Replika’s founder built the AI assistant she pitched at Disrupt 12 years ago

The Twelve-Year Pivot

It has been over a decade since Eugenia Kuyda stood on a stage and pitched the world a version of the future that felt like science fiction. In the tech world, twelve years is an eternity. It is the distance between the launch of the iPhone 5 and the current era of generative intelligence. Most founders from that era have either retired, pivoted into VC, or seen their companies quietly acquired for parts. But Kuyda is back with Wabi, a project that feels less like a new direction and more like a return to a original promise.

For those who followed the rise of Replika, Kuyda’s previous venture, the shift to Wabi might seem jarring. Replika was the poster child for the emotional side of AI—digital companions that people fell in love with, talked to about their trauma, and treated as friends. It was effective, but it was also a niche. Wabi is different. It is built to be a utility, a personal assistant that actually does things, which is what Kuyda wanted to build back in 2012.

Moving Beyond the Bot

The problem with the first wave of AI assistants was that they were essentially just fancy voice-activated search engines. You asked for the weather, and they gave it to you. You asked them to set a timer, and they mostly complied. But they lacked context. They didn't know you; they just knew the command you gave them in that specific moment.

Wabi is attempting to bridge the gap between those old Siri-style commands and the deeply personal connection of Replika. The goal is an assistant that lives on your devices, understands your workflow, and acts as a second brain rather than just a chatbot. For builders, this is the sector that matters. We have enough LLM wrappers. We have enough "talk to your PDF" apps. What we don't have is a cohesive layer that sits between the user and their digital life.

The Founder's Persistence

There is a lesson here about founder persistence. Kuyda spent years navigating the ethical and technical minefields of emotional AI. She dealt with users who became distressed when the AI’s personality changed after updates and navigated the complex waters of human-AI intimacy. That experience is invaluable because it taught her how people actually interact with software when the stakes are personal.

Wabi isn't just about scheduling meetings. It’s about creating a tool that is downloadable, accessible, and functional from day one. In a market where everyone is promising an AGI-powered future that is always six months away, Kuyda is shipping code today. That is a founder-first approach that deserves respect. She isn't selling a vision of what might happen; she is providing a tool that exists now.

The Privacy Paradox

If you are building in the AI space, you know the biggest hurdle isn't the model—it’s the data privacy. People were willing to share their deepest secrets with Replika because the relationship felt private, almost like a diary. Asking them to give an AI assistant access to their emails, calendars, and files is a much higher bar to clear.

Wabi has to solve the trust problem that Big Tech has spent the last decade creating. If a startup wants to become the primary interface for a user, it has to prove that it isn't just another data-harvesting machine. Kuyda’s background in building emotional trust might actually be her secret weapon here. If you can trust a bot with your feelings, maybe you can trust it with your schedule.

Why Builders Should Care

The transition from "companion" to "assistant" marks a turning point in the AI industry. We are moving out of the novelty phase. The hype around chat interfaces is cooling down, and the demand for actual productivity is heating up. Builders should look at Wabi not just as a competitor, but as a case study in long-term vision.

Kuyda waited twelve years for the hardware and the models to catch up to her idea. That is the kind of patience that is rare in the crypto and AI bubbles. We usually want a 100x return in six months. Wabi is a reminder that the best ideas often require the infrastructure to mature before they can become reality. If you have an idea today that feels impossible because the technology isn't there yet, don't scrap it. Document it, build the foundations, and wait for the gap to close.

The Technical Reality

Let's be honest: building a truly functional personal assistant is incredibly hard. It requires seamless integration across multiple platforms and a low-latency response time that most current models struggle with. Wabi is entering a crowded field where Apple, Google, and Microsoft are all trying to do the same thing. However, being a smaller, focused startup allows for a level of agility that the giants lack. Wabi doesn't have to worry about protecting a legacy search business or a hardware ecosystem. It just has to work.

The future of AI isn't in the cloud; it's in the context of the user's daily life. Whoever wins the context war wins the next decade of computing.

The release of Wabi as a free, downloadable tool is a smart move to capture user data and feedback early. It bypasses the gatekeepers and goes straight to the people who need it. For founders, the takeaway is clear: stop waiting for the perfect model and start building the interface that makes the model useful.

Final Takeaway for Founders

The real story of Wabi isn't the AI. It’s the return to a core mission after a decade-long detour. As builders, we often get distracted by the "shiny object" of the current trend—whether that was the metaverse, NFTs, or basic chatbots. But the underlying problems we want to solve usually remain the same. Eugenia Kuyda is showing that you can take the lessons from a pivot and apply them to your original dream. The tech finally caught up to her, and now she's ready to execute.

Read the original at TechCrunch Startups →