AiThority Interview with Yoav Regev, CEO and co-founder at Sentra

Yoav Regev, CEO and co-founder at Sentra comments on the security protocols that data teams should focus on more as AI becomes mainstream to this critical business function in this AiThority interview:

____________

Hi Yoav, tell us about your role at Sentra and your journey through the tech ecosystem.

I am currently CEO and co-founder at Sentra. My journey to founding Sentra was shaped by decades of experience securing sensitive data in complex environments. I served as the head of the cyber department in Unit 8200, the elite Israeli Military Intelligence, for nearly 25 years before transitioning into entrepreneurship.

Throughout my tenure at Unit 8200, it was clear that sensitive data had become the most valuable asset to organizations and adversaries. I noticed that in the private sector, the enterprises that were leveraging data securely were generating new insights, developing new products, providing better experiences, and separating themselves from the competition. On the other side, as data became more valuable, it also became a bigger target for threat actors. As the volume and the impact of sensitive data grew, so did the importance of finding the most effective way to secure it.

After ending my service at  Unit 8200, I joined Sentra’s co-founders, Asaf Kochan, Ron Reiter, and Yair Cohen, to create a data security company for the cloud and AI era. We see that as the major problem for most of the organizations in the world. That is the main pillar driving their business.

We’d love the highlights from your fresh funding round, what can users expect in terms of product enhancements in the near future?

We recently closed a $50 million Series B, bringing Sentra’s total funding to over $100 million. The round was led by Key1 Capital and participated in by our existing investors, Bessemer Venture Partners, Zeev Ventures, Standard Investments, and Munich Re Ventures. Leading up to the funding, Sentra experienced a more than 300% year-over-year increase in revenue and the addition of multiple new Fortune 500 customers.

Building on that momentum, we launched our Data Security for AI Agents solution. Designed to address the emerging challenges associated with AI assistants, our approach ensures that organizations can embrace AI innovation securely and responsibly. Key capabilities include automatic discovery of AI agents and their connected knowledge bases along with classification of the sensitive data within, real-time monitoring for unauthorized data access, and detailed visibility into AI-generated responses to prevent data leaks and ensure compliance.

As data’s journey evolves, so does Sentra’s product roadmap. Users can expect that we’ll continue to innovate our portfolio to reflect the needs of data security, privacy and governance teams – while also doubling down on the core capabilities that set Sentra apart: best-in-class data discovery, highly accurate classification, and broad coverage across all environments.

Also Read: AiThority Interview with Pete Foley, CEO of ModelOp

What should organizations be doing more of to ensure better data hygiene and data cleaning processes at a time when most boast of unhealthy data?

The number one step to fixing your data security issue is recognizing that data is your most valuable asset — and it is continually moving around your clouds and greater ecosystem, so a new more agile/scalable approach is required. Once you accept that, organizations can focus on a few key steps:

1. Get full visibility of their entire sensitive data. Before any meaningful data security work can begin, organizations must have a clear, real-time view of where sensitive data lives across their cloud, SaaS, and on-prem environments. Without this visibility, it’s impossible to assess risk, apply proper controls, or meet compliance requirements. Discovery must be continuous, not a one-time effort.
2. Automate security tasks. Even with the proliferation of AI, some organizations are hesitant to adopt the technology for their security stack. I recommend that security teams overcome this fear and use AI and other automation tools to eliminate repetitive and resource-intensive tasks such as data discovery and classification.
3. Uplevel sensitive data protection. Ensure proper data security posture by identifying sensitive data no matter where it resides. Put controls in place so that sensitive data is only accessible to authorized personnel. Continually monitor data access for unusual activity. Automate the creation of support tickets for security incidents, initiate automated remediations via integrations with the security stack controls, and prioritize high-risk alerts.
4. Implement risk-based permissioning. There have to be clear procedures for managing authentication credentials. Apply actions based on risk levels. For example, quick access revocation for low-risk cases and verification for critical credentials.
5. Have concrete data mapping strategies in place. With well-defined data mapping strategies, organizations can ensure data is stored in the appropriate places and complies with regulations.
6. Assign accountability. Encourage your employees, regardless of role, to take personal responsibility for data security.

Fortunately, data security posture management (DSPM) solutions can automate and tackle all of these steps for organizations, reducing the burden on security teams.

What added practices should data and marketing/ops teams be focusing on when they use AI to enable different types of workflows?

Before data and marketing teams incorporate any AI into workflows, organizations must sit down and outline a proactive security approach for AI moving forward. With this, companies can make sure that AI enhances, not compromises, security. To do this, organizations should:

• Create strict guidelines for data sharing and data hygiene within AI platforms
• Share clear AI usage policies based on zero trust and least privilege principles
• Ensure they control which data gets into the AI systems/models
• Integrate AI security into company-wide cybersecurity training to educate employees on the latest AI threats

Also Read: AiThority Interview with Dr. William Bain, CEO and Founder of ScaleOut Software

What other security protocols should data teams be conscious about as AI becomes more mainstream to their processes?

No model is entirely immune to privacy and security risks in real-world scenarios, so leveraging automated solutions for ongoing monitoring is crucial to maintaining AI security.  It’s important to have security embedded into AI applications from the start. Doing so sets developers and security teams up for success long before an AI application goes to market. Key steps include identifying where sensitive data resides and ensuring good security posture, removing or de-identifying sensitive data from training sets, testing models for adherence to privacy regulations during pre-production, and implementing continuous monitoring throughout the development lifecycle.

Five thoughts on the future of AI before we wrap up?

1. AI regulations are coming. Colorado set the standard last year for the first comprehensive AI legislation focused on consumer protections and safety. In 2025 alone, there have been at least 550 AI bills introduced in 45 states and Puerto Rico. Just like we saw with GDPR, HIPAA, and CCPA in the data realm, we’re going to see organizations having to navigate AI governance as lawmakers work to create policy to keep the technology safe.
2. AI is going to increase instances of shadow and duplicate data. As AI adoption continues, data will proliferate faster than we have seen with the cloud, leaving shadow data in its midst. Shadow data is any data that exists outside of a secure data management framework. Because it often exists without the knowledge of or proper management by the security team, it is considered a top target for threat actors. Organizations have to use security controls that stay with the data — regardless of where it goes.
3. Least privilege access will move from a nice-to-have to a must-have for AI systems. Security teams have to enforce the principle of least privilege to AI systems. This looks like only giving AI models access to the data it needs and no more, ultimately minimizing risk of misuse, data leakage, and breaches.
4. Protecting the integrity and privacy of data in large language models (LLMs) will become essential. Organizations need to have responsible and ethical AI applications, and the only way to do this is to hold a steadfast commitment to integrity and privacy. By implementing some of the best practices I mentioned above, organizations can mitigate risks associated with data leakage and unauthorized access.
5. We’re only beginning to understand AI’s potential — and its downfalls. Agentic AI is coming and its autonomy is capable of transforming business critical operations, increasing productivity, and lowering costs. However, its autonomy also introduces significant security risks. It’s going to require collaboration across the security ecosystem to keep AI threats at bay.

[To share your insights with us, please write to psen@itechseries.com]