Back to Blog
Minute Read

Introducing Linea AI: A revolutionary AI-first approach to data security

Abhi Puranam

The AI revolution that kicked into overdrive last year has taken the world by storm, and as the development and productization of Large Language Models and other AI technologies continue – the revolution shows no signs of slowing down.

In this article

Last year, we rapidly built and released Cyberhaven for Generative AI – to help organizations understand how their employees were using these new AI products and control risky data flow. But behind the scenes, our research and engineering team have been studying and applying the underlying principles of the recent advances in AI technology to augment Cyberhaven’s Data Detection and Response capabilities with an AI-first approach to data security. We are excited to share the results of their efforts - Linea AI.

What is Linea AI?

Linea AI is an end-to-end AI agent for data security, acting both autonomously and in concert with your team to detect, prioritize, triage, investigate and prevent risks. Here’s what it can do:

Detect risks

Linea AI analyzes all data flows within your company and detects insider risks that traditional policies and other AI techniques overlook.

Cyberhaven’s approach to data security so far has centered around giving security teams detailed context about data movement and usage in the form of data lineage. By collecting every event surrounding data usage and connecting the resulting dots, Cyberhaven enables security teams to uncover unknown unknowns surrounding data usage, classify any type of sensitive data, and rapidly investigate employee behavior.

With this robust knowledge of how data moves in corporations of all types and sizes, we’ve built a new type of model at the heart of Linea AI. Similar to how Large Language Models predict the next word in a sentence, our first-of-its-kind Large Lineage Model (LLiM) predicts the next likely action in corporate workflows. You can learn more about our approach to building this model in our engineering blog by Cyberhaven co-founder and CTO Volodymyr Kuznetsov.

This new model enables the detection of risky deviations in data handling, accurately identifying risks by understanding in detail the sensitivity of data involved in a flow and how typical a destination may be for that data. Consider the following examples from Linea AI’s beta:

In this example, Linea AI utilizes a semantic analysis of data lineage to flag this data movement as risky based on the true origin of the data being a draft of the board deck and a non-corporate Slack message being an atypical manner of sharing this kind of data. 

In this example that another data security tool may incorrectly alert on, Linea AI correctly determines that this  data movement is low risk because it understands this firmware file is from a commercially available 3-D printer and that the installation process of such firmware requires a USB.

This is the power of risk detection in Linea AI, access to rich context and metadata regarding how files and data move and the intelligence and knowledge to reason what the sensitivity of the data and riskiness of any action is – enabling more accurate risk detection than legacy approaches.

Prioritize risk

Linea AI understands risk the way the best security analyst would, and prioritizes the incidents with the greatest impact to the business so your people can focus on what matters most. Linea AI looks at risks it detects and those generated by your policies and ranks them so your analysts can stay on top of the most pressing risks to your business.

Analyze and summarize incidents

Linea AI changes the game when it comes to incident response. Completely autonomously, Linea AI performs its own investigation into the lead-up to an incident and looks for additional context in other data flows the user has performed. Linea AI also analyzes behavior observed across your company concerning similar data types and applications to uncover additional context. It summarizes all of these details for you so you can review the results and make a determination on how valid the incident is and what next steps are required.

Smart remediation

Linea AI automatically generates the ideal remediation steps based on the details of the incident and its severity and automates those responses. So if the right response is to contact the employee’s manager and HR, it automatically extracts contact information from your employee directory and drafts an email for you – all you have to do is click send.

Guided prevention

Finally, Linea AI facilitates real-time intervention and policy deployment, proactively reducing risk and bolstering your company’s security posture.

Linea AI in action

During the closed beta, Navan experienced transformative benefits from Linea AI. The platform delivered greater coverage, detecting 40 risks in a month that would have been unnoticed by policies, with an 85% accuracy rate. Moreover, Linea AI accelerated resolution times by 5x, analyzing broader context to help analysts understand incidents and take appropriate action.

You can learn more about these results by watching the Linea AI keynote, featuring Prabhath “PK” Karanth, VP of Security and Trust at Navan.

{{ promo }}

Looking ahead with Linea AI

Linea AI is set to become generally available in May 2024. As we prepare for this exciting launch, we remain committed to driving the AI revolution in data security and helping companies innovate securely.

Watch the Linea AI launch event
Watch now