The Guy Who Sold to GM for $2.1B Is Building a Gaming Social Network

01 /The Concept

When Kyle Vogt – the founder who built Cruise into a $2.1 billion acquisition by General Motors (with a full merger completed this year) – announces a new startup, the market he's targeting tends to have at least several billion dollars in headroom.

Fifth Door is a platform where people can build games with AI and play them. The platform hasn't launched yet, but the company just announced a $20M seed round.

Fifth Door functions as a social network for game enthusiasts. Users can build games without any programming using AI tools, remix other people's games, and play a daily stream of experiences that appear in their feed – similar to how a social feed works, but with games as the currency instead of posts.

Here's how the founder explains the platform's reason for being:

As social isolation increases and in-person interaction becomes rarer, the need for fun that's connected to real human relationships has never been greater.

Games have always delivered that. But creating games was a one-way street – only trained engineers and dedicated studios could do it.

We're entering a new era of gaming. Anyone can now build games with AI. You don't need a studio.

The line between game creator and game player is dissolving. If you genuinely love games, you can now do both – make them and play them. The only resource required is imagination.

And the scale of the audience you're building for is completely optional. You can make a game for your family, for your college friends, for your neighborhood group – full of inside references only they would understand. That's where games are going. And that's where Fifth Door starts.

02 /Why It Matters

Good ideas rarely occur to only one person. A very similar platform is being built by Y Combinator graduate Sagaland, which published its launch information just two weeks ago.

Sagaland's specific focus is "playable books" – text-based quests that users navigate as readers. The same core insight applies: users don't just play these quests, they create them using the platform's AI engine.

And Then ([covered here](/review/a-vot-jetim-malo-kto-zanimaetsja-nu-i-ochen-zrja)) graduated from a16z's accelerator in October and operates in the same space, but with a voice-first approach. Players can navigate quest storylines entirely through speech, with AI dynamically generating responses adapted to each player's choices.

Zooming out further, Wabi ([related review](/review/zabud-pro-app-store)) – which raised $20M in a round led by a16z in early November – is building the generalized version of the same idea under the banner of "the era of personal software." The difference is scope: Wabi covers any type of software, Fifth Door only covers games.

Both platforms are expressions of the same trend, which was articulated in August by an a16z partner in an essay titled "Disposable Software."

The argument: software creation used to be expensive, slow, and inaccessible. AI has made it fast, cheap, and available to anyone – including people who have never written a line of code. This makes it possible to build software "for the occasion" – for a specific purpose, for a handful of people, for a moment. Everyone can now be a creator, not just a consumer.

The next evolution of this trend is turning software into a medium for social interaction. People create, others use, others remix for their own creations – the same cycle that governs how content works in any social network.

The same paradigm shift is playing out in video. AI has dramatically reduced the barrier to video creation, and OpenAI's Sora app – which feels like TikTok but where clips are generated rather than filmed, complete with the ability to insert yourself and your friends – is one early expression of that.

03 /Opportunities

There's no shortage of AI platforms for building software, games, and video. What distinguishes the startups described today is that they're testing the ceiling – exploring what these AI creation tools enable beyond just making individual artifacts. Can they become the foundation for a new generation of social networks? A new kind of app marketplace?

Wabi's bet is that people will eventually remember the App Store the way they remember cable television – a curated, limited channel lineup produced by expensive professional studios. On that timeline, Wabi is YouTube to the App Store's broadcast TV.

These paradigm shifts can happen in unexpected categories. Here's a rough illustration of the pattern:

Many authors today use AI to turn a few core ideas into a full-length text with explanations and examples. Many readers use AI in the opposite direction – to turn a full book into a short summary. Some then ask ChatGPT how to apply the book's key ideas to their specific situation.

That creates a strange gap worth examining. Why not build a platform where authors publish only key insights – and readers receive AI-generated texts of whatever length they want, adapted to their own interests and context? Every reader would get a book written individually for them, built on the author's unique perspective and ideas.

Something like this is already happening on And Then. Quest authors provide only the story structure. The AI generates the actual narrative, adapted to each player's responses in real time. Every player effectively plays a different game – and even the same player gets a different experience every time, depending on their choices.

The broader point: AI isn't just a tool for making existing things more efficient. It's a medium for creating things that couldn't exist before. Personalized books, for example, exist in limited form as consulting relationships – but at scale and at low cost, they simply weren't possible without AI.

So the question worth spending time on: what new things can be built with AI that couldn't be built before? Because someone else is already asking the same question.