Patricia Thaine, Private AI: “When building something, only jump on waves if it fits your vision”

Chances are that you’ve already used generative AI services such as ChatGPT for things you shouldn’t. Perhaps you copied and pasted information that belonged to your employer, personal data that wasn’t yours to share. While this is a new threat, it joins a long line of such problems. So how do you, as a business, retain visibility and control?

We suspect this is the kind of question that keeps Patricia Thaine, the Co-Founder and CEO of Private AI, awake at night. For, as we’ve frequently seen in our TakeOff interviews with tech founders, Patricia saw the opportunity through a personal passion. She is a Computer Science PhD Candidate at the University of Toronto (on leave) and a Vector Institute alumna. Tellingly, her R&D work is focused on privacy-preserving natural language processing.

It’s partially through this insight that Patricia founded Private AI. A Microsoft-backed startup that has already raised over USD $10 million of funding…

What’s your elevator pitch?

At Private AI, we’re revolutionising data privacy in an era of chaotic data lakes, constant breaches, compliance violations and increasingly stringent data protection regulations. With just a few lines of code, organisations can integrate Private AI’s products anywhere within their software pipeline to have minute control and visibility over the type of personal data they collect, store and share – all across 49 different languages and multimodal sources like text, audio, images and documents. 

What made you launch a startup?

Patricia Thain CEO Private AI
Patricia Thaine, Co-Founder & CEO of Private AI, is also a Computer Science PhD Candidate at the University of Toronto

Our inspiration stemmed from the pressing need for developers to seamlessly incorporate privacy into their software. This integration is the difference between accessing important data needed to improve their systems or not getting any data at all.

So, we embarked on our journey in 2019 to address this critical gap in the privacy landscape, where not even the very core requirement of data protection regulations — the identification of personal information — was being addressed for 80-90% of the data collected, since it consists of unstructured data.

What problem are you trying to solve?

Our mission is to seamlessly enable a privacy layer for any software, on any device, for any type of data. The core problem we are solving first is that of identifying personal information within data streams accurately, encompassing a wide range of formats such as text, documents, audio and images.

Can you talk us through your journey so far? What’s a major milestone you’ve reached?

We started in 2019 with an idea and a passion for responsible innovation. In October 2020, we had our first customer, and now, less than three years later, we have expanded our reach across Europe and beyond, with over a billion API calls.

Our journey, albeit still new, consolidated us as the best privacy-preserving technology product in the world, proven by both internal benchmarks and numerous POCs, where our accuracy always comes on top. Our newest product, PrivateGPT — the privacy layer for ChatGPT — has proven to be a great success within the first three months since its launch.

On the fundraising side, the first major milestone happened in 2021, with our USD $3.2 million seed funding co-led by M12 (Microsoft’s fund) and Forum Ventures. A year later, in 2022, we raised our Series A funding round, securing $8 million.

The pinnacle of our journey so far materialised earlier this year, when we received recognition from The World Economic Forum, being nominated as a 2023 Technology Pioneer. This nomination not only acknowledges our current value but also opens doors for more innovation and opportunities on the horizon.

Who are your main competitors and what distinguishes your startup from them?

The main competition we come across for our core product of personal information identification tends to be AWS Comprehend, Google DLP, Microsoft Presidio and internal solutions. So far, we have always come out on top of these during a POC thanks to our high accuracy, thorough multilingual support, and (in the case of Comprehend and Google DLP) our ability to deploy directly within customer environments and never getting sent any of their data.

The toughest sale is when folks have an internal solution that doesn’t work very well, but the original creator is still at the company and doesn’t want all the work they put into it to go to waste. We also find that the difficulty of getting this core problem right can occasionally be vastly underestimated and several companies who believed they could build the tech themselves end up coming back to us after a few months or a couple of years.

In terms of our latest product, PrivateGPT – which removes personal information and (soon) company confidential information before sending prompts through an organisation’s LLM of choice – it’s still a bit early to tell. Professionals are still grappling with what they want out of LLMs. That said, lots of landing pages have popped up, and we often get asked how we compare with products that don’t even exist yet.

How has the startup scene in Toronto helped your own startup’s development?

Toronto is home to a number of excellent investors and accelerators. We got very significant initial (and continuous) support from Forum Ventures, which was an introduction from one of our lawyers at Osler, Chad Bayne. Forum Ventures focuses on early-stage B2B SaaS startups. The Creative Destruction Lab was also incredible at exposing us to some of our early adopters and even to someone who ended up becoming our board member.

Where do you hope your startup will be in ten years?

In ten years, we hope our name will be a synonym for privacy, security and trust. We hope to be recognised as an authority in privacy-enhancing technologies and serve as the definitive privacy layer for software.

Additionally, even though we currently already support over 50 languages, we envision a future where the language you speak doesn’t determine the amount of privacy that you’re entitled to due to technological barriers — we see ourselves becoming the privacy solution across the globe, with organisations proudly displaying a Privacy Powered by Private AI logo on their products.

What would you say to potential investors reading this interview?

Before proceeding with any investment, ask the company about its policies regarding customer data management. Any mishandling can lead not only to significant fines but can also cause irreparable damage to a brand’s reputation, leading to a deterioration of client trust and, ultimately, a decline in revenue.

Privacy is no longer a nice to have. Now, in addition to being a key competitive advantage, it has finally become an essential aspect to address in technology acquisition processes. AI is evolving rapidly, and this advancement has instigated major concerns regarding privacy. Generative AI like ChatGPT has already been under scrutiny by regulatory bodies, and AI regulations are a subject of constant debate for businesses and individuals alike.

Investing in a company that prioritises data privacy and acknowledges it as a core business principle is likely to outlast its competitors.

What advice do you have for aspiring entrepreneurs and anyone looking to launch their startup?

Think of the long term. When building something, only jump on waves if it fits your vision.

Our thanks to Patricia for taking the time to do this interview. If you’ve enjoyed this interview as much as we did, you might want to consider checking out the following:

Avatar photo
Tim Danton

Tim has worked in IT publishing since the days when all PCs were beige, and is editor-in-chief of the UK's PC Pro magazine. He has been writing about hardware for TechFinitive since 2023.