47: - A Friend That Brings Us Closer

1. Key Themes

The Loneliness Crisis is Existential, Not Sentimental

Paul frames the global isolation trend not as a feel-good social problem but as a civilizational threat. He marshals hard data to make the case that this is urgent.

"The share of American men have like zero close friends went from like 3% to 15%... half of American adults report being lonely... feeling lonely is the equivalent to like smoking 15 cigarettes a day... a world that is like completely isolated is like a world in which humans go extinct." - Paul Scherer 00:49:20

"Seoul in Korea has like 0.5 fertility rate. What this means in practice is that every generation is 33% smaller than the last one, which means Seoul is like 10 generations away from being like not there anymore." - Paul Scherer 00:51:18

The "Mutual" Part Matters More Than the "Friend" Part

The counterintuitive core of Eigen's product philosophy is that the network — the shared context across all users — is the real innovation. The "friend" interface is just the most natural UI for that network.

"A lot of what we're thinking about and what we're creating starts actually counterintuitively, not with the friend part, but the mutual part... right now we're like four engineers and like three of them are working on the network piece." - Paul Scherer 00:35:03

"You are building a social network in a sense that is inside of this person we all know... We're building a person." - Paul Scherer 00:43:06

AI as Antisocial by Design — and the Alternative Incentive Structure

Paul's clearest contrarian thesis: most AI companion products are structurally incentivized to increase time spent alone with the AI. Eigen's design is the opposite — the product fails if it doesn't make users more social with each other.

"Zach went on a podcast and he literally said humans have a capacity for five friends, five close friends, but the average American only has two. We're going to build the other three. His best case is for you to spend your Friday night talking to this thing." - Paul Scherer 00:33:35

"We are a platform, but we're also a participant... If this mutual friend that we're building is a massive asshole, guess what? You're just not going to talk to him... so the self-serving incentive of this person is to be not an asshole." - Paul Scherer 00:54:04

2. Contrarian Perspectives

The Social/Media Conflation Destroyed the Internet's Promise

Most people assume social platforms got more social over time. Paul argues the opposite — they became media platforms optimized for popular content, not your actual social graph — and this happened not through malice but through pure incentive mechanics.

"If you scroll on Instagram, it's actually not about your friends, right? It's about popular content or popular people... they didn't do that for Instagram because they were being like, wouldn't it be cool if we were really evil. They're just like, well, that just works." - Paul Scherer 00:19:24

Personality is a More Flexible Interface Than Graphical UI

Paul makes a subtle but profound claim: a consistent personality can serve as a universal interface across use cases in a way that no graphical UI can. This is why ChatGPT — the only non-networked mainstream consumer product — is actually an outlier, and why Eigen's bet is to collapse many apps into one person.

"How much more or less flexible is personality as an interface than graphical interfaces?... Consumer networks are very unbundled right now... basically every single mainstream consumer piece of software that is on your phone right now is networked except for ChatGPT. It's like the only non-networked mainstream scale product." - Paul Scherer 00:39:41

Building a Great Product Is Just as Hard as It's Always Been

Against the prevailing Twitter narrative that AI has commoditized product building, Paul pushes back hard — and backs it up with a challenge to name a single transformative consumer product from the last year that isn't the LLM itself.

"Tell me all the great products that have come out since strategy we teach dropped. I don't know that there's a single new product in my life other than the LLM itself... It's still just as hard to innovate on product." - Paul Scherer 00:31:44

The Friend vs. Servant Distinction Is the Critical Design Choice Most Are Getting Wrong

Most AI products, including therapist bots and companion apps, are assistants — not friends. Paul argues this single distinction determines whether AI makes us more or less human.

"I would push back because I don't think Claude is a friend. Claude is a servant... And I think you lose a lot of the important parts of like a friend the second it becomes an assistant." - Paul Scherer 00:41:36

We Have a Belonging Problem, Not a Loneliness Problem

This is a reframe with significant product implications — the solution isn't to make people feel less alone in isolation (which AI companions attempt), but to make them feel embedded in a network of mutual recognition.

"I think we have a belongingness problem and much less a loneliness problem." - Paul Scherer 01:15:20

3. Companies Identified

Eigen

Description: Early-stage AI company building what they call a "mutual friend" — a single AI entity with a distinct personality that knows all its users and creates social serendipity between them.
Why mentioned: It is the subject of the episode; Paul is the founder. The product is in private beta. They raised a $15M seed round from Benchmark.
Quote: "We're building a mutual friend that is where a lot of what we're thinking about and what we're creating starts actually counterintuitively, not with the friend part, but the mutual part... every time we build something, we build it in a way where it is built on top of the network, where it's shared between all of the people." - Paul Scherer 00:35:03

Notion

Description: Collaborative workspace platform that has integrated AI agents deeply into its product, repositioning itself as an agent hub for individuals and teams.
Why mentioned: Presenting sponsor of the podcast; Notion co-founder Akshay Kothari introduced Paul to host Jackson Dahl.
Quote: "Notion also made a bunch of new announcements on its developer platform recently... the way that they're pushing what you can do with agents is really remarkable." - Jackson Dahl 00:02:47

Waymo

Description: Alphabet's autonomous vehicle company operating robotaxis in San Francisco.
Why mentioned: Used as an analogy for how products that seem terrifying in description become instantly intuitive in use — which Paul believes Eigen will follow.
Quote: "If I tell my mom about Waymo, she's really scared... and then you just force her to take a Waymo and it's like, I don't want to do this. It's like 10 seconds of the ride. It's like the most normal thing in the world." - Paul Scherer 00:37:17

4. People Identified

Sarah (from Benchmark)

Description: Partner at Benchmark who led or participated in Eigen's $15M seed round.
Why mentioned: Referenced for her insight about the "aura of success" that precedes the team, and for the quality of investor-founder relationship Paul describes.
Quote: "One of the most profound things that Sarah and I talked about when just after we sort of announced the round... it is very clear that the aura of success like preludes us... that's good in some ways because it means you get to hire really amazingly talented people." - Paul Scherer 01:27:12

Akshay Kothari

Description: Co-founder of Notion.
Why mentioned: Introduced Paul Scherer to podcast host Jackson Dahl; also referenced as a sounding board for Paul's ideas about purpose and restlessness.
Quote: "I was talking about this with Akshay a while ago and I was like, it's like short-term paranoia, but like long-term, everything is exactly the way it should be." - Paul Scherer 00:07:38

Zach Sims

Description: Entrepreneur referenced for his thinking on the filter bubble / personalization problem.
Why mentioned: Paul credits him for developing the idea that optimizing individual bubbles comes at the cost of collective coherence.
Quote: "I have to give some credit to Zach Sims on this... what if the world that's perfect for each of us isn't the world that's perfect for all of us." - Paul Scherer 01:16:13

Ben Silbermann (referenced as "Ben Solomon" likely meaning Ben Silbermann)

Description: Co-founder and former CEO of Pinterest.
Why mentioned: Paul cites his interview technique for filtering candidates who are just looking for good jobs versus those who are mission-aligned.
Quote: "He was always asking people... if you weren't working at Pinterest, where would you work? And sometimes there'd be people like, 'I'd probably work at Stripe'... and he was like, 'oh, you're probably not going to be the right person.'" - Paul Scherer 01:19:55

Demis Hassabis

Description: Co-founder and CEO of Google DeepMind.
Why mentioned: Used to illustrate the same point about mission-alignment vs. opportunity-seeking — Hassabis identified that Zuckerberg was an opportunity-seeker, not a singular-mission person.
Quote: "Demis Hassabis made this point when they were meeting with Google and Facebook about acquiring DeepMind... he asked Zuck about VR or something and he gave equally great answers. And he's like, 'oh, you're just looking for opportunities.'" - Paul Scherer 01:20:50

Kevin Kelly

Description: Founding executive editor of Wired Magazine, futurist author.
Why mentioned: Jackson calls him out for the subtle but important framing of "AIs" (plural) rather than "AI" (singular), pointing at a future of many distinct AI entities rather than one monolithic intelligence.
Quote: "The one person I've noticed who doesn't talk about it like that is Kevin Kelly... Kevin talks about AIs, which is a subtle but important difference." - Jackson Dahl 00:26:37

5. Operating Insights

Use "What Would a Real Person Do?" as a Universal Product Design Filter

Paul's team applies a single governing constraint across all product decisions: would a real person do this? This collapses complex AI design choices into intuitive human behavior patterns and prevents the assistant-mode drift that plagues competitors.

"If someone was really mean to me, how would I react? And it's just like, you sort of think about that and it's really helpful because you just, it's really that simple... that paradigm has already been established and so it's very intuitive for people. They don't have to learn anything new." - Paul Scherer 00:44:34

Build Personality from Underlying Traits, Not Prescribed Behaviors

Rather than scripting responses ("you love bananas"), Paul's team invests in identifying the motivations, beliefs, and behavioral traits that would cause a person to naturally express those responses. This is more robust and more scalable as the product grows.

"You have to figure out what are the motivations, incentives, feedback loops, behavioral traits that would then lead to a person that sort of embodies this behavior... you have to create a person that likes bananas — not because it's prescribed, but because I created a person that likes bananas." - Paul Scherer 01:11:36

Hire for "Thinks in Experiences, Not Technology" — and Screen for Singular Mission Fit

Paul's talent filter is people who start with the desired experience and reverse-engineer the technology, not the other way around. Combined with the Pinterest-style interview tactic, this creates a team that is both product-native and unshakeably mission-aligned.

"I love people who always just start with the product... they think in experiences, not in technology... ultimately they always start and come back to: here's this part of the experience that I want to enable." - Paul Scherer 01:21:40

6. Overlooked Insights

Consuming and Creating Are the Same Action — This Unlocks a New Platform Dynamic

Paul makes a brief but enormously significant observation that in Eigen's design, the act of asking a question IS an act of contribution to the network. This inverts the classic cold start problem of social networks and removes the "why would I post?" friction entirely.

"For the first time ever, the creation process might be the same as the consumption process, because you asking a question teaches me something about you — like, 'hey, what restaurant should I go to?' Oh, Jackson's going to a restaurant. It's like a platform where consuming is actually also creating." - Paul Scherer 00:58:14

This is a genuinely non-obvious insight: every social platform in history has struggled with the 90-9-1 rule (90% lurkers, 9% light contributors, 1% heavy creators). If Eigen collapses that distinction — if merely using the product is simultaneously contributing to it — they may have structurally solved the hardest problem in consumer social before anyone noticed.

The Brain's Default State is Social Cognition — and No Product Has Been Built for That

Paul briefly mentions neuroscience research showing that the brain's idle/default state is social reasoning — mapping relationships, recalling conversations, updating mental models of people. Every consumer product is built for task cognition. Eigen may be the first product designed to serve the brain's actual default mode.

"Most of your brain's activity is like social cognition... mapping people in terms of networks... it's actually the like default state of your brain... when people go from doing a task, the brain lapses back into social cognition, which is your idle state — updating your mental model of these relationships and recalling conversations you've had with people." - Paul Scherer 01:04:48

This suggests the TAM for what Eigen is building is not "AI companion" or "social network" — it's the ambient background state of human consciousness. That is an extraordinarily large surface area, and nobody in the conversation called it out.