Ollie is betting its deal with privateness may help it win the AI assistant race

For an AI assistant to grow to be really helpful, it first has to know so much about you. Ollie, a private assistant for on a regular basis life, is betting that doesn’t imply you need to hand over all of your knowledge and sacrifice your privateness within the course of.

Whereas some enterprise-focused AI assistants have knowledge privateness protections in place, Ollie is among the many first mainstream, family-focused AI to attain SOC 2 compliance — a framework that enables an organization to display, by way of an impartial audit, that it has formal controls in place to guard buyer knowledge and function its methods securely.

Attaining this was an necessary milestone for Ollie’s crew, because it alerts to potential customers that the assistant isn’t attempting to scoop up your private knowledge for different functions, like AI coaching.

“We essentially suppose that belief and privateness are completely crucial, and that’s why our enterprise mannequin is a subscription, as a result of we wish our customers to know that Ollie works for you,” defined Ollie co-founder and CEO Bill Lennon in an interview with TechCrunch. “We’re not sharing your knowledge with anybody,” he pressured.

There are various private AI assistants positioning themselves as consumer-friendly options targeted on integrating with customers’ every day lives through textual content messages, which is why Ollie’s privateness angle might show a major differentiator.

At present, Ollie competes with different current text-based assistants like Poke (acquired by Cognition), Fambot, Ohai, Folk, Saner.ai, and Tomo. There are additionally a lot of assistants targeted extra on work, calendar administration and e-mail, like Town, Lindy and Reclaim.ai. And naturally, there’s Instinct, the AI assistant that just landed $350 million in funding at a $2.5 billion valuation earlier than it even launched.

To say the area is heating up is placing it mildly, in different phrases.

By comparability, San Diego-based Ollie has raised a way more modest $7.5 million seed spherical, largely from Khosla Ventures, together with AI House.

Regardless of the utility these assistants present, shoppers who flip to those instruments are wrestling with how a lot privateness and private knowledge they need to surrender within the course of. Intuition, for instance, came under fire for its overly broad Terms of Service and privacy policy that granted it a “perpetual and irrevocable” license to “entry, use, host, cache, retailer, reproduce, transmit, show, publish, distribute, and modify” any of the person’s supplies, together with for coaching its AI fashions. Yikes.

Lennon is taking a distinct method with Ollie. At present, this household assistant connects to calendars and e-mail to prepare household schedules and preserve its customers up to date about their days. It additionally presents instruments that may enable you plan meals, store for groceries, monitor to-dos, guide appointments, and pay payments through group chats. Later, it could assist handle family budgets, too.

But, all this entry requires belief, which Ollie claims, is its value-add in comparison with its rivals. “We’re not sharing your knowledge with anybody. That is tremendous delicate, and that’s essential to win the belief of the customers,” Lennon mentioned.

Along with the information protections offered by SOC 2 compliance, Ollie doesn’t request usernames or passwords to finish its duties. When Ollie must log in to an internet site on the person’s behalf, it does so utilizing its personal browser within the cloud, and sends the person a hyperlink to a distant session within the browser. The method is comparable when the assistant must make a cost or buy on the person’s behalf.

Whereas this methodology presents safety, the draw back is that customers at all times should log in to finish sure duties. Nonetheless, that’s one thing Lennon believes expertise will enhance upon in time.

“That is… new territory. I feel sooner or later, we are going to do some type of laborious tokenization, in a safe means, so that you’re not going to should re-enter [your information] each time,” he mentioned. “That’s frontier stuff… we need to discover the proper person expertise that balances comfort and belief.”

Constructing that belief will likely be necessary earlier than customers belief Ollie to entry extra delicate supplies, like their financial institution accounts, even when that’s by way of a connector like Plaid.

Lennon, who has a PhD in AI, additionally has a background in fintech, having beforehand offered his startup Groundwork, a neobank for nonprofits, in 2021. That have might assist as Ollie strikes additional into cash administration.

Nonetheless, it’s not but clear if individuals even need any of this help.

Lennon is hopeful, although, saying that Ollie’s retention curves are according to the main AI subscriptions by way of paid subscribers. (He didn’t share what number of customers or paid prospects are actually utilizing Ollie, saying the area is just too aggressive.)

These methods may also be hit-or-miss at occasions. Regardless of the hype round Intuition, certainly one of my first exams with it was to cost out resorts then guide one, and it quoted me all of the fallacious charges. Ollie, in the meantime, skilled an infrastructure outage with its textual content supplier after I was testing it, and it stopped responding.

We requested the founder how Ollie is dealing with such issues, given that the majority shoppers solely give a brand new app or expertise one shot. If it’s damaged, they merely transfer on.

Lennon agreed this is a matter with AI total. “That is the problem with LLMs, basically — as a result of they’re stochastic [i.e., involve probability], they’re inherently unreliable,” he mentioned. “Now we have to primarily construct the harness — the agent harness — in a defensive approach to catch and stop these issues… It’s virtually like there’s simply 1,000 cuts that you just’ve obtained to resolve first.”