Ollie bets on privacy to win the AI assistant race

Ollie, a personal AI assistant focused on organizing family life, has become one of the first mainstream, family-focused AI assistants to achieve SOC 2 compliance — an independent audit confirming it has formal controls to protect customer data and operate its systems securely. The San Diego-based startup is betting that privacy, not more aggressive data collection, will differentiate it in an increasingly crowded field.

Ollie co-founder and CEO Bill Lennon told TechCrunch that trust and privacy are central to the company’s approach, and that its subscription-based business model — rather than an ad- or data-driven one — is meant to reassure users that “Ollie works for you.” “We’re not sharing your data with anyone,” he said.

Ollie connects to calendars and email to organize family schedules, and offers tools to plan meals, shop for groceries, track to-dos, book appointments, and pay bills through group chats; it may later help manage household budgets. Rather than asking for usernames and passwords, Ollie logs into websites on a user’s behalf using its own cloud browser and sends a link to a remote session — a more secure approach, though one that still requires users to log in to complete certain tasks. Lennon said the company expects to eventually use some form of secure tokenization to reduce that friction.

Ollie competes with a growing field of text-based personal assistants, including Poke (acquired by Cognition), Fambot, Ohai, Folk, Saner.ai, and Tomo, as well as work-focused tools like Town, Lindy, and Reclaim.ai. Rival Instinct raised $350 million at a $2.5 billion valuation before it even launched, but drew criticism for terms of service that granted it a “perpetual and irrevocable” license to use, host, and modify user materials, including for training its AI models. Ollie, by contrast, has raised a comparatively modest $7.5 million seed round, largely from Khosla Ventures and AI House.

Lennon, who holds a PhD in AI and previously sold his fintech startup Groundwork in 2021, said Ollie’s retention curves are consistent with leading AI subscriptions among paid users, though he declined to share subscriber numbers, citing competition. He acknowledged the reliability challenges facing AI agents generally, noting that large language models are inherently unreliable because they are probabilistic, which is why Ollie has built its agent harness defensively to catch and prevent failures — a problem he compared to solving a thousand small cuts one at a time.

Why it matters

As AI assistants push deeper into sensitive personal data — calendars, email, payments — SOC 2 compliance and a subscription model could become a competitive edge against rivals whose broader data-usage terms have drawn scrutiny.