Ollie has achieved SOC 2 compliance, becoming one of the first mainstream family-focused AI assistants to secure an independent audit confirming formal controls over customer data.
For a personal assistant to function effectively, it must access significant amounts of user information. Ollie argues that this requirement does not necessitate surrendering all data to the provider.
While some enterprise tools offer privacy protections, Ollie distinguishes itself by ensuring the service does not harvest personal details for AI training or other corporate purposes.
“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,” Bill Lennon, co-founder and CEO, told TechCrunch.
“We’re not sharing your data with anyone,” he added.
Many current assistants integrate with daily life via text messages. Ollie’s privacy stance positions it as a potential differentiator in a crowded market.
Competitors include Poke, acquired by Cognition, alongside Fambot, Ohai, Folk, Saner.ai, and Tomo. Work-focused tools like Town, Lindy, and Reclaim.ai manage calendars and email. Instinct recently secured $350 million in funding at a $2.5 billion valuation before launching.
San Diego-based Ollie has raised a $7.5 million seed round from Khosla Ventures and AI House, a modest sum compared to the capital flowing into rivals.
Users often struggle with how much personal data to surrender. Instinct faced criticism for its Terms of Service, which granted a “perpetual and irrevocable” license to access, use, host, cache, store, reproduce, transmit, display, publish, distribute, and modify user materials, including for model training.
Lennon’s approach differs. The assistant connects to calendars and email to organise family schedules. It offers tools for meal planning, grocery shopping, task tracking, appointment booking, and bill payments via group chats. Future updates may include household budget management.
All this access requires trust. Lennon states this is the value-add compared to rivals.
“We’re not sharing your data with anyone. This is super sensitive, and that is necessary to win the trust of the users,” Lennon said.
Beyond SOC 2 compliance, Ollie does not request usernames or passwords for tasks. When logging into a website on the user’s behalf, it uses a cloud browser and sends a link to a remote session. Payments and purchases follow a similar process.
This method ensures security but requires users to log in for specific tasks. Lennon believes technology will improve this aspect over time.
“This is… new territory. I think in the future, we will do some form of hard tokenization, in a secure way, so you’re not going to have to re-enter [your information] every time,” he said.
“That’s frontier stuff… we want to find the right user experience that balances convenience and trust.” Lennon said.
Building this trust is essential before users allow access to sensitive materials, such as bank accounts via a connector like Plaid.
Lennon holds a Ph.D in AI and has a fintech background. He previously sold Groundwork, a neobank for nonprofits, in 2021. That experience may aid Ollie as it expands into money management.
It remains unclear if consumers want this level of assistance.
Lennon remains hopeful, noting Ollie’s retention curves match leading AI subscriptions regarding paid subscribers. He did not disclose the number of users or paid customers, citing a highly competitive space.
These systems can be inconsistent. A test of Instinct to price and book a hotel resulted in incorrect rates. Ollie suffered an infrastructure outage during my testing, causing it to stop responding.
We asked the founder how Ollie handles such issues, given that consumers typically give a new app only one chance before moving on.
Lennon agreed this is a broader AI problem.
“This is the challenge with LLMs, in general — because they’re stochastic [i.e., involve probability], they’re inherently unreliable,” he said.
“We have to essentially build the harness — the agent harness — in a defensive way to catch and prevent those things… It’s almost like there’s just 1,000 cuts that you’ve got to solve first.” Lennon said.
What it means
Users now have a choice between assistants that trade privacy for convenience and those that prioritise data security. Ollie’s model requires manual login steps for tasks currently, but it avoids training data harvesting. Success depends on whether users will tolerate the friction of logging in repeatedly in exchange for guaranteed privacy.
