The startup Wabi has repositioned its product from a tool that creates apps through text prompts to a personal artificial intelligence agent that combines conversation, task execution, and the creation of user interfaces when needed. The new product is called Wabi 2.0, and it is currently available through invitation codes distributed via X.
Founder Eugenia Kuyda, who previously founded the smart companion company Replika, described the new release as “a personal agent that does things on your behalf and builds the interface you need at the right moment.” This means that users do not interact with only a chatbot; they can also receive an interactive interface connected to the task they request.
From Building Apps to Executing Tasks
Wabi had focused on enabling anyone to create apps using text prompts, within the category of natural-language-based programming tools. In the new release, the company seeks to place the app and the agent within a single experience, rather than presenting itself solely as a direct competitor to app-building tools.
According to the vision put forward by Kuyda, a user might ask the agent to accomplish a task, prompting it to create an interface suited to that request. Proposed use cases include creating a calorie tracker, a weightlifting exercise log, and a weight tracker within a health-related conversation.
Other examples include creating a calendar for children’s activities, a language-learning app, and a tool for tracking family-friendly local events, with these interfaces linked to ongoing contexts or conversations about family, household management, and children’s interests.
Why Does This Change Matter?
The fundamental change does not lie in adding a single feature, but in redefining where the app belongs within the AI experience. Instead of carrying out a long series of requests within a single chat history, Wabi is attempting to turn the agent’s results into interfaces that can be opened and used visually, such as lists, calendars, and tracking tools.
Kuyda says that agents based solely on conversation may become less practical as the history and ongoing tasks accumulate within a single stream. In her view, the more powerful agent will combine conversation and software, so that the apps, tasks, and people connected to the user’s life are in one place.
Competitive Context and Current Limitations
This shift comes amid growing interest in AI agents, including Meta’s products called Muse and Instinct, and as companies such as OpenAI move toward integrating interactive experiences into general-purpose chatbot interfaces. The article stated that OpenAI announced an update to ChatGPT apps during a DevDay event that allows apps to provide interactive experiences within ChatGPT, while highlighting app names in the sidebar.
However, Wabi 2.0 is not yet generally available; access is limited to users who obtain invitation codes. The article also does not provide technical details about how the interfaces are built, the limits on what the agent can execute, or data and privacy policies—points that will practically determine how suitable this model is for everyday use.