📰 Key Takeaways

Wabi, the AI startup that let anyone build an app with a text prompt, is making a modest pivot in response to the rise of AI agents like Meta’s Muse and Instinct. The company announced this week that Wabi is transforming into a kind of AI messaging assistant, while still holding onto its original ability to help users build apps and get things done. Founder Eugenia Kuyda (who previously founded the AI companion startup Replika) describes this new version, Wabi 2.0, as “a personal agent that does things for you, and builds the interface you need in the moment.” The value proposition is similar to the original Wabi, but the way users interact with it is different — Wabi 2.0 is no longer positioning itself against other “vibe-coding” tools, but going head-to-head with AI agents that fold conversation and productivity tasks into a single interface. The pivot also echoes what companies like OpenAI are thinking about — how apps get embedded into their general-purpose chatbot interfaces. OpenAI announced updates to ChatGPT’s app/plugin features on the same day at Dev Day, letting apps deliver interactive experiences inside ChatGPT, with app names now showing more prominently in the sidebar. Kuyda explained on X that pure chat-based agents get worse the longer you use them — your history and in-progress tasks get buried in an endless scroll. People don’t just want to type; they want to tap, scroll, and browse. She gave examples of users building a calorie tracker, a workout log, or a weight tracker inside a health-themed thread, or a kids’ schedule calendar, a language-learning app, and local family activity info inside a family-themed thread. Wabi 2.0 is currently invite-only, with invite codes being handed out via X.

💬 JudyAI Lab Take

Wabi, the AI tool best known for “build an app with a text prompt,” announced this week that it’s transforming into Wabi 2.0 — a personal agent that generates its own working interface as it gets things done. This shift is worth watching for anyone building AI products, because it points to a structural pain point emerging in pure chat interfaces.

Founder Eugenia Kuyda nails the core issue: pure chat-based agents get worse the longer you use them, with history and in-progress tasks buried in an endless scroll. Users don’t just want to type — they want to tap, browse, and interact with an actual interface. That echoes OpenAI’s own update to ChatGPT’s app features on the same day at Dev Day, letting apps deliver interactive experiences inside the chat interface and show up more prominently in the sidebar. Put together, these two moves point to a trend: the next front of competition for AI agents may not be conversational ability itself, but whether they can generate a usable interface beyond the conversation.

If you’re designing an AI product, ask yourself: on someone’s tenth conversation, are they still retyping the same thing all over again?

📅 Source Info

🔗 Further Reading