In this episode of CyberBytes, we’re joined by Keith Stewart, Founder of Humanix, to unpack one of the biggest blind spots in cybersecurity today: the human layer.

Fresh off the RSA Innovation Sandbox, Keith shares what it takes to distill a deeply technical product into a three-minute pitch and why that challenge reflects a much bigger problem across the industry.

We dive into the infamous MGM Resorts breach, a $100M+ attack that didn’t rely on zero-days or sophisticated malware, but a simple help desk phone call. It’s a stark reminder that today’s attackers aren’t just hacking systems, they’re manipulating people.

Keith breaks down:

• Why social engineering continues to be the #1 attack vector

• The limitations of traditional security awareness training

• How attackers exploit trust, urgency, and human behavior to bypass even the strongest controls

• Why the industry has been “blaming the victim” instead of solving the problem

We also explore how Humanix is approaching this differently, treating social engineering as a detection and response problem, not just a training issue. By using AI to analyse conversations in real time, they aim to identify and stop attacks before damage is done.

Looking ahead, the conversation shifts to the rise of AI-driven help desks and agentic systems, and why this could massively expand the attack surface if left unchecked.

Keith's LinkedIn - https://www.linkedin.com/in/keithrstewart/

Humanix - https://www.linkedin.com/company/humanixai/

Steffen's LinkedIn - https://www.linkedin.com/in/steffenfoley/

Aspiron Search - https://www.linkedin.com/company/aspiron-search/

Today on CyberBytes, we’re joined by Itamar, Co-Founder and CEO, and Russell, VP of Marketing, at Token Security.Fresh off the RSA Conference 2026, where they were named one of just 10 finalists in the prestigious Innovation Sandbox, they’re building at the forefront of one of cybersecurity’s biggest shifts: securing AI agents.While much of the industry is focused on guardrails and prompt security, Token Security is taking a different approach. Their belief? AI agent risk is fundamentally an identity and access problem. And they’re helping organisations understand, manage, and secure what these agents can actually do.Itamar brings a deep technical background, having built security products since the age of 18 and served in Israel’s elite cyber unit 8200, before going on to build and exit a startup alongside one of the founders of Waze. Russell pairs that with hands-on security experience and a track record of scaling messaging and go-to-market at high-growth cybersecurity companies.In this episode, we unpack their journey to RSA Sandbox, how to cut through the noise in a crowded AI security market, and why changing how people think about the problem is the real key to standing out.Itamar's LinkedIn - https://www.linkedin.com/in/itamar-apelblat/Russell's LinkedIn - https://www.linkedin.com/in/rmiller/Oliver's LinkedIn - https://www.linkedin.com/in/oliverlegg/Nina's LinkedIn - https://www.linkedin.com/in/nina-korfias-39981311/Aspiron Search - https://www.linkedin.com/company/aspiron-search/Aspiron Influence - https://www.linkedin.com/company/aspiron-influence/

In this episode, we sit down with Gal Shafir, CEO and Co-Founder of Fig Security, fresh out of stealth and already making waves in the cybersecurity space.From leading global security architecture at Google to building and exiting Simplify, Gal brings deep SecOps expertise and a clear vision for the future of security operations.We dive into Fig Security’s journey from idea to RSA Sandbox finalist in just 12 months, and why they’re taking a fundamentally different approach to the crowded AI SOC market.Instead of focusing on alerts and analysts, Fig is tackling the real problem: the broken, complex infrastructure underneath modern security operations. The result? A new way to think about trust, data integrity, and how AI can actually work in the SOC.We also explore: • Why most AI security tools are solving the wrong problem • The growing complexity of SecOps infrastructure (and why it’s holding teams back) • What SOC modernisation really looks like over the next 2–3 yearsIf you’re a founder, operator, or security leader trying to understand where the market is heading, this is a must-listen.

In this episode, we sit down with Mike McKenna (CRO & Co-Founder of Manifold Security) fresh off their $8M seed round, to unpack what’s really happening at the cutting edge of AI security.AI adoption is exploding, agents are everywhere, and security teams are being told one thing: “Get AI done.”But here’s the problem…No one fully understands the risks.Mike breaks down: • Why AI agents are creating an entirely new attack surface • The shift from securing what models say → to securing what agents actually do • How Manifold is building a low-friction approach to AI security • Lessons from scaling Protect AI during the early generative AI wave • What it takes to build and sell in a category that didn’t exist 18 months ago • Why AI security talent doesn’t exist yet… and how to hire anywayWe also dive into: • The reality of competing in a noisy “AI security” market • Founder-market fit and why this team came back together • How to win your first 50 customers in a brand new categoryIf you’re a cybersecurity founder, operator, or investor, this is a must-listen.Mike's LinkedIn: https://www.linkedin.com/in/mikemckenna9/Manifold Security: https://www.linkedin.com/company/manifoldsecurity/Oliver's LinkedIn: https://www.linkedin.com/in/oliverlegg/Aspiron: https://www.linkedin.com/company/aspiron-search/

Fresh off winning Most Innovative Startup at the RSA Innovation Sandbox, we sit down with the Head of Marketing, Story Tweedie-Yates, at Geordie AI to break down what great marketing actually looks like at an early-stage cybersecurity company.Joining just four weeks before RSA, she was tasked with shaping the story, refining the pitch, and helping translate a highly technical product into something clear, compelling, and stage-ready.In this episode, we dive into: • How to turn complex cybersecurity products into simple, powerful narratives • The process of building a founder’s story and coaching them for high-stakes moments • Why positioning is still broken in most startups and how to fix it • How to think about messaging across different personas without losing clarity • The balance between demand generation, PR, and credibility in today’s market • What’s changed in marketing over the last few years, especially with AI and searchWe also explore her upcoming book, and why most technical founders struggle to communicate what they’ve built and how marketers can bridge that gap.If you’re a marketing leader, founder, or operator in cybersecurity or AI, this episode is a masterclass in storytelling, positioning, and building credibility from day one.Story's LinkedIn: https://www.linkedin.com/in/story-tweedie-yates-550b9234/Geordie AI: https://www.geordie.ai/Nina Korfias: https://www.linkedin.com/in/nina-korfias-39981311/Aspiron Influence: https://www.linkedin.com/company/aspiron-influence/

In this episode, Joseph sits down with Rod Schultz, CEO of Bolster AI, to unpack how cybersecurity has evolved from its earliest days to the AI-powered battlefield we’re seeing today.Rod shares his journey from joining Cisco in the late 90s, where security was still an unproven concept, through to working on cryptography at Apple, digital rights management during the rise of iTunes, and leading product security at Zoom during its explosive growth in the pandemic.But the real focus of the conversation is what’s happening now.Rod explains why: • Cyber and fraud are now converging into one problem • Attackers are targeting humans, not infrastructure • AI is accelerating both the volume and precision of attacks • And why CISOs are being forced to rethink their role entirelyOne of the biggest shifts? Security teams are no longer just protecting the network. They’re now responsible for protecting customers, brand trust, and a rapidly expanding “human attack surface.”If you’re a cybersecurity leader, founder, or operator trying to understand where the industry is heading, this is a must-listen.Rod's LinkedIn - https://www.linkedin.com/in/rodschultz/Bolster AI - https://www.linkedin.com/company/bolster-ai/Joseph's LinkedIn - https://www.linkedin.com/in/cybercoops/Aspiron Search - https://aspironsearch.com/

In this episode, Joseph sits down with Jonathan Kreiner, CTO and Co-Founder of Orion Security, fresh off the back of a $32m Series A and his Forbes 30 Under 30 recognition.In this conversation, we cover: • Why traditional DLP fails, from unscalable policies to 80–97% false positives • Why false positives are a symptom, not the root cause • How Orion is building an AI-native DLP with no policies or regex, using AI agents to judge intent in real time • The analogy between antivirus vs EDR and what it means for data security • AI agents as the next insider threat • Real-world deepfake hiring incidents and layered security in Fortune 500 environmentsJonathan's LinkedIn - https://www.linkedin.com/in/jonathan-kreiner/Orion Security - https://www.linkedin.com/company/orionsec/Joseph's LinkedIn - https://www.linkedin.com/in/cybercoops/Aspiron Search - https://www.linkedin.com/company/aspiron-search/

In this episode, we sit down with Brian Carbaugh, CEO and Co-Founder of Andesite, to unpack one of the most unconventional founder journeys in cybersecurity.Brian spent 25 years at the CIA, serving in special operations, covert action, and ultimately leading the agency’s Special Activities Center. Today, he’s building a well-funded AI-for-SOC company in one of the most crowded and competitive markets in tech.We dig into:• How a career in intelligence shaped Brian’s approach to leadership, risk, and company-building• Why Andesite is taking a contrarian stance on AI in the SOC by putting analysts, not automation, at the centre• What’s broken about today’s “replace the analyst” narrative in cybersecurity• How the AI SOC market has exploded and what actually differentiates vendors beyond hype• Why Andesite chose a company-builder and foundry model over traditional venture funding• The strategic decision to prioritise compliance and trust from day one• The realities of transitioning from national security to startup chaosThis conversation goes deep on human-first AI, trust in cybersecurity, and what it really takes to build an enterprise-grade company in a noisy market.Brian's LinkedIn - https://www.linkedin.com/in/brian-carbaugh-38b339243/Andesite - https://www.linkedin.com/company/andesiteai/Oliver's LinkedIn - https://www.linkedin.com/in/oliverlegg/Aspiron Search - https://www.linkedin.com/company/aspiron-search/

Security teams are drowning in vulnerabilities.Millions of alerts. Tiny teams. Attackers moving faster every year.In this episode, we sit down with Harry, Co-founder and CEO of Maze, to unpack why vulnerability management is fundamentally broken and how AI agents are starting to change the game.Harry shares his journey from leading product at Tessian to building Maze, the moments that convinced him this problem had to be solved, and why most AI security products are more marketing than substance.We cover:Why having a million “critical” vulnerabilities is uselessHow AI agents can identify the handful that actually matterWhat most security vendors get wrong about risk scoringHiring for potential in early-stage startupsWhy UK cyber startups must think US-first from day oneIf you care about AI in security, building high-impact teams, or staying ahead of attackers in a rapidly changing threat landscape, this one’s for you.Harry's LinkedIn - https://www.linkedin.com/in/harrywetherald/Maze - https://www.linkedin.com/company/mazehq/Steffen's LinkedIn - https://www.linkedin.com/in/steffenfoley/Aspiron Search - https://www.linkedin.com/in/steffenfoley/

In this episode of CyberBytes, we sit down with Alex, co-founder and CEO of Arcade.dev, to unpack the real story behind MCP (Model Context Protocol) and why it has become one of the fastest-adopted standards in AI.Alex explains how the industry is moving past “wow” chatbots toward action-taking agents that can actually send the email, update the CRM, and automate real workflows. He also shares the unexpected pivot that led Arcade to its core product. What started as a site reliability agent turned into a breakthrough “tool execution layer” that separates an agent’s brain from its hands, making AI actions deterministic, secure, and production-ready.We dig into why most enterprise agent projects stall before production, the three bottlenecks killing ROI (accuracy, token cost, and security), and how MCP done right is not just “APIs wrapped for LLMs” but intention-based tooling built around what users want to achieve. Plus, Alex drops hard-earned lessons from his previous startup exit to Okta, including why early founders should sell fake demos first and why the best infrastructure products feel like black magic.If you are building agents, investing in the space, or trying to understand what the next 6 to 12 months of AI adoption looks like, this one is for you.Alex's LinkedIn - https://www.linkedin.com/in/alexsalazar/Arcade.dev - https://www.arcade.dev/Steffen's LinkedIn - https://www.linkedin.com/in/steffenfoley/Aspiron Search - https://www.linkedin.com/company/aspiron-search/