AI Startup Jazz Secures $61M to Rethink Data Loss Prevention

A data loss prevention startup founded by a former Axonius leader raised $61 million to analyze every data transaction across systems, people and business processes.

See Also: AI Is Transforming the Chief Data Officer Role

The Glilot and Team8-led Seed and Series A funding will help New York-based Jazz develop agents that analyze the data itself, the systems involved, the people interacting with the data and the business processes surrounding the transaction, said co-founder and CEO Ido Livneh. Jazz also collects signals from endpoints to understand the sequence of events surrounding data access and movement.

“DLP is not a new problem. Every board, every security organization knows about the risk of DLP,” Livneh told Information Security Media Group. “But the fact of the matter is that the existing solutions don’t work, and when you talk about DLP with CISOs it’s really hard to find anybody that loves it. Everybody hates their DLP, to be honest, because they still get hit again and again and again.”

Jazz, founded in 2024, employs 44 people and emerged from stealth in conjunction with the funding announcement. The company has been led since its inception by Livneh, who spent eight months as vice president of product for asset intelligence startup Axonius and nearly three years as vice president of product for data security posture management startup Laminar prior to its August 2023 sale to Rubrik for $104.9 million (see: Rubrik Buys Startup Laminar to Unify Cyber Posture, Recovery).

Where Traditional Data Loss Prevention Comes Up Short

Traditional DLP systems rely heavily on pattern matching, keyword rules and regular expressions, which Livneh said can sometimes spot obvious cases of data exposure but lacks the contextual understanding necessary to determine whether the data movement actually represents a security risk. Because rule-based systems can’t fully interpret context, they tend to generate large numbers of false positives.

“The way we see the market, about 30% even tries to implement and run a DLP program, and even they know that what they have in place is best effort at best,” Livneh said.

Jazz’s architecture consists of multiple artificial intelligence agents that analyze a single data transaction from different perspectives to reconstruct the full story behind data movement, Livneh said. The company’s system performs a full contextual investigation of every data transaction using multiple AI agents, with each responsible for data analysis, system context, human behavior and business process context, he said.

“What we’ve built is a DLP investigator that does that work automatically for you in autonomy and it’s deployed at scale,” Livneh said. “It investigates in-depth every data transaction there is. It understands what happened at the level of the data, the systems, the people and the business process, and alerts your organization so you understand what happened, why it happened and the intent of the actor.”

Jazz attempts to encode policies in natural language, which helps the AI system interpret the intent behind the organization’s security policies rather than simply checking for rule violations, Livneh said. For example, if a new business workflow emerges that was never defined in the policy documents, the system can still reason about whether the activity aligns with organizational expectations.

“We have a natural language policy engine that helps describe what’s acceptable and what’s not acceptable in the company,” Livneh said. “That allows Melody to run a human-like assessment of the situation and make decisions on situations that are probably not even explicitly mentioned in the policy. It bridges the gap between day-to-day practices and the policies written in the company.”

How Sensitive Data Is Now Leaving Organizations

The explosion of generative AI tools and SaaS applications has dramatically increased the number of potential data leakage pathways, which makes traditional DLP tools even less effective, Livneh said. AI assistants, productivity tools and SaaS platforms often require users to upload documents, code or internal information and can therefore become channels through which sensitive corporate data leaves.

“Every week we hear about new tools being deployed, especially in the AI and GenAI world, and employees are adopting them even without the approval of the company,” Livneh said. “These are new ways in which data could be shared outside the organization and security teams have a really hard time keeping track and securing all these vectors.”

Sensitive data can leave organizations when engineers upload proprietary code bases and internal documentation into personal AI accounts to generate improvements or alternative versions of the software. Instead of relying on official company CRM platforms, employees sometimes store customer information in personal tools such as Apple Notes or spreadsheets that sync to personal cloud accounts.

“We’ve seen stories about engineers taking the entire code base of the company and a few strategic documents, putting that into a personal account in Claude Code and asking it to rebuild the company’s product,” Livneh said. “We’ve seen sales people managing shadow CRMs and having all the accounts they’re managing on their personal Apple Notes that sync to their personal iCloud account.”

With Jazz, Livneh said humans remain responsible for final decisions involving remediation actions such as disconnecting a device from the network or responding to confirmed incidents. The platform allows administrators to interact with the system by reviewing alerts, discussing uncertain cases and clarifying whether specific behaviors should be considered acceptable, which helps refine policy interpretations.

“There is always a human involved,” Livneh said. “Nobody is disconnecting a computer from the network without somebody making that decision. Melody shows the admin situations that it thinks are outside of policy or that it is not sure of, and they have a discussion about it. Over time, within a few weeks of working with it and talking with it, it feels like it’s molding itself around your specific business.”