Direct answer: Wabi, a startup that describes itself as a personal agent, has raised $55 million at a $1.25 billion valuation. The idea is simple to state and hard to build: instead of one person typing into one chat window, Wabi builds small, purpose-made interfaces — mini-apps — and shared spaces where friends, couples, families and groups can plan things together.
What happened
Wabi announced a $55 million funding round on 29 October 2026, valuing the company at $1.25 billion. The round was led by IVP. Other participants named in the announcement included Andreessen Horowitz (a16z), Forerunner and the Amazon Alexa Fund,
alongside angel investors Tobi Lütke, Marc Benioff, Marissa Mayer, Dick Costolo and Alex Pall.
That list of names is worth pausing on, because it tells you something about how the round is being read in Silicon Valley. IVP is a long-established venture firm. Andreessen Horowitz is one of the most closely watched firms in AI. The Amazon Alexa Fund’s presence points to an interest in how people actually live with voice and ambient computing, not just how they use a website. And the angels are people who have run large consumer platforms themselves: Lütke founded Shopify, Benioff founded Salesforce, Mayer ran Yahoo, Costolo ran Twitter, and Pall is part of The Chainsmokers. None of that guarantees Wabi will succeed. It does show that experienced operators were willing to put their names to this particular bet.
The timing matters too. Wabi had a public launch in mid-October 2026, about ten days before the funding announcement on 29 October. In other words, this is not a round raised on a slide deck alone. It follows a real product reaching real users, even if the announcements covered here do not give user numbers, revenue figures, or a detailed breakdown of how that launch went. Those gaps matter, and we will come back to them in the limits section below, because a $1.25 billion valuation invites questions that a press announcement alone cannot answer.
Wabi was founded by Eugenia Kuyda, who previously founded Replika, the AI companion startup. That history is central to understanding Wabi, and it is also the source of one of the most interesting tensions in the story. Replika became known for one-to-one companionship: a person talking to an AI. Wabi moves in a different direction. Its argument is that personal AI will matter most between people — in the group chat, the household, the running club, the book club — rather than in a private conversation between one person and one machine.
TechCrunch reported the round and the valuation, and the company announcement carried by PRNewswire set out the investor list and the product vision in the company’s own words. TechCrunch also placed Wabi in a wider market context, which we cover below. Both sources agree on the headline numbers: $55 million raised, $1.25 billion valuation, led by IVP.
It is worth being precise about what has not been claimed. The sources supplied for this article do not state a total amount raised across all rounds, so this article does not invent one. They do not give user numbers, revenue, retention, or a launch date for any specific country, so this article does not invent those either. When a startup story is moving fast, the temptation is to fill gaps with confident-sounding detail. The honest version is to say what is known, say what is not known, and explain why the known part still matters.
How Wabi actually works
Wabi combines three things: an AI agent, personalised mini-apps, and shared spaces. Each part on its own sounds familiar. Put together, they describe something different from a standard chatbot, and different again from the product Wabi started out as.
Start with the agent. Wabi describes itself as a personal agent, and the current version is sometimes called Wabi 2.0. The job of the agent is not only to answer questions. It is to understand what you are trying to do and then build the interface you need for that task. If you want to plan meals for the week, you do not just get a list of suggestions in a chat thread. You get a mini-app for meal planning. If you want to keep track of local restaurants you like, you get something built for that purpose. If you are planning date nights, working through a fitness plan, or running a specialised trivia game, the interface is shaped around that activity.
That is the second part: mini-apps. Suggested uses mentioned in the coverage include meal planning, tracking local restaurants, planning date nights, fitness plans and specialised trivia games. The point of a mini-app is that it is small, specific and disposable in a good way. You do not need to download a separate app, learn a new system, or pay for a subscription for every little job in your life. The agent builds a lightweight tool for the job, you use it, and you move on. In principle, that lowers the friction that stops people using software for the small, social, everyday tasks that never quite justify a full app.
The third part is shared spaces, and this is where Wabi’s argument gets most distinctive. A shared space is a place where a group can use a mini-app together. Friends planning an evening out, a couple organising a household, a book club choosing its next read, a running group sorting out routes and times, a family sharing the household load — these are all groups where the work is coordination, not just information. Someone has to suggest, someone has to respond, someone has to keep track, and everyone has to see the same plan. Wabi’s bet is that personal AI should live in that shared space, helping the group decide and organise, rather than sitting in a private chat with one person at a time.
Kuyda put the philosophy in a single sentence in the announcement: “AI shouldn’t just save people time. It should help them spend it better, doing more of what they love, with the people they love.” That sentence is doing a lot of work. It reframes AI from a productivity tool — faster answers, fewer keystrokes — into a social tool. Time saved is only useful, on this view, if it turns into time spent well, and time spent well usually involves other people.
Alex Lim of IVP, the lead investor, praised Kuyda’s sense of where AI is headed and a vision that goes beyond the text box or the command line. That phrase — beyond the text box — is the clearest summary of the product argument. For the last few years, the dominant way most people have met AI is a box you type into and an answer that comes back. Wabi is arguing that the next step is not a better box. It is software that builds the right surface for the task, whether that surface looks like a planner, a tracker, a game, or a shared list, and that lets a group stand around that surface together.
It helps to contrast Wabi with two things it is not. First, it is not simply a chatbot with a nicer personality. A chatbot is organised around conversation: you ask, it answers, the thread grows longer. Wabi is organised around tasks and groups: you have something to arrange, and the product builds a tool and a shared place for arranging it. Conversation may still be part of it, but conversation is not the whole product. Second, Wabi is not, in its current form, a general tool for anyone to build apps using prompts, which is how it originally started. That earlier version let people build apps by describing them. The shift to Wabi 2.0, an AI messenger that builds the interfaces you need as part of being your personal agent, narrows the focus. Instead of asking users to become app builders, Wabi builds the app for them, in the moment, for the task at hand. That is a meaningful change in who the product is for. The first version asked for creativity and effort from the user. The current version asks for a task and a group, and takes on the building work itself.
There are practical questions this raises, and they are worth stating plainly even before we get to the formal limits section. How good are these mini-apps, really? A meal planner built on the fly is only useful if it remembers your constraints, handles changes without breaking, and stays simple enough that a group will actually use it. A trivia game is only fun if the questions are good and the format holds up over repeated play. A shared space is only helpful if everyone in the group can get into it easily, without creating accounts, downloading apps, or learning new habits. None of those questions are answered by a funding announcement. They are answered by using the product over weeks, with real groups, doing real planning, and seeing what survives contact with busy lives. That is the test Wabi now faces, and it is a harder test than raising money.
Why investors paid a $1.25 billion valuation
A $1.25 billion valuation for a company that launched publicly about ten days before its funding announcement will strike many readers as extraordinary. It is worth asking, carefully and without hype, what investors think they are paying for.
The first part of the answer is market context. TechCrunch noted that the viral AI agent Instinct raised another billion at a $10 billion valuation in September, and that companion startup Tolan just raised $30 million. Those are very different companies, but together they paint a picture of a market where investors are willing to pay large prices for consumer AI products that show signs of catching on. Instinct, at a $10 billion valuation, suggests that a consumer agent product with momentum can command a very large number indeed. Tolan, at $30 million, shows that the companion space Kuyda knows well is still attracting fresh capital. Wabi sits between those poles: a consumer AI product, with a social twist, at a valuation that is large but far smaller than Instinct’s.
The second part is the founder. Kuyda founded Replika, one of the earliest and best-known AI companion products. That gives her two things that are hard to buy: experience building consumer AI that people form habits around, and a record of understanding what users actually want from a personal AI, as opposed to what technologists think they should want. Investors often say they back founders as much as ideas, and in this case the founder’s history is directly relevant. She has already built in the personal AI space, seen what works and what does not, and is now making a different bet: that the future is social, not solitary. IVP’s praise for her sense of direction fits that reading. It is not just praise for a product demo. It is a bet on judgement formed over years in this exact field.
The third part is the size of the prize if the bet is right. If personal AI becomes something people use every day, with their friends and families, then the company that owns that layer could be very large. It would sit between people and their plans, their households, their leisure and their social lives. That is a position of real commercial value, whether the money eventually comes from subscriptions, partnerships, or something else entirely. Investors paying $1.25 billion are not valuing today’s revenue, which has not been disclosed. They are valuing the chance that Wabi becomes that layer for a large number of people. Valuations like this are always, in part, a price placed on a possible future.
That is also why the valuation should be treated with caution. A valuation is not a measurement of worth in the way revenue or profit is. It is a negotiated price for a minority stake, agreed between a company and investors who share an interest in the number looking strong. It reflects optimism, competition between investors to get into the round, and a belief about the future. It does not prove that the product works well, that users will stay, or that the business model will hold up. Many companies have raised at large valuations and then struggled. Some have raised at modest valuations and grown steadily. The valuation tells you what a small group of sophisticated investors were willing to pay on one day in October 2026. It does not tell you what Wabi is worth to an ordinary user deciding whether to try it, or to a small business deciding whether to pay attention.
There is also a competitive shadow over the whole story. Bloomberg, as noted in the TechCrunch coverage, counts OpenAI and Meta among the competitors in personal assistants. That is a serious list. OpenAI and Meta both have enormous distribution, vast resources, and their own plans for assistants that help people across messaging, search, and everyday tasks. Apple and Google, too, are building assistants into the phones and operating systems that most people already carry, and that competitive pressure is part of the honest picture even though the supplied sources name OpenAI and Meta specifically. A startup in this space is not competing in an empty field. It is competing against companies that can put an assistant in front of billions of people without asking them to download anything new. Wabi’s answer has to be that it does something those giants do not do well: build small, social, task-specific tools for groups, with a feel and a focus that a general assistant cannot match. That is a plausible answer, but it is an answer that has to be proved in use, not just stated in a launch announcement.
What it means for UK small businesses
For a UK small business, Wabi is not a tool to buy today. UK availability and pricing were not confirmed in the announcements covered here, and this article makes no claim about either. What Wabi is, instead, is a useful signal about where customer-facing AI is heading, and a set of practical ideas that any small firm can use now, whatever happens to this particular startup.
The first lesson is about mini-apps. Many small businesses still present themselves to customers as a single, general thing: a website, a social media page, a chat thread. Wabi’s bet is that customers actually want something narrower: a small tool for a specific job, at the moment they need it. A restaurant does not need a general AI. It needs a way for a group of friends to pick a time, see a menu, and book a table without twenty messages. A fitness studio does not need a chatbot. It needs a way for members to see a class, check who else is coming, and reserve a place. A local shop does not need a virtual assistant in the abstract. It needs a way for a customer to check stock, reserve an item, and arrange collection. The mini-app idea is really just that principle applied properly: build the smallest useful tool for the job, and make it easy to share. Small businesses can apply that idea today with simpler tools, without waiting for Wabi or any other startup. The question to ask is not “should we have AI?” It is “what is the one small job our customers keep struggling with, and what is the smallest tool that would make it effortless?”
The second lesson is about group planning. Much of the spending that matters to small businesses is social: meals out, classes, trips, celebrations, gifts bought together, services booked for a household. Yet most business tools are built for one person at a time. A booking system assumes one booker. A shop assumes one buyer. A class timetable assumes one member. Wabi’s shared spaces point to a gap. If a business can make group decisions easier — if it can let a group agree, book, and pay with less friction — it can win work that a competitor loses to indecision and endless messaging. That does not require an AI agent. It can be as simple as a shared booking link, a group voucher, a clear way to split a bill, or a timetable that shows who else is attending. The technology matters less than the insight: customers often decide together, so make it easy to decide together.
For hospitality and leisure, the application is immediate. A café, restaurant, or pub lives or dies on group bookings. A tool that lets a group pick a date, see availability, agree on a menu or a set price, and confirm without a chain of phone calls would be worth more than a general chatbot that answers opening hours. Even without Wabi, a business can move in that direction: a single booking page that handles groups properly, clear information about capacity and dietary options, and a confirmation that everyone in the group can see. The lesson from Wabi is not the technology. It is the focus on the group’s decision, not just the individual’s enquiry.
For fitness and wellbeing, the parallel is clear. Classes, courses, and memberships are often social. People join because friends go, or because a partner or family member goes too. A shared space where a small group can see a timetable, plan attendance, and stay motivated together is a more useful idea than a one-to-one chat with an AI about exercise. Again, the point can be applied without Wabi: a class page that shows who is booked, a simple way to invite a friend, a group challenge or a shared plan for a term. The technology is secondary. The social structure is the product.
For retail and local services, the idea translates differently but still usefully. A customer choosing a product, a gift, or a service often checks with someone else: a partner, a parent, a friend, a colleague. A product page that is easy to share, with clear information, clear prices, and a simple way to reserve or buy, helps that conversation. A service business can do the same: a clear booking flow, a way to share options with a household member, and a confirmation that everyone can see. Wabi’s mini-apps are, at heart, a reminder that the customer is rarely alone. Businesses that design for the group, not just the individual, will often convert better.
There is a final, practical point about shareable information. Wabi’s shared spaces work only if the information inside them is clear, current, and trustworthy. The same is true for any small business tool. A mini-app, a booking page, or a shared plan is only as good as the information it contains. Prices must be right. Availability must be real. Terms must be clear. If a group makes a decision based on wrong information, the business pays for it in complaints and lost trust. That is a mundane point, but it is the one that matters most in practice. AI tools can make information easier to share, but they cannot make wrong information right. The discipline of keeping information accurate remains the business’s job, whatever tool it uses.
There are also limits to what a UK business should take from this story. First, Wabi is not a proven product at scale. It launched publicly in mid-October 2026, and the funding was announced about ten days later. That is a very short track record. Second, the shared-space idea raises obvious privacy questions. A space shared by a group will contain information about several people, and any business thinking about similar tools needs to be clear about who can see what, how data is stored, and how a person can leave a space or delete their information. Under UK GDPR, those are not optional extras. They are core design questions, and getting them wrong can damage trust and create legal risk. Third, the business model is not yet clear. It is not stated in the supplied sources how Wabi will make money at scale, and a small business should not build a plan around a product whose pricing and terms are unconfirmed.
For readers who want to go further, the practical next step is not to wait for Wabi. It is to look at the small, specific jobs where AI agents can already help a UK business, and to test one of them properly. That might be handling enquiries, qualifying leads, booking appointments, or following up after a sale — the kinds of repetitive, well-defined tasks where an agent can save time without taking over judgement. Our guide to AI agents for small business sets out how to think about that in practical terms, and our coverage of OmniChat’s $10m Series A for AI agents in chat commerce shows a different, more immediately commercial example of agents being used where customers already are. Wabi points to a possible future. Those examples show what is already usable today.
Honest limits / what to watch
It is easy to be carried away by a $1.25 billion valuation. It is harder, and more useful, to be clear about what Wabi has not yet proved.
First, a valuation is not proof of a product. As noted above, it is a price agreed between a company and its investors. It does not tell you how many people use Wabi every day, how long they stay, or whether they would pay for it. Those numbers have not been disclosed in the sources supplied for this article, and until they are, any judgement about Wabi’s success is provisional.
Second, real use is still to be proved. Wabi launched publicly in mid-October 2026. That is a very short time for habits to form, for problems to emerge, and for the product to be tested by ordinary users rather than early adopters. The mini-app idea sounds appealing. Whether mini-apps are consistently useful, whether they are easy to create, and whether people actually return to them are questions that only time and use can answer. The early coverage describes what Wabi is intended to do. It does not yet show, in detail, what it does in daily life for a large number of people.
Third, privacy in shared spaces is a genuine challenge. A shared space, by definition, involves more than one person. That raises questions that a one-to-one chatbot does not: who can see what, who controls the space, what happens when someone leaves, and how personal information is handled when it is shared by design. These are not abstract concerns. They affect whether people will trust the product with the very things — plans, preferences, household details, social connections — that make it useful. For UK users and businesses, UK GDPR adds a further layer: any tool that handles personal data in shared spaces needs clear, lawful handling of that data, and users need to understand what they are sharing and with whom. Wabi’s approach to these questions will matter as much as its product design, and it is not yet fully described in the supplied sources.
Fourth, competition is formidable. As noted above, OpenAI and Meta are counted among the competitors in personal assistants, and Apple and Google are building assistants into the devices and systems people already use. Wabi does not need to beat all of them at everything. It needs to be clearly better at one thing — social, task-specific mini-apps for groups — than a general assistant that tries to do everything. That is a demanding standard, because general assistants are improving quickly, and because users often prefer one familiar tool to several specialised ones. If a general assistant adds a good enough version of shared planning, Wabi’s advantage narrows.
Fifth, the business model is unproven. It is not stated in the supplied sources how Wabi will make money at scale. Subscriptions, partnerships, and other models are all possible, but none is confirmed here. A product can be loved and still fail as a business if the economics do not work. Investors have funded the attempt. They have not yet proved the model.
Finally, there is a broader question about what personal AI should be. Wabi’s vision is optimistic: AI that helps people spend time better, together. That is an attractive idea. It also raises questions about dependence, attention, and what happens when a tool that organises your social life becomes indispensable. Those are not reasons to dismiss Wabi. They are reasons to watch how it develops, and to ask, as users and as businesses, what kind of personal AI we actually want. The most useful response to a $1.25 billion valuation is not awe, and not scepticism for its own sake. It is curiosity, tempered by the questions above.
What would prove, or break, the bet? A few markers are worth watching. First, retention: do people keep using Wabi after the novelty wears off, and do groups keep returning to the same shared spaces? Second, mini-app quality: are the tools people build actually good, or do they feel generic and shallow? Third, privacy and trust: does Wabi handle shared data in a way that users understand and accept, including under UK GDPR for UK users? Fourth, a business model: does a clear, sustainable way of making money emerge? Fifth, differentiation: does Wabi stay clearly different from general assistants, or does it get absorbed into features that larger companies offer for free? If those markers turn positive, the valuation will look prescient. If they turn negative, it will look like another example of enthusiasm outrunning evidence. For now, the honest position is in between: an interesting product, a credible founder, a large bet, and much still to prove.
FAQs
What is Wabi?
Wabi describes itself as a personal agent. It combines an AI agent with personalised mini-apps and shared spaces, so groups such as friends, couples, book clubs, running groups and families can plan and organise things together, rather than each person typing into a separate chat.
How much did Wabi raise, and at what valuation?
Wabi raised $55 million at a $1.25 billion valuation. The round was led by IVP, with participation from Andreessen Horowitz, Forerunner, the Amazon Alexa Fund and several well-known angel investors.
Who founded Wabi?
Wabi was founded by Eugenia Kuyda, who previously founded Replika, the AI companion startup. That background matters because Wabi’s bet is about personal AI used between people, not only one-to-one conversation with a chatbot.
How is Wabi different from a chatbot?
A chatbot mainly answers in a text thread. Wabi is designed to build the interface you need for the task at hand, such as a mini-app for meal planning or tracking local restaurants, and to give a group a shared space to use it together.
Is Wabi available in the UK, and what does it cost?
UK availability and pricing were not confirmed in the announcements covered here, so UK readers should treat both as unconfirmed for now. This article does not make any claim about a UK launch date or price.
