Automated AI red teaming and runtime protection across the AI application lifecycle

SPLX, previously SplxAI, was founded in 2023 by Kristian Kamber and Ante Gojsalic and operated from Zagreb and New York. It raised roughly $9M, including a $7M seed round led by LAUNCHub Ventures in March 2025 with Inovo Venture Partners, SCV, DNV, Runtime Ventures and Rain Capital participating, and was acquired by Zscaler in November 2025, where the technology extends the Zero Trust Exchange with AI asset discovery, red teaming and governance. The product covers the AI lifecycle rather than a single stage, which is what distinguishes it from tools that only scan a model or only filter traffic.

The offensive half runs automated adversarial testing against models, agentic workflows, MCP servers and existing guardrails, using both standard and customer-specific attack datasets to probe for prompt injection, hallucination, data leakage and unsafe tool use. Results come back as vulnerability findings paired with concrete remediation, including hardened system prompts, and as benchmarks that let teams compare model behaviour before committing to one. The defensive half discovers AI assets across the enterprise, publishes an AI bill of materials, filters inputs and outputs in production, and maps controls against regulatory frameworks so that compliance evidence is produced continuously. SPLX also maintains Agentic Radar, an open-source scanner for agentic AI systems.