Ollie bets on privacy as its edge in the family AI assistant race

A personal AI assistant can't be useful unless it knows everything about you — calendar, emails, shopping lists, bills. Ollie, which targets families rather than enterprises, argues that knowledge shouldn't require surrendering privacy. The San Diego company recently completed SOC 2 compliance, an independent audit framework proving formal controls exist to protect customer data and secure systems. It claims to be the first in the consumer family category to do so; enterprise-focused competitors already hold similar certifications.

Privacy as business model, not slogan

Founder and CEO Bill Lennon says the subscription model follows directly from that stance: the user is the customer, not the product. "We don't share your data with anyone," he emphasizes, adding that the goal is to build trust around especially sensitive data. SOC 2 compliance is meant to signal that the company doesn't collect personal information for model training or other commercial purposes — a promise many players in the space avoid making explicitly.

Crowded market, very deep pockets

The category is heating up. On the consumer side: Poke (acquired by Cognition), Fambot, Ohai, Folk, Saner.ai, and Tomo. In the work-and-scheduling lane: Town, Lindy, and Reclaim.ai. Looming over all of them is Instinct, which raised $350 million at a $2.5 billion valuation before launch. By comparison, Ollie closed a modest $7.5 million seed round led by Khosla Ventures and AI House. The capital gap doesn't necessarily reflect a capability gap, but it dictates a very different pace of development and distribution.

When the privacy policy becomes the bad headline

Instinct drew sharp criticism for terms granting it a "perpetual, irrevocable" license to access, use, store, reproduce, transmit, display, publish, distribute, and modify all user content — including for model training. That is exactly the language Ollie tries to distance itself from. The functionality is similar: calendar and email integration for family scheduling, meal planning, shopping, tasks, appointments, and payments via group chats, with future plans for household budget management. The difference lies in data-access architecture.

Cloud browser instead of passwords

When Ollie needs to log into a site on the user's behalf, it doesn't ask for credentials. Instead it spins up a cloud browser, performs the action, and sends the user a link to a remote session where they complete the login or payment themselves. The approach prevents the company from holding keys to external accounts, but it creates friction: the user must intervene manually for any action requiring authentication. Lennon calls it "new territory" and expects hard tokenization to eventually enable full automation without compromising the safety model. For now it's still "things at the technology frontier," in his words, and the challenge is finding the user experience that balances convenience and security.