For an AI assistant to be truly useful, it first needs to know a lot about its users. Olly, your personal assistant in everyday life, is betting that doesn’t mean you don’t have to hand over all your data and sacrifice your privacy in the process.
While some enterprise AI assistants have introduced data privacy protections, Ollie is one of the first mainstream family AIs to achieve SOC 2 compliance. SOC 2 is a framework that allows companies to demonstrate through independent audits that they have formal controls in place to protect customer data and operate their systems securely.
Achieving this was an important milestone for Ollie’s team because it shows potential users that the assistant is not trying to collect personal data for other purposes, such as AI training.
“Fundamentally, we believe that trust and privacy are absolutely essential. That’s why our business model is subscription, because we want you to know that Ollie is working for you,” Bill Lennon, co-founder and CEO of Ollie, explained in an interview with TechCrunch. “We will not share your data with anyone,” he stressed.
There are a number of personal AI assistants that are positioning themselves as consumer solutions focused on integrating into users’ daily lives via text messages. That’s why Ollie’s privacy perspective can be a key differentiator.
Currently, Ollie competes with existing text-based assistants such as Poke (acquired by Cognition), Fambot, Ohai, Folk, Saner.ai, and Tomo. There are also a number of assistants that focus on work, calendar management, and email, including Town, Lindy, and Reclaim.ai. And, of course, there’s Instinct, the AI assistant that raised $350 million in pre-launch funding at a $2.5 billion valuation.
In other words, to say the space is heating up is an understatement.
By comparison, San Diego-based Ollie raised a much more modest $7.5 million seed round, primarily from Khosla Ventures along with AI House.
Despite the usefulness these assistants offer, consumers who utilize these tools struggle with how much privacy and personal data they are willing to give up in the process. For example, Instinct came under fire for its overbroad terms of service and privacy policy, which grant a “perpetual and irrevocable” license to “access, use, host, cache, store, reproduce, transmit, display, publish, distribute, and modify” your materials, including to train AI models. oh yeah.
Lennon has a different approach to Ollie. Now, the family assistant connects to calendars and email to organize family schedules and provide users with daily updates. It also provides tools to help you plan meals, shop for groceries, track to-dos, make reservations, and pay bills via group chat. It may also be useful for managing your household finances later on.
But all this access requires trust, which Olly argues is an added value compared to its competitors. “We don’t share your data with anyone. It’s highly sensitive and that’s necessary to earn the trust of our users,” Lennon said.
In addition to protecting your data with SOC 2 compliance, Ollie does not require a username or password to complete tasks. When Ollie needs to log in to a website on behalf of a user, he logs in using his own browser in the cloud and sends the user a link to the remote session within the browser. The process is similar if your assistant needs to make a payment or purchase on your behalf.
Although this method provides security, the drawback is that users must always log in to complete certain tasks. Still, John Lennon believes that technology will eventually improve.
“This is… new territory. I think in the future we’ll do some kind of hard tokenization in a secure way so we don’t have to re-enter (information) every time,” he said. “It’s cutting-edge stuff…We want to find the right user experience that balances convenience and reliability.”
It’s important to build trust before users trust Ollie to access sensitive information such as bank accounts. Even if it’s through a connector like Plaid.
Lennon, who has a PhD in AI, also has a background in fintech, having previously sold his startup Groundwork, a neobank for nonprofits, in 2021. That experience could help Mr. Oly move further into money management.
Still, it’s not yet clear whether people want this kind of support.
But Lennon is hopeful, saying Olly’s retention curve is in line with major AI subscribers in terms of paid subscribers. (He declined to say how many users or paying customers Ollie currently has, saying there is too much competition in the space.)
These systems can also be hit or miss at times. Despite Instinct’s hype, one of the first things I tested with Instinct was to look up hotel prices and book, and the prices I saw were all wrong. Meanwhile, Ollie experienced an infrastructure failure with his text provider and stopped responding while I was testing it.
Given that most consumers only try new apps and technology once, Ollie asked the founders how they address such issues. If it breaks, they simply move on.
Lennon agreed that this is an issue with AI overall. “This is the challenge with LLM in general: LLM is probabilistic (i.e. involves probabilities) and is therefore inherently unreliable,” he said. “We basically have to build a harness, an agent harness, in a defensive way to catch and prevent these things. It’s like we only have 1,000 cuts to solve first.”
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
