The NuGuard AI marketing leader on technical fluency, building demand for a new category and keeping the human connection in AI driven outreach.
Sam Kumarsamy’s background spans over two decades driving international growth across marketing, sales, and business development for organizations including Deepwatch, OpenText, MobileIron, Gigamon, Infoblox, Check Point, Cisco, and E&Y, alongside several venture-backed cybersecurity startups that achieved successful exits. Kumarsamy now directs go-to-market strategy and execution at NuGuard AI, a self-funded company focused on ensuring the safety and security of agentic AI before production.
He sat down with Energize Marketing’s Solange Deschatres, a fellow cybersecurity marketing veteran, and their conversation ranged freely from the technology behind an emerging security category to the day-to-day realities of creating demand with a lean team.
In this Q&A, Kumarsamy discusses why cyber marketers need firsthand technical knowledge, how NuGuard is introducing the issue of “behavior gap in agentic AI,” what his early AI-assisted outreach is and is not showing, and why personal relationships still matter when every buyer is receiving more automated messages.
Energize Marketing: You’ve always made a point of understanding where technology is going. How has the rapid change in AI altered your approach to marketing?
Sam Kumarsamy: Marketing has changed within the last six months. It’s incredible. Last year at Deepwatch, I was using ChatGPT for content creation and then adding my own brand voice. Now you can create agents (using a variety of LLMs) to do your content creation and campaigns. But you need to know where you can use AI and where you need the human touch. Experiment today. At least automate some of the mundane tasks like curating insights and research. Don’t keep going into a prompt and doing research each time manually. Try to create an agent for that.
I’m learning too. I’m by no means an expert. My colleagues are teaching me new tricks. But whether you’re in marketing or any type of high tech, you need to learn the basics of AI and apply yourself.
Energize: Cybersecurity buyers can move from a technical question to a business risk question in the same conversation. How do you speak credibly to both audiences?
Kumarsamy: I believe in doing the demo myself. I happen to have a technical and a business background. For the technical people, you need to be able to show the demo and the speeds and feeds. For the business audience, you need to talk about outcomes. How are you improving the risk posture? How are you mitigating risk?
When I talk to developers, it’s important to understand and speak their language like accelerating their time to production. With decision makers like CTOs and CISOs, the discussion is about the risk posture and audit ready evidence reports before an application is released. You need to understand both sides.
Energize: What would you tell a good marketer who is new to cybersecurity and wants to develop that technical depth?
Kumarsamy: Go and read a lot. In your spare time, go to YouTube and learn the basics of cybersecurity. Allocate some time every day. Attend local meetups and listen to what the latest trends are. Be part of that. Be proactive. Don’t just sit and rely on AI because you’re going to get questioned about what you know and what you produced.
You don’t need to know coding to start learning AI either. Learn the basics. Learn to create an agent, even if it’s just scheduling your day. Go beyond prompting. The more you learn, the more technical you are, the more useful you can be and indispensable to your employer.
Energize: If a marketer uses AI to draft cyber content, how can they catch an answer that sounds good but misses something important?
Kumarsamy: Product marketers still need technical depth. Before creating cybersecurity content, study the subject or ask a knowledgeable expert to verify the facts. When using tools such as Claude or ChatGPT, challenge their output rather than accepting the first response, which often reflects what your prompt suggests you want to hear.
I always double-check the content to see if it’s missing anything by putting the same question in several LLMs. I use them for different reasons: one for content, another for diagrams, for example. Then I put my personal brand on it. Some of the research statistics could be old, so question those too. You need to know enough yourself to question what comes back.
Energize: NuGuard talks about a “behavior gap” in AI applications. How do you explain that problem to an audience that may still be learning the category?
Kumarsamy: Awareness is the first thing. When you say “behavior gap,” it’s not a category that is clearly defined yet. Before you release an AI application, you’ve got to look at its behavior, test and validate it to make sure it’s behaving as intended. When you set up behavior policies or guardrails, make sure you define what it should do and what it shouldn’t do.
Consider a banking application. It may process financial transactions, but a policy could require human approval for amounts above a set threshold. Likewise, a financial application should not provide healthcare advice. These familiar examples show why clear boundaries must be defined and tested.
Energize: How do you build demand when the buyer may not yet be searching for “behavior validation” or have a budget assigned to it?
Kumarsamy: We’re starting with awareness and education. We’ve spoken with analyst firms like IDC and Omdia about the agentic behavior gap and validation because a third-party perspective helps people take a new problem seriously. Both firms have written a profile on NuGuard AI covering the agentic AI behavior gap. If it’s just the vendor promoting a category, someone may say, “I’ve never heard of it.” If an analyst says it’s worth watching, that creates awareness among buyers.
At the same time, the first 10 customers typically come from your personal connections. The first four proofs of concept we’re doing are through design partners who are developing customer facing agents. We’re also doing outbound demand, and we’re a product-led growth company. Developers can download our open-source CLI software from GitHub and we have had over 50K downloads so far, and the enterprise version targeted at CTOs, CISOs and VP of Engineering has a free trial. Rather than building a sales team right now we can ask interested prospects to download and test it themselves.
Energize: You’re speaking to developers, security leaders and executives. How does the path from initial interest to a buying conversation differ across that group?
Kumarsamy: The influencer is usually a developer or an engineer who is comfortable using the NuGuard open-source CLI version. NuGuard helps developers accelerate time to production by validating the agent behavior using red teaming and pen testing. The CTO, CISO and VP of engineering need to see the agentic risk posture and audit ready report aligned with OWASP Top 10, NIST RMF and PCI . The enterprise NuGuard AI Trust Platform was purpose built to address these executive needs. So, in summary the free version provides the influencers a way to try it, and you meet the needs of the decision makers with the enterprise SaaS version
For us, the enterprise customer is the eventual goal, but at this stage we’re also targeting SMBs that are developing customer facing agents for regulated industries like fintech, health tech and eCommerce. We are focused on who is building these applications and who has an urgent reason to validate them.
Energize: You’re using AI tools in your own outreach. What have you learned so far about whether those signals translate into real interest?
Kumarsamy: I’m still in the experimentation process. I can’t tell you, “Oh, this AI is working great.” The AI-native GTM tools can give you information about a company, contact, AI buying signals and intent I’m using that to focus outreach. But I’m still personalizing each message myself because AI should just give you a starting point.
I started this campaign two weeks ago, so I can’t tell you for sure. Maybe a month down the road I can tell you how many responded. Some executives have connected with me on LinkedIn, and we are following up with requests to meeting or trial our platform. A LinkedIn connection doesn’t mean anything on its own. People are buried in their own work, and you must give them time to respond.
Energize: What happens to outreach when buyers are getting so many automated emails and LinkedIn messages?
Kumarsamy: Everybody is getting bombarded with email. I get connections saying, “We can help you with demand generation, call us,” and then I receive another message every day. It gets annoying. You must think about how you space those cold and LinkedIn emails.
People can sometimes recognize an AI generated campaign even if you put a personal touch on it. So, try to meet, even over Zoom. Put a face to a voice. If you meet someone and explain what you do, there is that human touch. All four proofs of concept we’re doing now came through our previous personal contacts. Nothing came through email so far. That’s why I say the first 10 customers typically come from your personal connections.
Of course, you can’t rely only on that to build a pipeline unless you have a big sales team. But for an early stage, self-funded company, all the executives must wear several hats. I’m doing marketing, business development, sales and events. As a marketing leader, you must wear several hats.
Energize: What convention in AI or cybersecurity marketing would you challenge right now?
Kumarsamy: People are developing agentic AI at super speed and saying, “They have a budget for AI, let’s sell them AI.” But you need to ask whether it’s safe and secure for them to adopt. Ensure the agentic AI policies are in place and tested and validated before you pitch it.
In your marketing, when you talk about how AI is improving everything, show that you’ve done the research and have working examples because you’re testing it yourself. For critical actions, you still need humans in the loop. I think we’ll get to close to 100% AI automation soon, but I would be hesitant to trust it 100% today.
Energize: Buyers are also using AI to research vendors and make shortlists. How does that change the content you create?
Kumarsamy: We’re thinking about SEO alongside answer engine optimization and generative engine optimization. In our whitepapers, we ask questions and answer them. That’s more educational, and it helps explain a new topic: What is the behavior gap? What does behavior validation mean?
We try to use the terms people already know, such as AI security or AI governance, while introducing new terms like “AI oversight” and “behavior validation.” Sometimes a new term creates curiosity. You must start somewhere and take the risk and experiment. We’re trying to make our content useful for people looking for answers, whether they find it through search or an AI tool.
For Kumarsamy, the work of marketing an emerging cyber category begins with learning the technology well enough to question it. AI has accelerated his research, content production and outreach, but he is still testing which signals lead to conversations. The early opportunities he described came through people who already knew the team, a reminder that technical credibility and human relationships remain central to how new ideas reach the market.