Enterprises are accumulating vast amounts of data to propel their AI endeavors, yet concurrently grappling with concerns regarding data accessibility and privacy. PVML presents a compelling remedy by melding a ChatGPT-like tool for data analysis with the security assurances of differential privacy. Leveraging retrieval-augmented generation (RAG), PVML facilitates access to corporate data without the need for data relocation, thereby mitigating additional security risks.
The Innovation Behind PVML: Differential Privacy in Action
Recently securing an $8 million seed funding round led by NFX, with contributions from FJ Labs and Gefen Capital, PVML stands as a testament to innovation in data privacy and accessibility. Founded by the dynamic duo Shachar Schnapp (CEO) and Rina Galperin (CTO), PVML draws from their extensive expertise in computer science and AI, honed through their tenure in prominent corporations.
“A lot of our experience in this domain came from our work in big corporates and large companies where we saw that things are not as efficient as we were hoping for as naïve students, perhaps,” Galperin stated. “The main value that we want to bring organizations as PVML is democratizing data.”
Differential privacy, a cornerstone of PVML’s approach, ensures individual user privacy within expansive datasets while furnishing robust mathematical guarantees. This technique introduces controlled randomness into the dataset without compromising data analysis integrity.
In contrast to conventional data access methods fraught with inefficiencies and overheads, PVML’s differential privacy framework offers a streamlined solution. By preserving the original data integrity, PVML obviates the need for cumbersome data redaction, thus facilitating seamless integration into AI applications.
“The current knowledge about differential privacy is more theoretical than practical,” Schnapp emphasized. “We decided to take it from theory to practice. And that’s exactly what we’ve done: We develop practical algorithms that work best on data in real-life scenarios.”
Empowering Data Security, Collaboration, and AI Integration
PVML’s differential privacy extends beyond theoretical discourse, manifesting in tangible benefits for data analysis. Employing retrieval-augmented generation (RAG), PVML empowers users to engage with data securely, minimizing the risk of sensitive data leakage.
In addition to enhancing data security, PVML’s innovative approach unlocks new avenues for data sharing and monetization. PVML spurs AI integration across sectors by promoting collaboration and enabling controlled data access for third parties.
“In the stock market today, 70% of transactions are made by AI,” noted Gigi Levy-Weiss, NFX general partner and co-founder. “That’s a taste of things to come, and organizations who adopt AI today will be a step ahead tomorrow. But companies are afraid to connect their data to AI, because they fear the exposure — and for good reasons. PVML’s technology safeguards data, enables monetization, and fosters future AI integration by democratizing access to information.
See also:
