Replika founder raises $20 million for Wabi, the ‘YouTube of apps’

Eugenia Kuyda saw the future of consumer AI before most. He founded Replika, the first major AI startup-partnership, in 2017 — years before ChatGPT started. Today, it has 35 million users.

Now Kuyda is back with a new startup called Wabi, which he describes as a YouTube for apps – a social networking platform where anyone can use prompts to instantly create mini-apps and share them with friends. Wabi, which went into beta last month, heralds another shift in consumer AI: one where personalized software becomes the norm.

Wabi raised $20 million in pre-seed funding from an amazing list of angels including AngelList co-founder Naval Ravikant, Y Combinator CEO Garry Tan, Twitch co-founder Justin Kan, Replit CEO Amjad Masad, Notion co-founder Akshay Kothario, Sarah co-founder Neuravi. Guo.

“[Kuyda] it was early and right for AI partners, even though it wasn’t obvious at the time,” Anish Acharya, general partner at Andreessen Horowitz, told TechCrunch. “It’s very rare to find someone who has a track record of anticipating what consumers will want, and we believe they’re doing it again.”

Kuyda is entering a hot market. Vibe coding tools like Cursor and Lovable have attracted significant VC interest, while code-free AI platforms including Emergent, Replit, and Bloom are racing to enable non-technical users to build messaging apps. The Wabi difference: a complete platform for creation, discovery and hosting — no app store required.

“This is really about helping people who have nothing to do with coding or the tech world very quickly build apps from their everyday lives,” Kuyda, who last week joined us on stage at Disrupt to discuss AI companions, told TechCrunch. “All you have to put in is ‘build me an AI therapy app’ and that’s it. It’ll suggest features and you can brainstorm, but it’ll build you an app. You don’t have to be great at prompting. You never see the code.”

Earlier this week, Wabi rolled out some social features to beta users – like the ability to like, comment and mix on any existing app, as well as check out user profiles to see what others have liked, used or built.

X has been blowing up Wabi ever since it started giving out invites to select users. Several founders, designers, and investors from around the world have posted about Wabi’s ease of building apps for yourself. Even the Google DeepMind product lead Logan Kilpatrick yelled at Wambi.

“We think the social layer is absolutely critical because it allows for so much more creativity and discovery, and these mini apps become community starters or conversation starters,” Kuyda said.

Wabi’s Explore page currently features recent and popular apps, though Kuyda said it will become more algorithmic over time. The startup plans to roll out personalized onboarding in the coming weeks, automatically creating starter apps for new users.

Wabi’s core promise isn’t much different from ChatGPT’s GPT store or the bot from Quora’s Poe: Build mini-apps using messages that could solve small problems for you. Apps like Wabi have been able to package this promise well in terms of customers not having to build any technical infrastructure. Even if you enter a few sentences, Wabi handles things like creating an icon or setting up databases and deciding what the app’s user interface will look like.

Kuyda told TechCrunch that for apps that require anything to be built with AI, users can open the settings and choose their base model (like if they want to use ChatGPT or Gemini) and even rewrite the prompts that Wabi displays.

Creating a basic application is simple. However, you may need to debug the application to avoid errors, which is to be expected in a development lifecycle.

For example, we created an app that showed us a photo of a dog each day with a dog event. After a few days of use, we realized that the app was generating the same set of dogs. When we looked at another user’s daily news app, all the dates listed in the snapshots were October 1, 2023, and the news was a few weeks old. Furthermore, one of the news sources was, surprisingly, Wikipedia.

The onus is on the user to take care of maintaining the applications. Otherwise, you may find many unmanaged mini apps in the find section of these vibe coding apps.

Kuyda says it’s still early days for Wabi, and they’re still working out how to make sure the apps are ready out of the box. He noted that there are still model limitations that are being improved every day. He says a large portion of the $20 million will go toward building out Wabi’s product team.

Part of the funding is also intended to effectively subsidize the use of Wabi until the startup finds a monetization model. Kuyda says she is not interested in hosting ads on the platform, which leads to incentives that create dark patterns.

“I built Replika and never had any ads,” he said. “I think ads just create a really bad user experience. I like to create pleasant user experiences.”

Acharya believes that once network effects take off, it will be easy to make money. He sees a future where there will be an element of professionalism that happens on the platform, where many of the kids today who want to be TikTok stars could instead make software on Wabi.

“Think about the history of YouTube, it started as people putting on these choppy low-budget content experiences,” he said. “Now, 20 years later, it’s extremely high production value.”

Acharya added that there are even more opportunities with software because “video content has a declining value over time,” he said. “Software has compound value.” If someone creates the next successful app, it will continue to be relevant over time.

The idea fits perfectly with Acharya’s thesis about the future of “disposable software” — small, flexible apps that people can create and discard as easily as opening a new tab or having a quick chat with ChatGPT.

“I think software is the final frontier of participation,” Acharya said. “The Internet has been this driving force for participation … where anyone can post their thoughts. It’s kind of weird that the Internet is apparently all software, but so few people have been able to do that.”

So what does a Web 3.0 look like when everyone can create and share software in minutes?

“It seems the internet has become somewhat clinical – we all use the same Instagram, the same TikTok, we all have the same home screens, the apps have become quite monotonous,” he said. “I think the opportunity with Wabi is that it will restore some of that punk, weird, early ’90s fabric.”