Ollie is banking on privacy to gain an edge in the AI assistant competition.

For an AI assistant to be genuinely beneficial, it must first possess extensive knowledge about you. Ollie, a personal assistant designed for everyday tasks, asserts that this doesn’t require sacrificing your privacy or sharing all your data.

While several enterprise-oriented AI assistants implement data privacy measures, Ollie stands out as one of the first mainstream, family-focused AIs to attain SOC 2 compliance. This framework allows companies to prove, through independent audits, that they have formal controls to safeguard customer data and secure their operations.

This accomplishment marks a crucial milestone for Ollie’s team, signaling to prospective users that the assistant is not out to collect personal data for ulterior motives, such as AI training.

“We fundamentally believe that trust and privacy are vital, which is why our business model is subscription-based; we want our users to understand that Ollie is here for you,” stated Ollie co-founder and CEO Bill Lennon in a conversation with TechCrunch. “We’re not sharing your data with anyone,” he emphasized.

Numerous personal AI assistants are emerging as user-friendly options designed to integrate seamlessly into daily routines through text messaging, making Ollie’s privacy-focused approach a noteworthy differentiator.

Currently, Ollie faces competition from other text-based assistants such as Poke (now part of Cognition), Fambot, Ohai, Folk, Saner.ai, and Tomo. Additionally, there are many assistants centered around work-related tasks, calendar management, and email, including Town, Lindy, and Reclaim.ai. Not to mention Instinct, an AI assistant that secured $350 million in funding at a $2.5 billion valuation before its launch.

To say that competition in this space is intensifying is an understatement.

In contrast, San Diego-based Ollie has managed to secure a much smaller seed funding round of $7.5 million, primarily from Khosla Ventures and AI House.

Despite the advantages these assistants offer, users are grappling with the extent of privacy and personal data they are willing to forfeit. For instance, Instinct drew criticism for having overly broad Terms of Service and a privacy policy that allowed it a “perpetual and irrevocable” license to “access, use, host, cache, store, reproduce, transmit, display, publish, distribute, and modify” any of the user’s materials, including for AI model training. Yikes.

Lennon adopts a different strategy with Ollie. Currently, this family-centric assistant links to calendars and emails for organizing schedules and updates. It also provides features for meal planning, grocery shopping, tracking tasks, booking appointments, and paying bills through group chats. Future enhancements may include managing household budgets.

However, all this access hinges on trust, which Ollie positions as its primary advantage over competitors. “We’re not sharing your data with anyone. This is extremely sensitive, and it’s essential for gaining user trust,” Lennon noted.

In addition to the protections offered by SOC 2 compliance, Ollie does not ask for usernames or passwords to perform its tasks. When it needs to log in to a website on behalf of the user, Ollie utilizes its own cloud-based browser and sends the user a link to a remote session in that browser. This same method applies when making payments or purchases for users.

While this approach ensures security, it does mean that users must log in each time to complete certain tasks. Nevertheless, Lennon is optimistic that technology will advance in this area.

“This is… uncharted territory. I believe that in the future, we will implement some form of secure hard tokenization, enabling you to avoid re-entering your information every time,” he remarked. “That’s pioneering stuff… we aim to develop the optimal user experience that balances convenience and trust.”

Establishing that trust is crucial before users will allow Ollie to access more sensitive information, such as their bank accounts, even if it’s through a connector like Plaid.

Lennon, who holds a PhD in AI, comes from a fintech background, having previously sold his neobank startup, Groundwork, targeting nonprofits, in 2021. This experience could be beneficial as Ollie expands into money management.

However, it remains uncertain whether consumers truly desire such assistance.

Lennon remains optimistic, noting that Ollie’s user retention rates align with those of leading AI subscription services regarding paid subscribers. (He refrained from disclosing specific user or customer numbers, citing the high competition in this space.)

These systems can occasionally disappoint. Despite the buzz surrounding Instinct, one of my initial interactions involved trying to price out hotels and book one, and it provided all incorrect rates. Meanwhile, Ollie faced an infrastructure issue with its text provider during my evaluation, leading to a non-responsive assistant.

We inquired how the founder is addressing such challenges, given that most users only grant a new app or technology one chance; if it fails, they simply move on.

Lennon acknowledged that this is a prevalent issue with AI overall. “This represents a challenge with LLMs in general — since they are stochastic [i.e., involve probability], they are inherently unreliable,” he explained. “We must essentially create the harness — the agent harness — in a defensive way to capture and prevent these issues… It’s a matter of resolving a thousand minor inconveniences first.”