Today TechCrunch reported that startup Wabi raised $55M at a $1.25B valuation. The money came from IVP, Andreessen Horowitz, Forerunner, the Amazon Alexa Fund, and a whole lineup of star investors, including execs from Salesforce and Shopify.
The interesting part: just a month ago, Wabi was a completely different product. And its new version has existed for all of ten days.
Wabi's founder is Eugenia Kuyda. And here I want to give her credit upfront. In 2017, long before ChatGPT, she created Replika — one of the first mass-market AI companions. Today the service has tens of millions of registered users. Eugenia genuinely has a knack for spotting new forms of human-tech interaction before everyone else. And for raising very large sums of money behind her ideas.
But the Wabi story leaves me somewhat puzzled.
A mini-app builder that already exists everywhere
Originally, it was supposed to be a kind of YouTube for apps. Anyone could type what they needed, and the AI would build a small app you could share with friends — a habit tracker, a restaurant vote, a trip planner.
Great idea. Except you can already do this in Claude Code, Replit, Lovable, Rork, and a bunch of other tools. A landing page for a trip with friends can be built in literally a few minutes today, the link dropped into Telegram, and never touched again.
Why would you need a separate platform for that?
Second pivot: an agent that works for everyone
In September, Eugenia apparently decided to rethink the product again. Now Wabi is a personal AI agent and a messenger at the same time. It's supposed to help people organize their lives, spin up interfaces on the fly, and coordinate shared tasks.
Say you and your friends are planning a trip. Wabi helps come up with a route, build a schedule, pick restaurants, split up tasks — all inside a shared chat.
And this is where a genuinely interesting idea shows up: AI not for me, but for us. Not a personal assistant you talk to alone, but a shared intelligent intermediary between people.
But there's a problem. A pretty big one.
Competitors got there faster
While Wabi was morphing from an app builder into an agent, a whole different set of competitors showed up on this market.
Instinct can already handle real errands through ordinary messages: booking doctor's appointments, handling returns, organizing tasks. In September it raised another $1B at a $10B valuation.
Meta launched Muse. OpenAI launched Dots. These companies have their own models, massive infrastructure, and — most importantly — access to an existing audience. Instinct doesn't even require a separate app: you can talk to it through WhatsApp, iMessage, or email.
And I keep coming back to the same question.
Why would I switch to Wabi and invite my friends there if I'm already talking to them on Telegram or WhatsApp? And I can build whatever app I need in Claude or ChatGPT and just send the link?
Sure, the average user doesn't want to deal with code, hosting, or databases. And if Wabi can make creating and sharing apps a completely invisible process, the product has a shot.
But right now I don't see a convincing answer to why any of this needs to happen specifically inside Wabi.
The company talks about strong initial interest but doesn't disclose concrete numbers on active users or retention. Meanwhile investors are already valuing it at over a billion dollars.
Maybe they're seeing something we're not. Or maybe they're betting primarily on Eugenia, who was already right once about the future of consumer AI.
And I'm fully prepared to be proven wrong — she might be right again. I really like the idea itself: personal software that spins up for a specific task and disappears once it's no longer needed.
But there's a paradox here.
If apps become so cheap and easy to make that you can build them anywhere, why do we need yet another platform for building them?