Zach Yadegari, the 19-year-old co-founder of the calorie tracking app Cal AI, has moved quickly from his first big exit to his next ambitious bet. His new company, Persona, is an AI agent startup that has already raised $10 million in a funding round led by Vine Ventures. Z Fellows founder Cory Levy and Collective Global also participated in the round.
The announcement comes only months after MyFitnessPal acquired Cal AI in March for an undisclosed sum. Cal AI had surged past the older calorie app in app store rankings and generated more than $30 million in annual revenue in under two years. The deal was significant enough to convince Yadegari and his co-founders to sell the business, according to statements made at the time.
Yadegari and his co-founders initially stayed with MyFitnessPal after the acquisition, continuing to work on Cal AI. But Yadegari left the larger company in June and immediately began developing his next idea. He is working with another co-founder who has not yet been publicly announced.
What Persona Is Building
Persona is a personal AI assistant that combines software with a physical wearable. The company plans to sell a $179 band that users can wear on their wrist. The assistant is similar in concept to other personal AI products such as Instinct or Meta's Muse, but Persona's hardware component is central to its pitch.
The band is not designed to be an always-on ambient recorder. Yadegari says it will not listen or record continuously. Instead, the user activates the assistant with a button, or the band can be configured to respond to a flick of the wrist. That design choice is meant to address one of the biggest concerns about wearable AI: the fear that a device is constantly capturing conversations and environmental audio.
Persona's privacy architecture also differs from some competitors. The AI model does not run on the device, and user data is not stored locally on the band. Instead, the model runs in Persona's cloud, and user data is sent there for processing. The company says it encrypts data in transit and at rest. Users can delete their data, and Persona promises that conversations will remain private and will not be sold to advertisers or data brokers.
A Different Approach to Privacy
The company is also building safeguards against phishing and prompt-injection attacks. Those two threats have become major concerns for AI agents, which can be manipulated into revealing information or taking unintended actions. Prompt injection, in particular, occurs when malicious instructions are hidden in content that an AI model processes, potentially causing the agent to ignore its original rules or leak data. AI labs have identified these risks as serious obstacles for autonomous systems.
Persona is also putting limits on what its agent can see and do with financial information. Yadegari says the assistant cannot see a user's credit card details. Purchases require the user's approval, and payment is handled through external providers such as Stripe Link. Stripe Link is a secure wallet designed for use by agents, which allows an AI assistant to complete transactions without directly exposing full card credentials.
That approach reflects a broader trend in agentic commerce. As AI assistants become capable of ordering food, booking rides, and buying products, companies are trying to create payment rails that balance convenience with security. The challenge is to let an agent act on a user's behalf without giving it unrestricted access to sensitive financial data.
How Persona Plans to Make Money
Although Persona says user data will not be sold to advertisers, the startup's current business model does rely on advertising. Yadegari says the company will show advertised products during shopping research. However, he claims the agent will present unbiased recommendations because it will not know which items are sponsored. For example, if a user is researching office chairs, the agent will know the user's preferences and show chairs that match those preferences. Some of those results may be sponsored, but the agent itself will not distinguish them when making recommendations.
That model creates an obvious tension. An AI assistant that is both a trusted advisor and an advertising channel must convince users that its recommendations are not shaped by commercial incentives. Persona's answer is to separate the recommendation logic from the ad targeting, at least in theory. Whether that separation holds up in practice will be a key test for the company.
The ad-supported approach also contrasts with subscription models used by many productivity and AI tools. A subscription business can align incentives with users by charging directly for the service. An ad business, by contrast, may create pressure to increase engagement or promote sponsored products. Persona is betting that it can generate revenue from commerce research without compromising the assistant's perceived objectivity.
Beta, Preorders, and Launch Plans
Persona is not waiting for its hardware to launch before testing the product. The company has introduced a free beta version that works through iMessage text messages. Users can interact with the assistant by sending texts, and the company says it has attracted a few thousand beta users. The band has also generated five figures in preorder revenue.
The wearable bands are expected to be available in December. An app is also in development. Once the app and band are both available, Persona will compete more directly with other consumer AI assistants and wearable devices.
Like other consumer AI assistants, Persona can order food or rides, take notes, book travel, and answer questions and emails. Those capabilities place it in a crowded field of AI agents that aim to simplify everyday tasks. The difference, according to Yadegari, is the combination of a wearable form factor, a privacy-conscious activation model, and a cloud-based assistant that can act across multiple services.
Competition in Wearable AI
Persona enters a market that has already seen significant experimentation. Amazon acquired the wearable AI device Bee in 2025, signaling interest from major technology companies in always-on or easily accessible AI companions. Meta has developed its Muse assistant, and other startups such as Instinct are pursuing similar ideas. The category has also seen cautionary tales, including early wearable AI devices that struggled with battery life, usefulness, and privacy concerns.
The failure or success of these products often depends on whether they solve a real problem better than a smartphone. A wearable AI assistant must justify why users should wear an extra device, charge it, and interact with it instead of simply pulling out their phone. Persona's answer is that it will eventually reduce the need to stare at a phone screen by handling tasks proactively and seamlessly.
Yadegari has described a future in which the many utility apps people use every day collapse into a single super app powered by an AI assistant. In his view, shopping apps, ride-hailing services, delivery platforms, and email clients will all be accessible through one intelligent interface. That vision is shared by several major technology companies and startups, but it remains far from fully realized. Building reliable agents that can navigate multiple apps, respect user preferences, and handle payments securely is a difficult engineering and trust challenge.
From Cal AI to Persona
Yadegari's track record gives Persona instant attention. Cal AI became a breakout success in the calorie tracking market, surpassing a long-established competitor in app store rankings and generating more than $30 million in annual revenue in less than two years. The acquisition by MyFitnessPal validated the company's growth and gave Yadegari and his team a significant exit at a young age.
Before Cal AI, Yadegari built his first business in the ninth grade. He sold it for $100,000 when he was 16. That early experience, followed by Cal AI's rapid rise, has made him one of the most closely watched young founders in consumer AI. His new venture has attracted backing from Vine Ventures, Cory Levy of Z Fellows, and Collective Global, suggesting that investors believe he can repeat his previous success in a much larger market.
The funding will likely help Persona scale its team, develop its hardware, and expand the beta product. The company is hiring and preparing for a December launch, but it faces competition from well-funded technology giants and a growing number of startups. It also must deliver on its privacy promises while operating an ad-supported business.
What to Watch
The most immediate question is whether Persona's band can stand out in a crowded wearable market. Its $179 price point is lower than some AI hardware devices but still requires users to adopt a new habit. The activation button and wrist-flick control are meant to make the device feel intentional rather than invasive, which could appeal to users who are uncomfortable with always-listening gadgets.
The second question is whether the assistant can perform tasks reliably enough to justify its existence. Ordering food, booking rides, and managing emails are useful capabilities, but they are also available through smartphones and existing apps. Persona needs to show that its agent can complete those tasks faster, more intelligently, or more conveniently than current alternatives.
The third question is trust. Persona's ad-supported model, cloud-based data processing, and payment integrations all depend on users believing that the assistant is acting in their interest. The company's encryption, deletion options, prompt-injection safeguards, and Stripe Link integration are designed to build that trust. But the real test will come when users begin relying on Persona for sensitive tasks and high-stakes purchases.
Persona has already raised $10 million, attracted a few thousand beta users, and generated five figures in preorder revenue. Its bands are expected to ship in December, with an app in development. Whether it becomes a defining consumer AI product or another experiment in wearable computing will depend on execution, adoption, and the company's ability to balance advertising with user trust.
Source:TechCrunch News