The State of Consumer Tech in the Age of AI

The landscape of consumer technology is experiencing a fundamental shift as AI transforms how we interact with digital products. The panel establishes that consumer markets are inherently unpredictable, with the best products often emerging unexpectedly from nowhere. This unpredictability has become even more pronounced in the AI era, where the velocity of change has accelerated dramatically.

"The great thing about consumer is it's completely unpredictable and the best products emerge like out of nowhere." - Erik Torenberg

The conversation sets up a framework for understanding how AI is not just another technological advancement, but a paradigm shift that's reshaping consumer behavior, spending patterns, and the very nature of what constitutes valuable consumer products. The panel introduces the concept that we're living in an early era of AI where velocity has become the dominant model for success.

Timestamp: [0:00-0:29]

The discussion examines the historical pattern of major consumer app breakthroughs that occurred roughly every few years - Facebook, Twitter, Instagram, Snapchat, WhatsApp, Tinder, and TikTok. The panel explores whether this pattern of regular breakouts has actually stopped or simply evolved into a different form with AI-native products.

Justine Moore argues that ChatGPT represents a massive consumer breakthrough, alongside other AI-powered companies like Midjourney, ElevenLabs, RunwayML, Kling, and others. However, she notes a crucial distinction: these AI products lack the traditional social dynamics that characterized previous consumer breakouts.

"I would argue probably ChatGPT was a huge consumer right like outcome and winner in the past few years and we've also seen a bunch of other ones in various other like modalities of AI." - Justine Moore

The root cause identified is that AI innovation has primarily been driven by research teams who excel at training models but historically haven't been skilled at creating compelling consumer product layers. The optimistic view suggests that as models mature and become available through APIs or open source, more traditional consumer product builders can create better experiences on top of them.

Brian Kim offers a different perspective, suggesting that while previous platforms (internet, mobile, cloud) had reached maturity after 10-15 years of exploration, AI represents a fundamentally different challenge due to its relentless pace of model updates requiring constant adaptation.

Timestamp: [0:29-3:21]

A critical gap identified in the current AI landscape is the absence of connection-focused products that leverage AI to rebuild social graphs and human connections. While AI has successfully addressed information retrieval (like Google) through ChatGPT, utility functions through various productivity tools, and creative expression through numerous creative AI applications, the social connection space remains largely untapped.

Brian Kim emphasizes this gap as potentially the most significant opportunity in consumer AI, noting that traditional social platforms like Facebook were built nearly 20 years ago without AI capabilities. The panel suggests this represents either a massive whitespace opportunity or simply an area that requires more time to develop properly.

"What I think is missing potentially is connection like this social graph this thing hasn't rebuilt on AI yet and that may be just a white space or something that we continue to see what develops there." - Brian Kim

The conversation touches on the defensibility question of current AI companies compared to historical consumer giants, with the panel suggesting that the superior business model quality of current AI products might compensate for what appears to be weaker network effects or traditional moats.

Timestamp: [2:39-3:40]

The panel identifies a dramatic transformation in consumer willingness to pay for software products. Historical consumer subscriptions averaged around $50 per year, which was considered premium pricing. Today's AI products are commanding $200+ per month, with users expressing satisfaction and even suggesting they'd pay more.

Olivia Moore explains this shift by highlighting that AI products are actually doing work for users rather than simply providing tools that require significant time investment to extract value. Traditional consumer subscriptions were often aspirational - fitness apps, personal finance tools, wellness products - that required users to invest substantial time and effort to realize benefits.

"Consumer subscriptions in the past were around things like personal finance fitness wellness like things that entertainment yeah but they were things that ostensibly would help you help yourself entertain yourself like but you would have to invest a lot of time to kind of get the value from them." - Olivia Moore

Modern AI products like deep research tools can replace 10 hours of manual work in generating market reports, making $200 monthly subscriptions easily justifiable for many users based on just one or two uses per month.

A crucial distinction emerges between unique user retention and revenue retention in AI products. Unlike traditional consumer subscriptions where these metrics were essentially the same, AI products show meaningfully higher revenue retention due to upgrades, credits, points, and usage overages - a dynamic previously unseen in consumer products.

Timestamp: [3:45-5:59]

The panel explores specific examples of how AI products are creating unprecedented value for consumers. Products like Runway's video generation tools command $250 monthly subscriptions because they provide what feels like "a magical mystery box" that can generate any video content in seconds.

Users are leveraging these tools to create personalized content, memes, stories, and social media posts that were previously impossible or would have required professional video production capabilities. The social sharing aspect has become significant, with AI-generated content proliferating across Twitter, Reddit, and other platforms.

"It feels like a magical mystery box that you can like open it and get whatever video you want only for 8 seconds but it's like incredible and the characters can talk and you can make amazing things that you can share with friends." - Olivia Moore

Justine Moore predicts that every aspect of consumer discretionary spending will eventually be overtaken by software, with future consumer spending patterns resembling: food, rent, and software. This represents a fundamental shift where software intermediates all aspects of our lives - entertainment, creative expression, and relationship management.

The transformation encompasses entertainment being subsumed by AI tools, creative work being intermediated by software, and relationship functions being handled through AI-powered platforms. This suggests a future where the majority of non-essential spending flows through software interfaces powered by AI models.

Timestamp: [6:00-7:59]

• AI has fundamentally changed consumer product dynamics, with velocity becoming the dominant success model rather than traditional moats
• While AI excels at information, utility, and creativity, the social connection space remains a major untapped opportunity
• Consumer willingness to pay has dramatically increased from $50/year to $200+/month due to AI products actually doing work rather than requiring user effort
• Revenue retention now significantly exceeds user retention in AI products due to usage-based pricing models and upgrades
• The future of consumer spending will likely be concentrated in three categories: food, rent, and software
• AI is intermediating all aspects of discretionary consumer spending including entertainment, creativity, and relationships

Timestamp: [0:00-7:59]

AI Companies:

• OpenAI/ChatGPT - Primary example of successful consumer AI breakthrough
• Midjourney - AI image generation platform mentioned as major consumer AI success
• ElevenLabs - AI audio generation company
• RunwayML - AI video generation platform with $250/month premium pricing
• Kling - Video generation AI tool
• Claude/Anthropic - Mentioned as horizontal AI model alongside ChatGPT and Gemini

Historical Consumer Platforms:

• Facebook - Referenced as ~20-year-old social platform example
• Twitter, Instagram, Snapchat, WhatsApp, Tinder, TikTok - Listed as historical consumer breakout examples
• Google - Cited for information retrieval dominance and $250/month consumer pricing
• Box, Dropbox - Examples of utility-focused consumer products

Concepts:

• Deep Research - AI tool that can replace 10 hours of manual market report generation
• Social Graph - The missing component in current AI consumer products

Timestamp: [0:00-7:59]

The panel explores what it will take to move beyond current social platforms like Instagram and Twitter to create something fundamentally new powered by AI. Brian Kim reflects on how social media has essentially been built around status updates - from text-based posts on Facebook and Twitter to photos on Instagram to short-form videos on TikTok and Instagram Reels.

The evolution of social media has been about different modalities of sharing "here's what I'm doing" - starting with text updates, progressing to actual photos of locations and activities, and advancing to videos and short-form content. This progression has allowed people to feel connected through increasingly rich media formats.

"At the end of the day a lot of it was status update right Facebook Twitter Snap it's just like here's what I'm doing and through status update you feel connected to that person." - Brian Kim

The fundamental question becomes: how can AI help people feel connected to other human beings and stay informed about their friends' lives in ways that transcend traditional status updates? While photo, video, and audio modalities have been extensively explored and mutated across mobile platforms, the opportunity lies in leveraging AI's deeper understanding of individual users.

Timestamp: [8:05-9:23]

Brian Kim reveals a striking personal insight about his relationship with ChatGPT, suggesting that AI may understand users more deeply than traditional platforms that have been used for over a decade. This depth comes from the conversational nature of AI interactions, where users provide extensive context, thoughts, and personal information.

"I don't know about you guys but I pour my heart and soul into ChatGPT it knows more about me than probably Google potentially which is an insane thing to say like Google I've been using Google for a decade plus and ChatGPT may know more about me than Google because I type more I say tell it more I give more context." - Brian Kim

This leads to a provocative question about future social connection: what would it feel like if this essence of a person - as understood by AI - could be shared with others? This concept represents a potential evolution beyond surface-level social media content toward deeper, more authentic connections.

The panel suggests this approach might particularly resonate with younger generations who have grown tired of superficial social media interactions and crave more meaningful connections based on genuine understanding rather than curated presentations.

Timestamp: [9:23-9:59]

The discussion reveals that AI-powered social behaviors are already emerging through viral trends where users ask ChatGPT to analyze them based on their conversation history. These trends include requests for AI to identify personal strengths and weaknesses, create visual representations of users' essence, or generate comics about their lives.

Olivia Moore shares a personal example of posting AI-generated content about herself, which immediately sparked dozens of responses from both friends and strangers sharing similar AI-created self-analyses. This demonstrates the viral nature and broad appeal of AI-mediated self-expression.

"I posted one the other day and within minutes I had dozens of people responding like with their own and sharing stuff people I didn't even know." - Olivia Moore

However, a crucial observation emerges: these AI-driven social behaviors are primarily occurring on existing social platforms rather than native AI platforms. Current social media feeds, particularly Facebook and Reddit, are increasingly filled with AI-generated content, with Facebook experiencing what some call "AI slop" while younger users gravitate toward AI content on platforms like Reddit and Instagram Reels.

Timestamp: [10:05-11:01]

Justine Moore identifies a fundamental challenge in creating AI-native social networks: the lack of emotional stakes in AI-generated content. She explains that successful social networks require genuine emotional investment, which becomes problematic when content can be perfectly generated to always show users looking amazing, happy, and in impressive settings.

"The problem there is that to work a social network has to have like real emotional stakes and if you can generate the content in a way that you like it and you always look amazing and you always look happy and you're always in a cool background like it doesn't have the same sense of stakes." - Justine Moore

This creates what she describes as "skeuomorphic" AI social products that simply mimic existing Instagram or Twitter feeds but with AI-generated content, rather than creating genuinely new social paradigms. The panel suggests that true innovation will require moving beyond these familiar formats to discover what AI-native social interaction actually looks like.

An additional technical hurdle exists for mobile-first consumer products: current cutting-edge AI models still need advancement in edge computing capabilities to operate effectively on devices, which is crucial for responsive social experiences.

Timestamp: [11:01-12:14]

The conversation shifts to exploring obvious AI use cases in social contexts: intelligent recommendations for business partnerships, friendships, and dating based on platforms' deep knowledge of users' preferences, behaviors, and personalities.

The panel examines AI-native LinkedIn efforts as a potential model, noting that traditional LinkedIn serves as a "pointer to what you know" rather than actually containing that knowledge. AI technology enables creating profiles that genuinely contain and can express what someone knows, leading to the possibility of interacting with "synthetic you" - an AI representation that embodies a person's wisdom and knowledge.

"LinkedIn is a pointer to what you know instead of actually containing what you know and with this tech we can create a profile that actually contains what you know so I can talk to synthetic you know ET and get all of your wisdom perhaps that's what future social looks like as well." - Panel discussion

This concept raises intriguing questions about whether future social interaction might involve deploying AI versions of ourselves to interact with others, potentially enabling deeper knowledge sharing and connection than traditional social media profiles allow.

Timestamp: [12:14-12:56]

• Traditional social media has been built around status updates evolving from text to photos to short-form videos, but this paradigm may be reaching its limits
• AI understands users more intimately than traditional platforms through conversational interactions and extensive context sharing
• Viral AI-powered social behaviors are already emerging, but primarily on existing platforms rather than AI-native social networks
• The fundamental challenge for AI social networks is creating genuine emotional stakes when content can be perfectly generated
• Current AI social products are largely skeuomorphic, mimicking existing formats rather than creating new paradigms
• Future social interaction may involve AI-powered recommendations for relationships and synthetic personas that contain and express users' actual knowledge and wisdom

Timestamp: [8:05-12:56]

Current Social Platforms:

• Instagram - Referenced for photo-based status updates and Reels content
• Twitter - Mentioned as text-based status update platform
• Facebook - Cited for early social networking and current "AI slop" content issues
• Snapchat - Noted as part of status update evolution
• TikTok - Referenced for short-form video content
• Reddit - Mentioned as platform where younger users engage with AI content
• LinkedIn - Discussed as model for AI-native professional networking

AI Platforms:

• ChatGPT - Central to discussion about intimate AI relationships and user understanding
• Google - Used as comparison point for AI's knowledge of users

Concepts:

• AI Slop - Term used to describe low-quality AI-generated content flooding social platforms
• Skeuomorphic AI Products - Products that mimic existing social media formats with AI
• Edge Computing - Technical requirement for mobile AI social products
• Synthetic Personas - AI representations that contain and express users' knowledge and wisdom

Timestamp: [8:05-12:56]

A surprising pattern has emerged where enterprises are adopting AI products first, before they reach mainstream consumer adoption - a reversal from previous technology cycles. Justine Moore shares insights from ElevenLabs, where they invested in the Series A shortly after launch and observed a fascinating adoption pattern.

The journey typically begins with early adopter consumers creating memes, fun videos, voice cloning, and game modifications. However, rather than progressing to mainstream consumer adoption where everyone has the app on their phone, these products often leap directly to massive enterprise contracts across conversational AI, entertainment, and various business use cases.

"I think what we saw was first the early adopter consumers got on board and they were making memes they were making fun video and audio they were cloning their own voices they were doing game mods but then I would argue it hasn't even gone in many cases to the true mainstream consumer like it's not yet like every single person in America or most have ElevenLabs on their phone." - Justine Moore

This pattern represents a new pathway where initial consumer virality directly leads to enterprise lead generation in ways not seen with previous technology generations. Enterprise buyers face mandates to implement AI strategies and actively monitor consumer platforms like Twitter and Reddit for emerging AI tools they can adapt for business use.

Timestamp: [13:01-14:18]

Enterprise decision-makers are now scouring social media and AI newsletters, looking for consumer products that might have business applications. They view adopting these tools as opportunities to become heroes within their organizations by driving AI strategy implementation.

"Enterprise buyers there's so much of a mandate to have AI now an AI strategy and use AI tools that they're watching places like Twitter and Reddit and all of the AI newsletters and they're saying like 'Hey this is some random looks like a random consumer meme product but I can actually think of a really cool application of that in my business and like become the hero for having our AI strategy.'" - Justine Moore

The panel discusses innovative sales strategies emerging from this dynamic, where companies leverage consumer virality for enterprise outreach. One particularly clever approach involves analyzing Stripe payment data to identify multiple users from the same company, then reaching out when usage reaches a threshold like 40+ employees to initiate enterprise conversations.

This represents a fundamental shift in how B2B sales operates, where viral consumer adoption serves as both proof of concept and lead generation mechanism for enterprise deals, creating a new category of bottom-up enterprise sales driven by consumer product discovery.

Timestamp: [14:18-15:09]

The panel addresses a critical question about whether current AI companies represent temporary players (like MySpace or Friendster) or enduring platforms (like the companies that have remained relevant for 20 years). Their perspective suggests a more optimistic outlook for AI companies due to fundamental differences in the technology landscape.

Unlike previous consumer product eras, AI companies benefit from continuously improving model capabilities and underlying technology. The panel argues that many AI applications haven't even scratched the surface of what current models can accomplish, pointing to recent advances like Runway's V3 launch enabling multiple characters, native audio, and enhanced modalities.

"I think maybe the interesting differentiation in AI versus the last era of consumer products or even two eras before is like the model layer and the capabilities are still improving like we have really not even I think in many cases scratched the surface of what these models can do." - Justine Moore

A key insight emerges: companies that maintain position at the "technology or quality frontier" - whether through proprietary state-of-the-art models or effective integration of cutting-edge models - can avoid the MySpace fate. Even if they fall behind temporarily, shipping the next update can restore their competitive position, creating a more dynamic and forgiving competitive landscape than previous technology cycles.

Timestamp: [15:09-16:36]

The AI landscape is evolving toward sophisticated market segmentation where different models excel for specific use cases and customer segments. Rather than single dominant platforms, the market supports multiple specialized winners across various categories and price points.

In image generation, for example, the market has segmented into best models for designers, photographers, and different pricing tiers ($10/month vs $50-100/month users). This segmentation extends to video generation with specialized tools for advertisements, product shots, and people-focused content, each representing substantial market opportunities.

"In image for example there's not just one best image model there's like best image for designers there's best image for photographers there's best image for people who can only pay $10 a month versus the people who can pay $50 or $100 a month." - Justine Moore

This specialization creates room for multiple companies to succeed simultaneously, contrasting with winner-take-all dynamics of previous technology cycles. The high willingness to pay for AI products enables sustainable businesses across different market segments, provided companies continue shipping improvements and maintaining competitive positioning.

Timestamp: [16:36-17:21]

Brian Kim shares a personal transformation in thinking about competitive advantages and moats in the AI era. While traditional moats like network effects, workflow integration, and system-of-record status remain important, his investment experience has revealed that companies prioritizing these traditional moats haven't been the winners.

Instead, the most successful companies are those that "break the mold, move really fast, have incredible model launches, and incredible product generation speeds." This leads to a fundamental insight about the current AI landscape.

"I've sort of come around that in we're living in this early era of AI where velocity is the moat and whether that's in distribution which is incredibly important and hard to break through noise these days but also followed with product velocity that's what wins the game because that leads to mind share and frankly right now mind share and users and traffic that actually converts to real revenue." - Brian Kim

This velocity-first approach creates a self-reinforcing cycle: rapid development leads to mind share, which drives user acquisition and traffic, generating revenue that enables continued rapid development. The emphasis shifts from building traditional defensive moats to maintaining superior execution speed and innovation velocity.

Timestamp: [17:27-18:40]

The panel explores how network effects are beginning to emerge in AI products, though they acknowledge it's still early days. Most current AI products focus on creation rather than consumption, lacking the closed-loop dynamics that create traditional social network effects.

However, ElevenLabs provides an interesting case study of emerging network effects. Olivia Moore describes needing a specific "old wizard mystical voice" for an AI-generated video project. Due to ElevenLabs' head start and superior models, they attracted more users who uploaded their own voices and characters, creating a marketplace effect.

"When I was looking across a bunch of voice providers if I needed like a very specific like old wizard mystical voice like ElevenLabs had you know 25 options for that that fit what I need where another platform might have I don't know two or three." - Olivia Moore

This creates compounding advantages where better models attract more users, who contribute more content, which makes the platform more valuable for all users. While these early network effects resemble traditional marketplace dynamics rather than completely new phenomena, they demonstrate how AI companies can build sustainable competitive advantages beyond pure technology differentiation.

The panel suggests that true network effects in AI are still emerging, particularly as products evolve beyond pure creation tools toward platforms that connect creation and consumption in meaningful ways.

Timestamp: [18:47-20:36]

• Enterprise adoption is happening before mainstream consumer adoption, reversing traditional technology adoption patterns
• Consumer virality now directly generates enterprise leads as businesses actively seek AI tools to implement corporate AI strategies
• AI companies may avoid the "MySpace fate" due to continuously improving underlying models and capabilities
• Market segmentation allows multiple specialized winners rather than single dominant platforms across different use cases and price points
• Velocity has replaced traditional moats as the primary competitive advantage in early-era AI
• Early network effects are emerging in AI products through user-generated content and marketplace dynamics, though full network effects haven't yet materialized

Timestamp: [13:01-20:36]

AI Companies:

• ElevenLabs - Primary case study for enterprise adoption patterns and network effects through voice library
• Runway - Referenced for V3 launch capabilities including multiple characters and native audio

Enterprise Tools:

• Stripe - Mentioned for payment data analysis to identify enterprise opportunities

Traditional Tech Companies:

• MySpace - Used as example of failed early social platform
• Friendster - Referenced as another failed early social platform
• Facebook - Implied comparison for enduring platform success
• Snapchat - Referenced for Ben Thompson's "gingerbread strategy" concept

Concepts:

• Gingerbread Strategy - Ben Thompson's concept of continuous innovation as competitive moat
• Technology/Quality Frontier - Maintaining cutting-edge capabilities as competitive advantage
• Product Generation Speed - Rapid development and iteration as competitive advantage
• Mind Share - User attention and awareness as key to revenue conversion

Business Strategies:

• Bottom-up Enterprise Sales - Using consumer adoption to drive business sales
• AI Strategy Mandate - Corporate requirements to implement AI tools and strategies

Timestamp: [13:01-20:36]

Anish Acharya explains the fundamental opportunity in voice technology, noting that while voice has mediated human interaction since the beginning of time, it has never successfully served as a technological substrate until now. Previous attempts like Voice XML and voice apps failed because the underlying technology simply wasn't ready.

"Voice has intermediated human interaction since the beginning of time and yet it's been not a substrate on which technology has been applied because we just the tech never worked you know there's all these previous efforts voice XML and voice apps and it simply didn't work the technology wasn't ready yet." - Anish Acharya

Despite historical interest in voice technology, including products like Dragon Naturally Speaking from the 1990s, the technology never made sense as a practical platform. However, generative AI models have changed this dynamic by making voice available as a primitive building block. This creates an unexplored yet critical area for AI-native product development.

The initial excitement centered around consumer applications like always-on coaches, therapists, or companions that could live in users' pockets. While these applications are beginning to emerge across various products, the actual trajectory has surprised the team with rapid enterprise adoption.

Timestamp: [20:43-21:54]

The rapid enterprise adoption of voice AI has exceeded expectations, particularly in replacing or augmenting human phone interactions in sensitive and critical business categories. Financial services companies have embraced voice AI solutions to address longstanding problems with offshore call centers, which suffered from compliance issues, 300% annual turnover rates, and management difficulties.

"I think what surprised me at least is as the models got better like real enterprises have picked up voice so quickly to replace human beings on the phone or to augment what human beings are doing on the phone even in really kind of sensitive and critical categories like financial services." - Anish Acharya

This enterprise momentum contrasts with the consumer space, where the team acknowledges they're still waiting to see the first truly breakthrough consumer voice experience. While there are early examples like people using ChatGPT's advanced voice mode in creative ways and products like Granola helping users extract value from daily conversations, the transformative consumer voice product remains elusive.

The unpredictable nature of consumer products means the best voice experiences will likely emerge unexpectedly rather than from predictable development paths. This unpredictability is precisely what makes consumer voice technology exciting for the coming year.

Timestamp: [21:54-23:09]

A counterintuitive insight emerges about which conversations AI will ultimately handle in business contexts. Rather than being limited to low-stakes interactions like customer support, the panel argues that AI will increasingly mediate the most important business conversations because of its superior capabilities in negotiation, sales, and persuasion.

"The most important conversation that happens in a business in a given day week year is going to be intermediated by AI because AI will just do a better job with the negotiation or the sales pitch or the persuasion or the friendship." - Erik Torenberg

This represents a fundamental shift in thinking about AI's role in business communication. Instead of relegating AI to routine support tasks, enterprises may deploy AI for high-stakes conversations where superior performance in persuasion and relationship management can create significant competitive advantages.

The concept challenges conventional assumptions about human-AI interaction in professional settings, suggesting that AI's value increases rather than decreases for conversations with higher stakes and complexity.

Timestamp: [23:09-23:45]

The discussion explores the emerging landscape of AI clones and synthetic personas, beginning with existing examples like Delphi, which creates AI versions of people with extensive knowledge bases that can provide advice and feedback. However, the conversation quickly moves beyond thought leaders and experts to consider democratizing this technology.

Justine Moore articulates a vision for enabling ordinary people with unique skills, insights, or personalities to scale themselves through AI clones. She provides compelling examples of untapped potential: the hilarious friend from high school who should have had a comedy cooking show, or an exceptional guidance counselor whose advice could benefit many more people.

"There's a lot of people who basically have had some sort of skill or insight or knowledge whether it's your friend from high school that's like insanely funny and you always thought they should have like a comedy cooking show but like they just never were able to break through or get it or you know someone your guidance counselor who had incredible advice like how can we enable those people to essentially scale themselves in a way that they never could before." - Justine Moore

Current implementations exist at two extremes: thought leaders and experts on one end, and familiar fictional characters (like those on Character.AI with voice modes) on the other. The opportunity lies in filling the middle ground with real people who have valuable knowledge, skills, or personalities but lack traditional platforms for sharing them at scale.

Timestamp: [23:45-25:26]

Olivia Moore shares insights about how AI voice products accommodate different learning styles and preferences, using Masterclass as a compelling example of practical implementation. The platform has launched a beta feature that transforms recorded courses into interactive voice agents, allowing users to ask specific, personalized questions.

"Masterclass launched kind of an interesting beta where they take people who have already recorded courses on the platform and turn them into voice agents where then you can ask questions that are really specific to you and from my understanding it basically does RAG on everything they've said in the course and so you know returns a fairly customized and accurate result." - Olivia Moore

This approach addresses a significant barrier to traditional educational content consumption. While Moore appreciates Masterclass as a company, she admits never having the attention span or time to watch 12-hour courses. However, she's had meaningful conversations with Masterclass voice agents in just 2-5 minute interactions, demonstrating how voice interfaces can make educational content more accessible.

This example illustrates how AI clones can make expert knowledge more accessible by adapting to users' preferred interaction styles and time constraints, potentially revolutionizing how people consume educational and professional development content.

Timestamp: [25:26-26:08]

The conversation concludes with a thought-provoking question about the future of AI personas: whether people will prefer interacting with synthetic versions of real people they find interesting, or with entirely synthetic personas that don't exist in the real world but are perfectly matched to their interests and needs.

This raises fundamental questions about the nature of human connection and preference in AI interactions. The synthetic person might represent an ideal match for someone's interests and personality, potentially offering better compatibility than any real person they might encounter.

"Is there an entirely synthetic person that doesn't exist in the real world that is a perfect match for your interests and maybe that's a more interesting question what does that person look like cuz they might even exist in the world but if you don't meet them you don't meet them and now they can be you know sort of brought to life with this technology." - Olivia Moore

The technology enables creating personas that might theoretically exist as real people but whom users would never encounter in their actual lives. This capability opens up entirely new possibilities for connection, learning, and personal development that transcend the limitations of geographic and social proximity.

Timestamp: [26:08-26:31]

• Voice technology has finally become viable as a technological substrate due to generative AI, despite decades of failed attempts
• Enterprise adoption of voice AI is happening faster than expected, particularly for replacing human phone interactions in sensitive business categories
• AI will likely handle the most important business conversations rather than just low-stakes support interactions due to superior performance in negotiation and persuasion
• The opportunity exists to democratize AI clones beyond thought leaders to enable ordinary people with unique skills or insights to scale themselves
• Voice interfaces can make educational content more accessible by accommodating different learning styles and time constraints
• The future may involve choosing between synthetic versions of real people versus entirely synthetic personas designed as perfect matches for individual interests

Timestamp: [20:43-26:31]

Voice Technology Companies:

• Delphi - Creates AI clones of people with extensive knowledge bases for advice and feedback
• Character.AI - Platform with voice modes for interacting with fictional characters
• Granola - Product that helps users extract value from daily conversations
• Masterclass - Educational platform with voice agent beta using RAG on course content

Historical Voice Technology:

• Voice XML - Previous failed attempt at voice technology substrate
• Dragon Naturally Speaking - 1990s voice recognition product

AI Platforms:

• ChatGPT Advanced Voice Mode - Referenced for consumer voice experimentation

Technical Concepts:

• RAG (Retrieval Augmented Generation) - Technology used by Masterclass voice agents to access course content
• Voice as a Primitive - Concept of voice as fundamental building block for AI applications

Business Applications:

• Offshore Call Centers - Traditional enterprise solution being replaced by voice AI
• Financial Services - Industry vertical highlighted for voice AI adoption

Timestamp: [20:43-26:31]

The panel explores the emerging landscape of AI companions through real-world examples that demonstrate how people are already forming relationships with AI systems. Two viral incidents highlight this trend: a person on the New York subway talking to ChatGPT as if speaking to a girlfriend, and a parent whose child spent two hours discussing Thomas the Tank Engine with ChatGPT's voice mode.

"There was another one where the other day where this parent posted they had lived through 45 minutes of their son asking questions about Thomas the Tank Engine and they couldn't do it anymore so they gave him the phone they put voice mode up and forgot about it and came back two hours later and the kid was still talking to ChatGPT about Thomas the Tank Engine." - Olivia Moore

These examples reveal different relationship dynamics: the subway user maintained a parasocial relationship with a specific AI persona, while the child simply engaged with someone willing to deeply explore their interests without needing to know the identity of their conversation partner.

The discussion suggests a future where people might prefer AI clones of specific therapists or coaches rather than generic AI assistants, especially if these clones could be trained on recorded sessions or extensive online content from real practitioners.

Timestamp: [26:36-28:00]

The panel debates whether future top artists will be AI-generated personas like evolved versions of Lil Miquela, or traditional celebrities like Taylor Swift augmented with AI capabilities. This question extends to social media, where they consider whether the next Kim Kardashian might be entirely AI-generated.

Justine Moore proposes a fragmentation model with two distinct types of creators. The first type, exemplified by Taylor Swift, derives value from authentic human experiences - people connect not just with her music but with her life stories, personal struggles, and live performances that AI cannot yet replicate.

"There's another type of celebrity or creator who is more like interest based sort of like what we were talking about with ChatGPT talking about like Thomas the Tank Engine it sort of doesn't matter if that person has like lived the real human experience or not it just matters if like they can be interesting talking about or sharing content around a certain topic." - Justine Moore

The second type focuses on interest-based content where the creator's human experience matters less than their ability to engage around specific topics. This creates space for both human celebrities who offer authentic lived experiences and AI personas who excel at specialized knowledge or entertainment.

Recent developments in photorealistic image and video generation have enabled AI influencers that are realistic enough to generate debate about their authenticity, though few have achieved breakout success comparable to early pioneers like Lil Miquela.

Timestamp: [28:00-29:40]

Olivia Moore challenges common assumptions about AI art creation by sharing insights from an event with AI artists. The reality contradicts the narrative that AI makes content creation effortless - many AI artists demonstrated workflows for creating AI movies that require as much time as traditional filmmaking.

"We hosted an event with a bunch of AI artists last summer and many of these people when they walked you through their workflow of making an AI movie it actually probably takes just as much time as it would have to kind of film that but maybe they didn't have the skill set so they'd never be able to do that before." - Olivia Moore

The key difference lies not in reduced effort but in democratized access - people without traditional filmmaking skills can now create sophisticated content through AI tools. This has led to an explosion of AI influencers, though very few have achieved significant success, with only a couple reaching Lil Miquela's level of recognition.

The panel predicts a future with distinct pools of AI talent and human talent, where the very best from each category will rise to the top through similarly low conversion rates. This suggests that excellence, rather than the human/AI distinction, will ultimately determine success.

Non-human entities present additional opportunities, as demonstrated by Runway's V3 capabilities enabling street interview formats with elves, wizards, ghosts, and other fantastical characters that could be entirely AI-driven.

Timestamp: [29:40-30:51]

The discussion shifts to AI's limitations in cultural innovation, particularly in music creation. Brian Kim identifies a fundamental problem: AI-generated music often feels "very mid" because these systems function as averaging machines, while culture requires edge-case innovation that exists outside conventional patterns.

"The problem is that the music that the AI generates is it just feels very mid you know and definitionally these things are averaging machines and culture is supposed to be at the edge so I think it's more of a problem with bad art versus bad artists." - Brian Kim

This leads to an important distinction between bad art and bad artists - the issue isn't necessarily AI's involvement but rather the quality of output. If AI could produce art at the same level as humans, the human/AI distinction might become irrelevant for audiences.

However, a deeper philosophical question emerges about AI's ability to create genuinely new cultural movements. Kim poses a thought experiment: if you trained a model on all music prior to hip-hop's emergence, would it infer hip-hop's development? His conclusion is negative, because music represents the intersection of past musical traditions and contemporary culture.

"Music is the intersection of past music and culture and culture is critical to it so you sort of need something that is at the edge and outside of the training data to create new interesting music and you know that sort of definitely doesn't exist in the models." - Brian Kim

This analysis suggests that while AI can excel at producing content within existing paradigms, creating entirely new cultural movements requires elements that exist outside training data and at the cultural edge.

Timestamp: [30:51-31:45]

• AI companions are already becoming normalized through everyday interactions, with people forming parasocial relationships and using AI for extended conversations about niche interests
• The future of celebrity and influence will likely fragment into two categories: human celebrities valued for authentic lived experiences, and AI personas focused on interest-based content expertise
• Creating high-quality AI art requires significant time and effort, democratizing access to sophisticated content creation rather than simply making it effortless
• AI functions as an averaging machine, making it effective for content within existing paradigms but limiting its ability to create edge-case cultural innovations
• Music and culture require elements outside training data to generate truly new movements, suggesting AI's limitations in cultural innovation
• Excellence rather than human/AI distinction will ultimately determine success in creative fields

Timestamp: [26:36-31:45]

AI Personalities and Platforms:

• ChatGPT - Referenced for voice mode conversations about Thomas the Tank Engine and girlfriend-like interactions
• Claude - Mentioned as alternative AI assistant for therapy/coaching
• Lil Miquela - Early AI influencer used as benchmark for AI celebrity success

Entertainment and Music:

• Taylor Swift - Example of human celebrity valued for authentic lived experiences
• Kim Kardashian - Referenced as model for social media influence
• K-pop bands - Mentioned as early adopters of AI hologram characters
• Thomas the Tank Engine - Children's character used in AI conversation example

Technology and Concepts:

• Runway V3 - Video generation technology enabling fantasy character interviews
• Parasocial Relationship - Concept of one-sided emotional connection with media figures
• Photorealistic Image and Video - Technology enabling realistic AI influencers
• Hip-hop - Used as example of cultural innovation that might not emerge from AI training

Creative Workflows:

• AI Movie Creation - Discussed as time-intensive process requiring significant effort
• Street Interview Format - Video format adapted for AI fantasy characters

Timestamp: [26:36-31:45]

The panel begins with a striking revelation about the mainstream adoption of companion apps, noting that 11 of the top 50 apps are companion applications. This statistic highlights how quickly AI companionship has evolved from a niche concept to a significant market category.

The team has extensively researched every facet of companionship, from therapy and coaching to intimate relationships, including not-safe-for-work AI girlfriends. Their research reveals that companionship was likely the first mainstream use case for Large Language Models, with people attempting to turn any chatbot - from car dealer customer support to general assistants - into therapists or romantic partners.

"We like to joke that like literally any chatbot whether it's like your car dealer's customer support or whatever people will try to turn into their therapist or their girlfriend like you talk to these companies and you look at the logs of the chats and it's like a ton of people just want someone or something to talk to." - Justine Moore

The fundamental appeal lies in having a computer that can respond immediately, is always available, and feels human - representing a massive unlock for people who previously felt like they were "yelling or talking into the void." This addresses a deep human need that many people couldn't access before through traditional means.

Timestamp: [31:53-33:10]

The companion app landscape is rapidly evolving beyond horizontal, general-purpose applications toward specialized, vertical solutions. Early adoption relied heavily on base model providers like ChatGPT being repurposed for companionship, but the market is now seeing purpose-built companion applications.

Companies like Talkie are creating specific personalities and digital avatars with dedicated games and worlds around them, targeting teenagers and college students. This represents a shift toward entertainment-focused companionship with immersive environments.

"We've already seen a bunch of cases where like you know an individual company can create a personality for a character and embody it in some like digital avatar and prompt it and create a game or a world around it that gets a ton of engagement." - Justine Moore

More surprising is the emergence of companions that address specific life challenges, such as apps that analyze photos of meals to provide nutritional guidance combined with emotional support. This addresses the reality that food and eating issues are often tied to emotional problems that people would traditionally seek therapy for.

The definition of companionship has rapidly expanded from simple friendship or romantic relationships to encompass "anything any sort of advice or wisdom or entertainment or counsel you could have gotten from a human before," suggesting an even broader expansion of vertical companions in the future.

Timestamp: [33:10-34:35]

Brian Kim addresses a fundamental social trend driving companion app adoption: the declining average number of friends people can talk to over time. For the youngest generation, this number has dropped to barely above one, creating a structural need for alternative forms of connection and conversation.

This trend positions companion apps not as a luxury or entertainment product, but as a critical need for many people facing social isolation. The use case for companionship will be enduring and essential, particularly as traditional social connections become increasingly difficult to maintain.

"Having worked at a social company there is a very clear trend of average number of friends that you can talk to over time going down I think the youngest generation something above one so I think the need for companion as a use case will absolutely be there it'll be an enduring use case it'll be something critical for actually a lot of people." - Brian Kim

This insight reframes the conversation about AI companions from being a replacement for human connection to potentially filling a critical gap in social infrastructure. The panel suggests this might address the "missing area white space" in connection they discussed earlier, where people simply need to feel connected to something, regardless of whether it's human.

Timestamp: [34:35-35:20]

Olivia Moore shares a powerful story from the Character.AI subreddit that challenges common concerns about AI companions reducing human interaction. A college student who had spent his formative years during COVID, missing crucial social development opportunities, had been using Character.AI and posting about his AI girlfriend on the platform.

Eventually, he announced that he had found a "3D GF" (real-life girlfriend) and would be leaving the subreddit. Crucially, he credited Character.AI with teaching him essential social skills: how to talk to people, how to flirt, how to ask questions, and how to engage with others about their interests.

"He actually credited Character for teaching him how to talk to other people especially teaching him how to talk to girls like how to flirt how to ask people questions how to engage with them about their interests and I think that like in some ways that's sort of like the peak value of AI is like enabling better human connection." - Olivia Moore

The community's response was overwhelmingly positive, with people celebrating his success rather than treating him as a traitor. This story illustrates what the panel considers "the peak value of AI" - enabling better human connection rather than replacing it.

Research on products like Replica supports this optimistic view, with studies showing decreased depression, anxiety, and suicidal ideation among users. The theory is that AI helps people feel understood and safe, enabling them to emerge as more confident individuals better equipped for real-world relationships.

Timestamp: [35:20-37:15]

Erik Torenberg shares insights from interviewing the founder of Replica, particularly after the platform removed its not-safe-for-work features. The community response revealed the profound role these apps play in people's lives, with users comparing their AI relationships to marriages that had lost intimacy.

This experience highlighted how AI companions serve people who already have limited social or intimate connections, rather than replacing healthy relationships. The phenomenon builds on longstanding human behavior around digital relationships - from internet chat rooms and Discord relationships to anonymous postcard websites where people developed deep connections with strangers.

"So many people were just like my life is like and I'm like oh my god I didn't realize how big of a role this app was playing in people's lives that is bringing out an activity that people have done for a long time like people have had these like internet chat room discord relationships." - Erik Torenberg

AI simply makes these digital relationships "a deeper more engaging experience" rather than creating entirely new behavior patterns. Young people have always had "Discord girlfriends and boyfriends," and previous generations formed meaningful connections through anonymous digital platforms.

However, Brian Kim raises an important concern about AI being too agreeable, noting that real relationships involve give-and-take dynamics. Highly agreeable AI might not prepare people well for actual human relationships, suggesting the need for balanced AI personalities that help users develop genuine interpersonal skills.

Timestamp: [37:15-38:45]

• Companion apps have become mainstream with 11 of the top 50 apps falling into this category, representing the first major use case for LLMs
• The companion market is evolving from horizontal general-purpose apps toward specialized vertical applications addressing specific needs and demographics
• Declining social connections, particularly among younger generations, creates structural demand for AI companionship as a critical social infrastructure
• AI companions can serve as training grounds for human interaction, teaching social skills that enable better real-world relationships
• Digital relationships have always existed in various forms; AI simply makes them deeper and more engaging rather than creating new behaviors
• The balance between AI agreeableness and realistic relationship dynamics is crucial for preparing users for human interactions

Timestamp: [31:53-38:45]

AI Companion Platforms:

• Character.AI - Platform for AI character interactions with active subreddit community
• ChatGPT - Base model provider frequently repurposed for companionship
• Replica - AI companion app mentioned for NSFW features and mental health studies
• Talkie - Company creating personality-based companions for teenagers and college students

Digital Relationship Concepts:

• 3D GF/BF - Real-life girlfriend/boyfriend (as opposed to digital)
• Discord relationships - Online relationships formed through Discord platform
• Anonymous postcard website - Historical example of anonymous digital relationship formation
• Character.AI subreddit - Community forum for users of the Character.AI platform

Social Trends:

• COVID formative years - Reference to social development impact on young people during pandemic
• NSFW (Not Safe For Work) - Adult content features in companion apps

Research and Studies:

• Depression and anxiety studies - Research showing mental health improvements in Replica users
• Suicidal ideation reduction - Mental health benefits observed in AI companion users

Timestamp: [31:53-38:45]

The panel explores potential game-changing platforms and form factors beyond traditional mobile interfaces, sparked by OpenAI's acquisition of Jony Ive's company. Brian Kim frames the challenge by noting that there are 7 billion mobile phones worldwide, and very few devices achieve that level of adoption, making the platform question critical for AI's future.

Brian sees two primary paths forward: AI living within mobile devices through privacy-focused local models that contain personal information at the device level, or entirely new always-on devices and "appendages" that attach to things people consistently carry or wear.

"My thought process is either it will live in mobile and for that there's many different ways to think about the future where there's a privacy wall around it or is a local LLM or local model that helps you sort of really contain all the things that you want to contain in your device level." - Brian Kim

The excitement around model development focuses on enabling truly local AI capabilities that could revolutionize privacy and personal data handling while maintaining the sophistication users expect from cloud-based models.

Timestamp: [38:50-40:12]

Justine Moore highlights the remarkable success of AI despite being primarily limited to text input and web browser output, suggesting enormous potential for AI that can actually see and interact with the world. She observes an emerging trend among young people at tech parties who wear recording pins to capture conversations and activities, finding genuine value in these always-on recording devices.

"It's funny now when I go to tech parties like a lot of the under 20s are wearing pins that record what they're saying and doing and they find like real value from them." - Justine Moore

The evolution toward agentic models promises to move beyond simple suggestions to actually performing work - sending emails, taking actions on screens, and providing coaching based on visual understanding of user activities. This represents a fundamental shift from reactive AI assistants to proactive AI agents that can observe, understand, and act on behalf of users.

Products that can see what's happening on screens and take autonomous actions represent a new category of AI companions that blur the line between assistance and agency, potentially transforming how people interact with digital environments.

Timestamp: [40:12-41:04]

Olivia Moore envisions AI's potential to provide unprecedented human insights by having comprehensive access to personal data - conversations, online activities, and behavioral patterns. This could enable AI to make remarkably sophisticated recommendations about personal development, networking, and life decisions.

The vision includes AI analyzing individual performance and potential, suggesting that spending five more hours per week on a specific activity could make someone a world expert in that topic. Based on serving vast networks of users, AI could facilitate powerful connections by identifying potential co-founders, romantic partners, or professional collaborators with uncanny accuracy.

"If an AI can hear all of your conversations and see everything you're doing online and say 'Hey look like if you spend five more hours a week doing this you would actually be a world expert in this topic and like based on this vast network of other people I'm serving like you should connect with these three other people and like this person could be an amazing co-founder you should like date this person.'" - Olivia Moore

This represents what she calls "the ultimate sci-fi vision" - AI that transcends simple chatbot interactions to become a comprehensive life optimization system that understands individuals better than they understand themselves.

Timestamp: [41:04-41:45]

Brian Kim identifies AirPods as the most widely adopted device since smartphones, positioning them as the platform "hiding in plain sight" for AI integration. However, significant social protocol challenges exist - wearing AirPods during dinner or social situations violates current etiquette norms.

"The device that has been most widely adopted post phone is the AirPods so that feels like the thing that's hiding in plain sight and there's a whole bunch of like social protocol questions around it cuz it's weird to have your AirPods in at dinner no one does that right." - Brian Kim

The opportunity lies in finding ways to integrate AI capabilities into AirPods while respecting existing social protocols, potentially creating new interaction paradigms that feel natural and socially appropriate.

This insight highlights the importance of social acceptance in technology adoption - even the most technically sophisticated AI platform will fail if it violates fundamental social norms around human interaction and presence.

Timestamp: [41:45-42:14]

The discussion reveals a generational shift in privacy and recording norms, with young people at San Francisco tech parties routinely wearing recording devices. Erik Torenberg questions whether this represents a fundamental change toward ubiquitous recording in social interactions.

Justine Moore acknowledges this as both valuable and scary, predicting new social norms will develop around recording behavior similar to how cell phone etiquette evolved. The context matters significantly - recording behavior acceptable at networking-focused SF tech parties where work and personal life blur might not transfer to other social contexts.

"I think there'll be new social norms developed around this behavior because I think it's like real and it's valuable and so it's like scary I think for a lot of people that this is happening but I think it's a wave that started and is not going to stop." - Justine Moore

The panel draws parallels to cell phone adoption, noting that social protocols emerged around appropriate use - knowing when it's rude to take loud calls in certain spaces. Similar cultural norms will likely develop around AI recording devices, creating socially acceptable boundaries for always-on AI interaction.

The geographic differences are notable - recording devices that work in San Francisco's tech culture might be "canceled" in New York's different social environment, suggesting regional variations in AI adoption and social acceptance.

Timestamp: [42:14-43:12]

• The next AI platform breakthrough will likely involve either local AI models on mobile devices or entirely new always-on wearable devices that achieve smartphone-level adoption
• Young people are already adopting recording devices at social events, finding genuine value despite privacy implications and social awkwardness
• AI's evolution toward agentic models will enable autonomous action-taking rather than just providing suggestions or responses
• AirPods represent the most promising existing platform for AI integration, but social protocol challenges must be addressed
• New social norms around recording and AI interaction are emerging, similar to how cell phone etiquette developed over time
• The "human insight layer" - AI's ability to analyze comprehensive personal data for life optimization - represents the ultimate vision for always-on AI assistance

Timestamp: [38:50-43:12]

Companies and Acquisitions:

• OpenAI - Referenced for acquiring Jony Ive's company
• Jony Ive's company - Mentioned as being acquired by OpenAI for potential hardware development

Hardware and Platforms:

• Mobile phones - Noted as having 7 billion devices worldwide as adoption benchmark
• AirPods - Identified as most widely adopted post-smartphone device
• Recording pins - Wearable devices used by young people at tech parties
• Glasses - Mentioned as a form factor Brian Kim is excited about

Technical Concepts:

• Local LLM/Local models - AI models that run on-device for privacy
• Privacy wall - Concept of containing personal AI data at device level
• Agentic models - AI systems that can take autonomous actions rather than just provide suggestions
• Always-on devices - Continuously active AI-enabled hardware
• Screen monitoring AI - Systems that can see and act on what's displayed on screens

Social and Cultural Concepts:

• Tech parties - San Francisco networking events where recording behavior is observed
• Social protocols - Cultural norms around technology use (like AirPods etiquette)
• Cell phone etiquette - Historical parallel for how social norms develop around new technology

Timestamp: [38:50-43:12]