive, alongside people that thought no one would ever use a self-driving car, wear a biometric ring, or use a smart watch. In the space of about a year, people went from having no thoughts whatsoever about AI, to routinely using AI chatbots as full time therapists and confidants. Why wouldn’t we take the next logical step and introduce our AI companions directly into our lives rather than having to manually relate to them the issues of the day?
Amid the backlash, a lot of people have said things like “how can I short this company” and “investors must hate him” after seeing the monster ad buy. Well, here I am, someone who invests for a living, explaining my investment and how I think about it.
I met Avi in 2024 on the advice of a good friend who knew I was interested in this stuff. Within about 15 minutes, I wrote him a $50k check. In my relatively short career as an angel investor (dating back to 2019), I’ve personally deployed over $2m into 53 different companies. (Keep in mind this is completely distinct from my firm Castle Island, where we have invested $100s of millions into crypto companies.) If you thought of my angel portfolio as a standalone fund, it would be top 1% in its vintage by DPI, TVPI, MOIC or any other VC metric (frankly, more like top 0.1% but I don’t have the numbers).
While a plurality of my angel deals are in crypto (startups), because that’s where I have best access, I have also done a lot in the food and beverage space, healthtech, consumer, and AI. My considerations for angel deals are completely different from my work at Castle Island. First, I do try to make money, so I don’t just write checks randomly. Second, I like to back my friends and support them. And third, I like to invest in concepts that a) I want to exist in the world and b) are interesting to talk about. That’s how I ended up investing in fish-killing robots, an app that helps you avoid seed oils, at-home blood testing devices, and an AI-powered investment bank.
A few people asked me about my investment in Friend given all the drama and I respond unhesitatingly that I’m glad I did it, and I would do it again in a heartbeat.
“Even with the big domain purchase and the ad buy?” To which my answer is “Yes, in fact, because of those things.”
Spending huge amounts of money on ad buys when you are still an early stage startup is obviously very risky. But I want my angel deals to take on the absolute maximum amount of risk possible. Let me explain.
The first reason is that, when you combine early-stage investments into a portfolio, you want your startups individually to be as risky as possible. This is not the case with all portfolios. For instance, if you are using leverage and trading liquid assets that you have to mark to market, you have to manage your risk or you will be liquidated. And if you have external LPs and the possibility of redemption (as with hedge funds), you might be forced to sell during a drawdown, so you end up unable to ride out a storm. However, in closed-end venture structures, and obviously with angel investing, you don’t have redemption, and you don’t have leverage, so there’s no risk of liquidation. It’s specifically this structure which allows VCs to take on outsized risk, and this is why it’s an attractive asset class. The freedom that VCs enjoy because LPs can’t redeem is why they can invest in counterintuitive or even crazy-sounding ideas.
And psychologically, because angel investments aren’t marked to market, you’re never going to get shaken out of your position. Most people don’t have the mental ability to tolerate massive portfolio downturns without being tempted to sell, but in startups you are generally locked up. If you marked to market some of our early stage companies they would be eye-wateringly volatile. The illiquidity is a feature, not a bug.
The second reason why you should back founders who do counterintuitive things is because that’s how you actually make money. If you had a rule that you were only willing to invest in “derisked” ideas, you’d only be investing at the Series B or C stage, when the concept was starting to work, revenue was in the door, the company was showing signs of profitability, and so on. But that involves paying multiples more. And if you were only willing to back previously exited founders (who naturally have a much higher probability of success), you’d be paying a much higher ticket price.
This is why people talk about non-consensus investing. Consensus ideas are expensive. When I first backed CoreWeave it was a GPU farm mining ETH, right as ETH was about was deprecate Proof of Work. Basically, their only client was about to disappear. But they had this wild notion (keep in mind this was 2018/9) that AI would become a thing, and that they’d be able to monetize those GPUs in new ways. No one except for them believed that at the time. Today the company is worth $60b.
Paul Graham has an excellent blog post about this entitled Black Swan Farming. At the time, a lot of future good ideas are indistinguishable from bad ideas. He says:
That’s made harder by the fact that the best startup ideas seem at first like bad ideas. I’ve written about this before: if a good idea were obviously good, someone else would already have done it. So the most successful founders tend to work on ideas that few beside them realize are good. Which is not that far from a description of insanity, till you reach the point where you see results.
This is why the failure rate you should target in your portfolio is not zero. If it’s zero, you are not taking on enough risk, and you are probably forsaking upside. Graham says:
If we ever got to the point where 100% of the startups we funded were able to raise money after Demo Day, it would almost certainly mean we were being too conservative.
In my own angel portfolio, I have a 9% failure rate, a 39% up round rate, with a 2% IPO rate (the average maturity of these checks is still just a couple years so these numbers will change a lot). I don’t have a strong intuition for what the failure rate should be, but I know it’s not 0.
Now does this mean that it’s always good when founders make absolutely unhinged decisions? No. But am I A-OK with Avi doing unconventional things like buying a flashy domain or doing a big ad buy? For sure.
The AI wearable space is still in its infancy, with probably only a few thousand adopters worldwide. That’s effectively 0% market penetration. In my view, the addressable audience for this is hundreds of millions of people, so the first challenge is simply making people aware that this kind of device exists.
And even though negative publicity isn’t normally seen as a good thing, with early-stage startups it’s asymmetric. Some percentage of people seeing the Friend discussion online will be curious and adopt the device, even if the messaging around it is negative. And Friend is indifferent to the negative perceptions from non-owners, since they are going to continue not buying the device. If you had the ability to reach 10 million people with a 95/5 hater/fan ratio, you’d rather do that than reach 10,000 fans with no backlash. I think Avi understands this perfectly well, which is why he has successfully taken advantage of outrage marketing on more than one occasion.
Any AI device is going to be inherently controversial because a significant fraction of the population is negatively disposed against AI. And any device that passively listens in public is going to be controversial by definition since it requires new intuitions around privacy. (Although as I explain, Friend doesn’t literally present you with a transcript of what has been said). So I don’t think you can actually market an AI wearable without significant backlash (see the reaction to the Humane pin). The options were: be meek and hope to find an early adopter wedge without upsetting anyone, or be loud, accept the negative publicity, and find your people that way. Avi was just the first one to do this at mass scale.
So if I’m choosing between a portfolio of Avi Schiffmanns doing unusual things or a portfolio of founders doing consensus paint-by-numbers startups, I’ll take the former every time.