Nov 5, 2025 · 50m · a16z
Seeing The Future from AI Companions to Personal Software
Eugenia Kuyda · 37m spoken Justine Moore · 4m spoken Anish Acharya · 2m spoken Erik Torenberg · 1m spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The a16z Podcast, Wabi Founder and CEO Eugenia Kuyda discusses the future of artificial intelligence, outlining a shift from conversational chatbots to personalized, user-generated mini-apps. She explores the democratization of software creation, critiques voice-only hardware narratives, and shares founder reflections from her early AI work with Replica and OpenAI.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The host holds 4.1% of the talking time here. How this is scored →
The host as informed peer 2.5 Guest teaching 2.8 Guest disagreement 0.6 The host pushing back 0.1
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 46:40 Debunking Voice-Only Hardware
Eugenia forcefully attacks the consensus industry trend of building screenless voice AI devices, calling it a mind trap caused by misunderstanding the movie Her.
Hardest push from the host ▶ 41:50 Highlighting OpenAI's Strategic MisstepAnish pushes back on the assumption that major labs were always right by citing Karpathy's admission that OpenAI's pivot to video games was incorrect while Eugenia's focus was right.
Biggest teaching moment ▶ 46:40 Why Screenless Voice Interfaces FailEugenia systematically breaks down why voice-only interfaces are impractical in public spaces, offices, and sleeping environments, educating the hosts on UI affordances.
The host holds their own ▶ 11:37 Mass Market Software ThesisAnish demonstrates deep sector expertise by framing software bottlenecks around the world's 20 million developers and explaining consumer software downstream effects.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Career Journey: From AI Companions to Personal Software | 1 | 3 | 0 | 0 | Erik asks open contextual questions about Eugenia's career arc. Eugenia explains her move from AI companions to personal software, arguing that current chatbots are like MS-DOS for AI. | |
| Reimagining Software: From Fixed Apps to Personal Operating Systems | 2 | 3 | 0 | 0 | Erik and Anish ask about the shift to personal operating systems and ephemeral software. Eugenia illustrates how user-generated software will replace fixed apps with a story about creating a custom puzzle game for her daughter in two minutes. | |
| Replacing App Store Bloatware with Custom Mini-Apps | 3 | 1 | 0 | 0 | Justine shares her personal experience replacing bloated App Store utilities with custom Wabi mini-apps. Eugenia validates this with her own story of creating a custom weightlifting tracker. | |
| Social Graphs, Remixing, and Creator Dynamics on Wabi | 3 | 2 | 1 | 0 | Anish brings up creator ratios and video generation benchmarks like Sora. Eugenia explains Wabi's social graph features while Anish and Erik connect it to their thesis on mass market software creation. | |
| 'Vibe Designing' and Mini-Apps as Community Hubs | 2 | 3 | 0 | 0 | Eugenia explains Wabi's design philosophy of hiding technical complexity and framing creation as 'vibe designing' rather than coding, comparing the approach to Canva. | |
| Mobile Guardrails vs. Web Vibe Coding | 3 | 2 | 1 | 0 | Justine notes how consumer web coding tools can break easily and praises Wabi's mobile guardrails. Eugenia expands on why platforms like Wabi must act as an organizational layer rather than sending raw web links. | |
| Deep Personalization as AI's Breakthrough Primitive | 4 | 2 | 0 | 1 | Erik prompts Eugenia with a quote from Hiten Shah regarding Wabi as a context and memory framework. Eugenia agrees and details how deep personalization goes beyond simple skinning or isolated app sandboxes. | |
| Multiplayer Mini-Apps and Social Feeds | 3 | 2 | 0 | 0 | Anish asks about multiplayer social features, and Justine brings up how teens post unstructured text prompts on TikTok. Eugenia explains how mini-apps package complex prompts into frictionless interfaces. | |
| Removing Friction: Why Mini-Apps Beat Command-Line Chatbots | 2 | 3 | 1 | 0 | Anish asks if users will find mini-app sharing intuitive. Eugenia explains that graphical user interfaces are far lower friction than command-line prompt copy-pasting. | |
| Software as Content: The 100x Expansion of Mini-Apps | 4 | 2 | 0 | 0 | Erik states the venture thesis that the world only has 1% of the software it needs. Eugenia expands on software becoming content for niche creators and community hubs. | |
| Empowering a New Creator Class for Software | 3 | 2 | 0 | 0 | Anish asks about a professional creator class forming on Wabi. Erik compares creator potential to MrBeast selling physical products, while Justine adds perspective on usage analytics for software creators. | |
| The Early Days of Replica & Pre-Transformer AI | 1 | 4 | 0 | 0 | Erik asks Eugenia to reflect on early AI companion development. Eugenia educates the hosts on pre-transformer NLP, Word2Vec in 2012, and early Google dialogue papers. | |
| Early OpenAI Partnership & The GPT-3 Breakthrough | 2 | 4 | 0 | 0 | Anish listens as Eugenia recounts her early partnership with OpenAI, being shown GPT-3 in 2020 by Sam Altman and Mira Murati, and fine-tuning DaVinci models with Greg Brockman. | |
| Greg Brockman's Apartment & OpenAI's Early Office Culture | 2 | 3 | 0 | 0 | Anish prompts Eugenia to share memories of working out of OpenAI's early office. She describes visiting Greg Brockman's apartment and OpenAI temporarily shifting away from language models. | |
| Founder Lessons: Capital, Conviction, and Going Big | 3 | 3 | 1 | 1 | Anish notes Andrej Karpathy's observation that OpenAI's video game push was a wrong direction, highlighting that Eugenia was right. Eugenia reflects humbly on capital allocation, conviction, and her journalism background. | |
| Debunking Voice-Only Hardware & The AI-First OS | 2 | 5 | 6 | 0 | Erik asks Eugenia to speculate on hardware futures. Eugenia strongly rejects the voice-only hardware trend, calling it a 'mind trap' based on misinterpreting the movie Her. |
Statements from this episode (35)
Kuyda: AI should not be restricted to smartphone apps
“Right now, AI is just an app on your phone. It should not be that way.”
Kuyda: AI builders wrongly view voice as the main interface
“There's a huge mind trap that exists among builders in the space where they somehow think that voice is the main interface.”
Kuyda: OpenAI research shows one-third of ChatGPT usage is writing help
“OpenAI published one of those research one of those studies showing that these are, in fact, the main use cases for Chagipty. I think a third of all, all of the use was around writing writing help.”
Kuyda: AI chatbots are in their MS-DOS era before a Mac OS moment
“The current chatbots are really the Microsoft DOS era for AI interfaces, and that there will be some sort of a Windows Mac OS moment.”
Kuyda: Nearly one billion people use AI tools for simple tasks
“Almost a billion people using these AI tools already, but they only, only use them for These kind of simpler use cases.”
Kuyda: Software creation will shift to user-generated mini-apps
“And so I think the same will happen with software. Right now we're Kind of stuck in this world where are these few apps developed by professional developers? And then eventually, of course, we're going to move on to this new world where apps are built by all o…”
Kuyda: Future OSs will dynamically generate context-aware, personalized apps
“So if you think about it, the operating system of the future, you open it up and you see your regular popular apps you use all the time. Maybe it's X, maybe it's Instagram, whatever it is. And then you see maybe some cool apps that you discovered your friend, …”
Kuyda: Early Wabi users build niche apps unviable on App Stores
“Right now we're already seeing tons of, you know, some of our first users building very specific apps that would never exist in On the App Store. They're just too small. They're too personalized, too niche.”
Moore replaced over a dozen App Store utilities with Wabi mini-apps
“There were, like, probably over a dozen apps that I downloaded, and in some cases paid for, that I just totally didn't need anymore, because I had built better versions of it on Bobby.”
Moore: On-the-fly mini-apps will displace ad-heavy long-tail mobile apps
“Instead of having to find this long tail app that's running all these crazy ads, pop-ups all the time, and is hard to use and is not personalized to you can just make exactly what you want on the fly and tweak it”
Kuyda: Fully original app creators on Wabi will stay under 10%
“I'd say like fully kind of original creators probably still under 10%”
Acharya: There are only 20 million software developers globally
“And software has just been so restrictive because there's only twenty million developers in the world.”
Acharya: Personal AI software is a consumer product, not a developer tool
“I think unlike many of the other products, and you should comment on this, that we've seen, it feels like this is not text to app. You know, it's not a developer or developer adjacent tool. It's truly a mass market consumer product for non-technical people.”
Kuyda: Apple App Store will never become social due to privacy focus
“If today the app store is very much not social and it probably won't ever be social because of Apple's mandate for privacy and how important that is for them”
Moore: Vibe-coded websites are not ideal for daily personal software
“We needed to move beyond this paradigm of like, great, you vibe code a website and it exists out there. But like, is that the best interface for you to be using a product you want to use daily and storing personal information on and having your own records and…”
Kuyda: User-generated software will thrive, but standalone web links won't scale
“I do believe that we're all going to be building some sort of software and using software that everyone builds, some UGC software, but I definitely don't believe in links that people will share with each other and with me relying on some random person somewher…”
Kuyda: Top vibe-coded women's dating app leaked sensitive user data
“And we've already seen this with some apps, vibe coded apps reaching the top of the app store. I think the one around women dating that all the super sensitive information got leaked. And that wasn't because they are bad actors. It's just they're not professio…”
Kuyda: Deep personalization is AI's breakthrough primitive
“And I feel like for AI, the super special thing is personalization, but really deep personalization.”
Kuyda: Driving AI via paragraph text prompts is worse than MS-DOS
“We pass around these text prompts, which is almost like Microsoft DOS commands, but worse, because at least the commands were, it was like sort of like you could learn them. There were a few, whatever, like they were short, they weren't that long. And now are …”
Kuyda: Shareable mini-apps will replace raw text prompts in AI creation
“Instead of that, if you could just quickly, you know, click on the link in comments on TikTok and open a mini app Where everything's already set up for you, just add a photo. And you can see examples, you can choose different styles and this and that, and then…”
Kuyda: ChatGPT prompt subreddits have millions of members
“There's, there are Millions of people in the subreddits like chagiptypromptgenius, which I love those.”
Torenberg: 99% of needed software will be built in next 5 years
“On the investment team, we've said that there's, you know, the world has one percent of the software that it needs and that the rest is going to be built in the next five years.”
Kuyda: Software will evolve from fixed apps into interactive content
“I do think we'll see just a completely different type of software, not just apps, not just stale fix apps, but more content, community build community starters, conversation starters, and just fun little toys.”
Kuyda: Software creation is the last frontier for content creators
“If you're really thinking about it as YouTube, that's sort of the last frontier. At this point, creators can make their own professional content. They can make videos, they can make shows, they can write, but they still cannot create software.”
Kuyda: Modern video platforms prioritize polished commercial content over friends
“Right now, a lot of the video platforms are very polished and very commercial. You don't even see your friends anymore. You don't see that much weird content. You just see very curated, polished stuff.”
Kuyda bet Replika on dialogue language models in 2015
“When we saw that in August, 2015, we just basically put all of our bet everything we had at the company on building those models. We said, okay, well, this is it. This is right around the corner. We need to really focus on building these language models, getti…”
Kuyda: Replika was OpenAI's largest GPT-3 API customer
“We were the biggest customer in terms of API calls, because we were the only chatbot available on the market that actually used Genitive AI.”
Kuyda: Replika owned all major AI app store search keywords
“Before that, we literally owned all the keywords like AI, chatbot, artificial intelligence on every single platform.”
Kuyda: Replika owned hundreds of .ai domains and let them expire
“We also owned like hundreds of .AI domains, which I just let expire because I felt like, no one wants them.”
Kuyda: OpenAI temporarily abandoned language models for video game agents
“And very quickly they stopped working on language models, and we were very upset because we really wanted to continue going there, but they didn't want to talk about any language models because no one was really working on them and that may have made us feel v…”
Kuyda: Replika raised only $11M in total funding
“With Replica, we only raised eleven million dollars, and I guess it's still all in the bank and, you know, we went for many, many years and built a big business”
Kuyda: Leading AI builders are technical savants who lack empathy
“Today what I'm seeing with AI, especially it's being built by a very specific type of Person personality. It's oftentimes these savants, these like brilliant geniuses physicists, mathematicians, and they're insane in building algorithms and math and kind of sc…”
Kuyda: Building screenless AI hardware is a major mistake
“There's a lot of people trying to build voice only devices. In my case, completely wrong. Like it can be a fantastic extra way to interact with the computer, but every single Alexa right now is, like, 75% of them are being shipped with the screen. Because even…”
Kuyda: 75% of shipped Alexa devices now feature screens
“Every single Alexa right now is, like, 75% of them are being shipped with the screen.”
Kuyda: Next-gen AI hardware will be local AI smartphones, not voice gadgets
“I do believe that the AI device is not about a voice driven thing. It's more about building this AI first operating system. Having all the models run locally as well. I think that there's a lot in that, like building truly an AI first smartphone and, you know,…”