San Diego-based AI startup Ollie is positioning itself as a privacy-focused alternative in the crowded personal AI assistant market, announcing in September 2026 that it has achieved SOC 2 compliance — an independent audit certification confirming formal controls are in place to protect customer data.
The family-oriented assistant connects to users’ calendars and email to organize schedules, plan meals, shop for groceries, track to-dos, book appointments, and pay bills through group chats. Future features may include household budget management. Despite requiring access to sensitive personal information to perform these tasks, Ollie says it does not use customer data to train AI models or share it with third parties.
“We fundamentally think that trust and privacy are absolutely imperative, and that’s why our business model is a subscription, because we want our users to know that Ollie works for you,” said co-founder and CEO Bill Lennon. “We’re not sharing your data with anyone.”
Ollie has raised a $7.5 million seed round, largely from Khosla Ventures along with AI House. The company operates in a competitive space that includes text-based assistants such as Poke, Fambot, Ohai, Folk, Saner.ai, and Tomo, as well as work-focused tools like Lindy and Reclaim.ai. Instinct, a rival assistant, recently raised $350 million at a $2.5 billion valuation before launching, though it drew criticism for a privacy policy that granted it broad, irrevocable rights to user data.
To avoid storing user credentials, Ollie handles logins and payments through a cloud-based browser session, sending users a link to complete sensitive actions themselves. Lennon acknowledged this adds friction but said the company plans to explore secure tokenization to improve the experience over time.
Lennon, who holds a PhD in AI and previously sold fintech startup Groundwork in 2021, declined to disclose Ollie’s current user or subscriber numbers, citing the competitive environment. He did say retention curves are consistent with leading AI subscription products. Reliability remains a challenge — Ollie experienced an infrastructure outage during testing — which Lennon attributed to the inherent unpredictability of large language models.
Source: TechCrunch