Company Profile: PVML
PVML is a deep-tech cybersecurity and data access startup headquartered in Tel Aviv, Israel. Founded in 2021 by Shachar Schnapp and Rina Galperin, the company focuses on the intersection of artificial intelligence and data privacy. PVML provides a platform that allows enterprises to connect, access, and analyze sensitive data sources?such as production databases?using AI and SQL without exposing raw information or violating privacy regulations.
The company emerged from stealth in early 2024 with $8 million in seed funding led by NFX, with participation from FJ Labs and Gefen Capital. PVML?s mission is to "decouple the data stack from the AI stack," enabling organizations to treat data as a strategic asset without the traditional security and compliance bottlenecks.
Continue…Core Mission and Technology Focus
The primary objective of PVML is to provide "The Virtual Database for AI." Traditional data anonymization methods (like masking or PII tagging) are often binary, manual, and prone to re-identification attacks in the age of generative AI. PVML addresses this by shifting the paradigm from protecting the data itself to protecting the output of queries.
Differential Privacy (DP)
The technological backbone of PVML is its proprietary implementation of Differential Privacy. This mathematical framework introduces carefully calibrated "statistical noise" to query results.
* Mathematical Guarantees: DP provides a provable guarantee that no single individual's data point can be extracted or inferred from the output of a query.
* Zero Overhead: Unlike homomorphic encryption, PVML?s DP technology does not require massive computational power, allowing for real-time analysis of live data.
* Accuracy Preservation: The system is designed to maintain the utility of the data, ensuring that while individual records are hidden, the overall statistical insights remain accurate for business intelligence and machine learning.
Products and Services
PVML offers a cloud-based platform that acts as a secure permission-enforced layer between an organization's raw data and its users or AI models.
1. AI-Ready Virtual Databases
PVML turns internal, fragmented data sources into a unified, secure virtual database that can be accessed without moving or duplicating any raw data.
* Direct Connection: Admins can connect any database (SQL, NoSQL, etc.) using a simple connection string, with no changes required to the existing infrastructure.
* Centralized Administration: Provides a single point to monitor and control all data access, offering a high-level view of queries, permissions, and user requests.
2. Analyze Data with AI (Natural Language Interface)
A standout feature of the platform is its ability to allow non-technical users to query complex databases using plain English.
* Text-to-Query Translation: The platform uses Large Language Models (LLMs) to understand natural language requests and translate them into SQL.
* Explainability & Anti-Hallucination: PVML generates a "recipe" for the query that the user can verify, ensuring the results are grounded strictly in the organization's real data rather than fabricated by the AI.
* Automated Visualization: The platform can automatically generate graphs and charts from the query results for easier insight consumption.
3. Automated Privacy and Permissions Management
PVML automates the "manual chaos" of data access requests.
* Just-In-Time (JIT) Access: Admins can grant temporary, scoped permissions for specific data tasks.
* Permission Enforcement: The platform automatically detects and fixes excessive, unused, or conflicting permissions with "one-click" security actions.
* Output-Level Protection: Privacy policies are enforced in real-time at the output level, regardless of whether the user is using a BI tool, an API, or the AI chat interface.
4. Data Monetization and Sharing
For companies looking to unlock revenue from their data, PVML provides a framework for secure collaboration.
* Safe Data Sharing: Enterprises can allow third parties or partners to run analytics on their sensitive datasets without ever exposing the raw underlying records.
* Monetization Framework: Transitioning from selling raw data (which is high-risk) to selling value-added insights (which is secure and compliant).
Strategic Standing and Partnerships
As of early 2026, PVML has integrated deeply into the global AI and cybersecurity ecosystem:
* Defense & Intelligence: In late 2025, VisionWave partnered with PVML to build secure, AI-powered defense intelligence systems, leveraging PVML?s ability to handle highly classified data securely.
* Multi-Industry Adoption: The platform is utilized by leaders in fintech, telecommunications, and healthcare, where the cost of a data breach is exceptionally high.
* Investor Support: In addition to institutional VCs, the company is backed by former executives from Facebook, T-Mobile, and Intel Ignite.
Leadership Team
- Shachar Schnapp, Ph.D. (Co-founder & CEO): A computer science expert specializing in Differential Privacy.
- Rina Galperin, M.Sc. (Co-founder & CTO): A veteran software engineer with a background in building large-scale data products for corporate environments.
