Podcast
AI Security, Startup Struggles, and Redefining Red Teaming
SplxAI founders share how they’re redefining AI security with continuous red teaming, workflow visibility, and secure-by-design development practices.
Kristian Kamber
Ante Gojsalić
DATE
Apr 6, 2025
TIME & LENGTH
52 min
STATUS
Available on demand
LANGUAGE
English
In this episode of The Pulse, host Peter James, CEO of AR Security, welcomes Kristian Kamber (CEO & Co-Founder of SplxAI) and Ante Gojsalic (CTO & Co-Founder of SplxAI) for a candid conversation about the state of AI security and their journey building SplxAI.
The discussion spans the early founding story of SplxAI – from a chance meeting to identifying the gap in production-ready security for AI systems – and follows their evolution from a European startup into a leading force in AI red teaming. Kristian and Ante break down the urgent need for transparency and compliance in today’s AI workflows, share actionable advice for other founders navigating the startup journey, and explain why traditional security vendors are unequipped for the risks introduced by generative AI.
From redefining red teaming for agentic systems to tackling RAG poisoning, hallucinations, and integrating red teaming into CI/CD pipelines, this episode offers a deep dive into what it takes to build secure-by-design AI systems and why the industry needs purpose-built security solutions for this new era.
Startup to Scale: Why AI Security Needs a New Playbook
AI Red Teaming must keep up the pace: Traditional testing can’t handle AI’s complexity. SplxAI focuses on continuous, adaptive red teaming for agentic workflows, that goes beyond black-box methods.
Visibility is key: AI security now means securing workflows and applications, not just models. Tools like Agentic Radar help map and monitor agent behavior to prevent data leaks and manipulation.
Secure from the start: SplxAI bakes continuous AI security testing into CI/CD pipelines, emphasizes scanning during key changes, and broadens AI security to include assessments for hallucinations, brand risks, and compliance gaps.
