The race to build the next generation of personal AI assistants is heating up, with major players and ambitious startups competing to create agents that can do more than simply answer questions. These systems are increasingly being designed to make calls, manage appointments, handle purchases, communicate with businesses, and complete everyday tasks on behalf of users.
Among the growing field of competitors, venture capitalist Vinod Khosla believes one relatively new company could have a major advantage: trust.
Khosla is backing Wajo, a startup founded by former Google and DeepMind engineer Shivani Poddar. His argument is straightforward: if people are going to give an AI agent access to their personal information and allow it to act independently, they need to feel confident that the system is working in their best interests.
Trust May Become the Biggest Differentiator
According to Khosla, Wajo has built its agent around privacy, safety, and user trust rather than treating those concerns as secondary features.
That could become particularly important as AI assistants gain more control over users’ digital lives. An agent that can read messages, contact companies, make purchases, or organize personal tasks has considerably more responsibility than a traditional chatbot.
Khosla also questioned whether consumers will be comfortable handing this level of access to advertising-driven technology companies. His concern is that users may hesitate to give sensitive personal data to businesses whose primary model depends on collecting information for advertising.
For Wajo, positioning itself as a trusted intermediary could therefore become a significant competitive advantage.
Wajo’s Agent Goes Beyond Chat
Wajo’s personal agent, called Fo, is designed to operate across several everyday communication and service workflows.
Users can interact with Fo through platforms such as iMessage and WhatsApp, while a web interface organizes potential tasks into categories including scheduling, gifting, and general life administration.
The agent can also communicate with businesses and individuals on the user’s behalf. That means Fo can potentially make phone calls, arrange services, cancel appointments, or handle other errands without requiring the user to manage every step manually.
Early testing has shown that the technology can already handle some real-world tasks successfully. For example, Fo was able to contact an airline to investigate an upgrade request.
However, the experience isn’t perfect yet. Some calls involving personal contacts have raised questions about how clearly an AI should identify itself when speaking for a user. As more AI agents gain phone-calling capabilities, disclosure and transparency could become an increasingly important industry issue.
When AI Needs a Human, Fo Can Adapt
One interesting feature of Wajo’s approach is that the system doesn’t necessarily have to complete every task entirely through automation.
When necessary, Fo can bring a human into the process to finish an assignment. This creates a hybrid model where AI handles the workflow but human assistance can step in when automation reaches its limits.
Fo can also generate a virtual credit card for purchases, allowing transactions to happen without directly exposing the user’s actual payment information.
The agent has its own email identity as well, enabling it to communicate with businesses and other people. If a website refuses to interact with an AI agent, Fo can ask the user to step in or create a shared conversation involving the user and the business to verify the request.
Wajo Is Growing Quickly
Wajo has attracted support from prominent technology figures, including Google’s Jeff Dean and investor Gokul Rajaram.
The startup has not disclosed how much funding it has raised or provided specific user numbers. However, Wajo says Fo is already being used across 107 countries and that its usage has increased tenfold since launching in late September.
The personal-agent market is still in its early stages, and competitors such as Meta’s Muse, OpenAI’s Dot, and Instinct are moving quickly.
But as these agents gain the ability to act independently, the winning product may not simply be the one capable of doing the most.
