Wabi Personal AI Agent Raises $55 Million at $1.25 Billion
Key Takeaways
- Wabi raised $55 million on October 29, 2026, at a $1.25 billion valuation, in a round led by IVP.
- Participants included Andreessen Horowitz, Forerunner and the Amazon Alexa Fund, plus angels including Tobi Lütke, Marc Benioff, Marissa Mayer, Dick Costolo and Alex Pall.
- Founder Eugenia Kuyda previously founded the AI companion startup Replika.
- Last month Wabi pivoted to Wabi 2.0, an AI messenger and personal agent that builds task-specific mini-apps.
- The valuation is an investor bet on the new agentic direction, not proof that the product has succeeded at scale.
Investors are spreading bets across the race to build a winning personal AI assistant, and Wabi is the latest to benefit. On October 29, 2026, the startup said it had raised $55 million at a $1.25 billion valuation, in a round led by IVP. The money follows a pivot announced last month, and it is best read as a bet on that pivot — not as evidence that the new product has already won users at scale.
Wabi was founded by Eugenia Kuyda, who previously founded the AI companion startup Replika. It originally focused on allowing anyone to build apps using prompts. Last month, however, the startup announced a pivot of sorts to become an AI messenger. The new Wabi, or Wabi 2.0, is described as a personal agent that builds the interfaces people need to get things done, as part of its role as their assistant.
What Wabi 2.0 is supposed to do
The idea is that instead of opening a different app for each task, a user talks to Wabi, and Wabi builds a small, task-specific interface — a mini-app — for that job. The company suggests Wabi could be used to build mini-apps for things like meal planning, tracking local restaurants, planning date nights, creating fitness plans and designing specialised trivia games. The interface is generated for the task, rather than chosen from a fixed catalogue.
That distinction is the heart of the pitch. Traditional assistants answer questions or trigger fixed actions. Wabi’s claim is that the assistant should also build the tool for the job, on demand. If it works, the user experience could feel less like managing apps and more like describing an outcome. If it does not work reliably, the generated interfaces could become another layer of friction — pretty, but slower or less trustworthy than the app they replace.
IVP general partner Alex Lim, in a statement quoted in the funding coverage, said Kuyda has a clear vision for how people will interact with AI, one that goes far beyond the text box or command line, and that Wabi is the expression of that vision. That is an investor’s endorsement, not an independent evaluation of the product, and it should be weighed accordingly.
Who is backing it, and why that matters
Other participants in the round included Andreessen Horowitz (a16z), Forerunner and the Amazon Alexa Fund, plus angel investors including Shopify chief executive Tobi Lütke, Salesforce chief executive Marc Benioff, former Yahoo head Marissa Mayer, former Twitter chief executive Dick Costolo and Alex Pall of The Chainsmokers. The breadth of that list — venture firms, a platform fund and prominent operators — suggests the round was competitive, but it does not disclose how much each participant invested, nor what milestones the valuation depends on.
The funding also arrives in a hot market for AI agents. The coverage notes viral AI agent Instinct raising another billion at a $10 billion valuation in September, and companion-maker Tolan raising $30 million. In that context, Wabi’s $1.25 billion valuation looks less like an outlier and more like part of a wider pattern: investors are paying high prices for early-stage agent companies, on the theory that one of them will become the default personal assistant. That theory may be right, but it is still a theory, and most early bets in such races do not win.
The agent question is broader than one company. Plato’s simulated workplaces address how agents can be trained and tested before they touch real systems, while GPT-6.1 Terra’s computer work shows how agents are increasingly expected to operate software directly. Wabi sits on the consumer side of the same shift: an agent that does not just answer, but builds.
What would prove the pivot works
A useful way to judge Wabi 2.0 over the coming months is to watch for evidence that is hard to fake with a demo. Do users return to the product after the first week, do they complete real tasks through generated mini-apps rather than reverting to the original apps, and do those mini-apps handle errors gracefully when a booking, a plan or a payment goes wrong? Retention, task completion and error handling are the measures that separate a genuinely useful agent from an impressive prototype. None of those measures was available in the funding announcement, which is why the valuation should be treated as a starting point for evaluation, not as a verdict on the product.
Limitations and unanswered questions
The announcement leaves the most important product questions unanswered. No user numbers, retention figures, revenue figures or mini-app quality metrics were disclosed in the coverage reviewed. No independent test of Wabi 2.0’s generated interfaces was available. The pivot is only a month old, so there is little public track record to judge it by. The $1.25 billion valuation is a price investors agreed for this round; it is not a measure of current revenue, profit or market share, and it should not be quoted as one.
For readers, the sensible stance is curiosity without credulity. Wabi’s pivot is an interesting idea, and the investor list gives it resources to pursue it. Whether generated mini-apps become a better way to get things done — or just a novel demo — will depend on reliability, speed, privacy and whether users actually return to the product after the first try.
