In this Executive Conversations episode, servicePath™ CEO Daniel Kube sits down with Daniel Thanos, the CEO and CTO of Sygen.ai, for a wide‑ranging discussion on agentic AI, cyber‑risk and the future of enterprise software. Thanos recounts his journey from a teenage hacker who soldered PC networks and wrote his first contract at 14, through Silicon Valley start‑ups, cryptography research and roles at GE and Arctic Wolf – all of which shaped his view that AI must be front and center in every boardroom.

🎙️ What You’ll Learn

AI as a double‑edged sword: Thanos describes how malicious actors are already using generative AI to design ransomware, calling out the supply‑chain of cyber crime – initial access brokers, code authors and “customer‑support” call centres that help victims pay ransom. Agentic AI (AI that can reason, plan and act through tools) magnifies these risks; without governance, an autonomous agent can become a “chaos monkey” that corrupts data or abuses privileged access. Why boards need a new AI‑risk framework:

• Traditional security tools can’t detect adversarial prompts or attacks hidden in multimodal content.

• NIST’s AI Risk Management Framework emphasizes building AI systems with trustworthiness in mind; the framework’s new generative AI profile helps organizations identify unique risks and propose risk‑management actions.

• The OWASP Top‑10 for LLM applications warns that prompt injection vulnerabilities occur when user inputs alter an LLM’s behaviour, potentially violating safety protocols.
• Attackers can hide malicious instructions in external files or images, leading to data disclosure or privilege escalation.

Hybrid AI beats monolithic models:

• Thanos advocates neuro‑symbolic systems – blending deterministic rules, knowledge graphs and classical AI with modern deep learning. This hybrid approach offers verifiability and auditability, essential for pricing and quoting.

• Companies should fine‑tune and distill smaller, specialized models, not just run one gigantic model. Large LLM APIs are expensive, opaque and still prone to hallucination.

Practical advice for leaders:

• Governance first: adopt frameworks like NIST AI RMF and MITRE’s ATLAS knowledge base; treat agentic systems as a new attack surface and build agentic SecOps to monitor and control tool‑use.

• Air‑gapped backups: follow classic ransomware hygiene by keeping critical data offline and immutable.

• Keep hiring and up‑skilling: junior engineers aren’t going away; pair them with AI tools and experienced mentors to create “cyborg” development teams.

• Split deterministic and probabilistic logic: use AI for guided selling, rule discovery and summarization, but retain deterministic calculators for mission‑critical pricing/config tasks.

A vision for secure, open AI: Thanos sees a future where open‑source AI flourishes through community‑built, smaller models; AI becomes a net benefit when engineered securely and transparently. He warns that heavy‑handed regulation could stunt innovation; robust security practices will help avoid the backlash that kills open ecosystems.

💬 Stand‑out Quotes

“The malicious AI supply chain moves faster than the defenders.”

“Prompt injection can lead to data disclosure, unsafe outputs and unauthorized tool use.

“Agents will become the primary way people consume software.”

“You still need Iron Man inside the Iron Man suit – humans + AI create advanced intelligence.”

👥 Join the Conversation

How is your organization preparing for agentic AI? Are you implementing frameworks like NIST AI RMF or testing for prompt injection vulnerabilities? Drop your thoughts and questions below. If you’re struggling with complex pricing, configuration or recovering from a failed CPQ implementation, see how servicePath™ CPQ+ (low‑code, CRM‑integrated) simplifies complex sales while keeping your pricing logic deterministic. www.servicepath.co