One AI assistant recently came under fire for burying a clause in its terms of service that granted itself a “perpetual and irrevocable” license to access, store, reproduce, and modify user data, including for training its own models. That assistant had just raised $350 million. So when a competitor shows up claiming privacy is its whole pitch, it’s worth asking whether that’s a real commitment or just a marketing angle.
Ollie is a family-focused AI assistant built around the idea that you shouldn’t have to hand over your personal data just to get help organizing your life. According to TechCrunch, Ollie is among the first consumer-facing, family-oriented AI assistants to achieve SOC 2 compliance, an independent audit framework that verifies a company has formal controls in place to protect customer data. That’s a meaningful credential in a space where most competitors are either vague about data handling or actively building business models that depend on it.
The assistant connects to your calendar and email to track family schedules and daily updates. It also helps with meal planning, grocery shopping, to-do lists, appointment booking, and paying bills through group chats. Budget management is on the roadmap too. That’s a significant amount of access to personal information, which makes the privacy question more than academic.
Ollie’s business model is a subscription, not advertising or data licensing. Co-founder and CEO Bill Lennon, who has a Ph.D in AI and previously sold Groundwork, a neobank for nonprofits, in 2021, is direct about why. “We’re not sharing your data with anyone,” he told TechCrunch. “Ollie works for you.” The subscription model is meant to remove the incentive to monetize user data in the background. That logic holds, as long as the company stays independent and the terms of service don’t quietly change.
One technically interesting decision: Ollie does not ask for usernames or passwords. When it needs to log into a site on your behalf, it uses its own cloud browser and sends you a link to a remote session. Payments work the same way. The tradeoff is friction. You have to log in manually each time to complete certain tasks. Lennon acknowledges this isn’t ideal, and says the company is working toward some form of hard tokenization that would remove the re-entry burden without compromising security. That’s a real engineering challenge, and it’s honest to frame it as unsolved rather than pretending it doesn’t exist.
Ollie is based in San Diego and has raised $7.5 million in seed funding, mostly from Khosla Ventures and AI House. That puts it in a very different position from competitors like Instinct, which raised $350 million at a $2.5 billion valuation before it even launched. The field also includes text-based assistants like Fambot, Ohai, Folk, Saner.ai, and Tomo, plus productivity-focused tools like Lindy, Town, and Reclaim.ai. It’s a crowded category, and most players are racing to scale fast, which tends to create exactly the kind of shortcuts that erode user privacy.
Whether Ollie can hold its position is an open question. Lennon says retention curves are consistent with leading AI subscriptions, but declined to share actual user numbers, citing competition. And the product has real reliability issues. TechCrunch reported an infrastructure outage with Ollie’s text provider during testing, leaving the assistant unresponsive. Lennon is candid about this too, pointing to the probabilistic nature of large language models as an inherent reliability problem the whole industry is working around. “It’s almost like there’s just 1,000 cuts that you’ve got to solve first,” he said.
The broader context matters here. Consumers are increasingly aware that AI assistants require deep personal access to be useful, and trust is becoming a real differentiator. SOC 2 compliance and a no-password policy are concrete steps, not just talking points. But they only mean something if the company’s terms of service, data retention policies, and third-party agreements hold up under scrutiny. For now, Ollie’s approach is worth watching, especially as the rest of the market continues to treat user data as a resource rather than a liability.