SPLX, a Zscaler Company

What SPLX, a Zscaler Company does

SPLX, a Zscaler Company, provides an end-to-end AI security platform for Fortune 500 companies and global enterprises. The platform secures LLM-powered systems across the entire lifecycle, from development to deployment. It combines automated AI red teaming, real-time threat detection and response, and compliance mapping to uncover vulnerabilities, block live threats, and enforce AI policies at scale. Built by AI security experts and red teamers, SPLX empowers security, engineering, and risk teams to confidently adopt LLMs, chatbots, and agents, protecting against prompt injection, jailbreaks, data leakage, off-topic responses, privilege escalation, and evolving threats. The platform offers full-stack security tooling for building, deploying, and scaling AI systems safely, including AI asset management, automated AI red teaming, AI runtime protection, AI governance and compliance, dynamic remediation, AI runtime threat inspection, and AI model security.

Products

• SPLX PlatformA unified platform to test, protect, and govern AI at scale, purpose-built to secure every stage of the AI lifecycle from red teaming and runtime security to governance and remediation.
• AI Asset ManagementAutomatically discovers every model, AI workflow, MCP server, and guardrail across an enterprise, building a complete AI-BOM, uncovering vulnerabilities, and ensuring compliance.
• Automated AI Red TeamingContinuously uncovers real vulnerabilities in AI systems using automated red teaming, powered by an extensive AI attack database and custom datasets. Includes 25+ predefined probes and customizable probes.
• AI Runtime ProtectionEnforces input/output guardrails during live deployments to detect malicious behavior, prevent prompt injections, and stop sensitive data leakage in real time with near-zero latency.
• AI Governance & ComplianceAutomatically maps AI systems to global and custom security standards, ensuring alignment with regulatory frameworks and internal policies at scale.
• Dynamic RemediationTurns red teaming insights into action with dynamic system prompt hardening and remediation steps, directly minimizing the attack surface by up to 95%.
• AI Runtime Threat InspectionDetects jailbreaks, prompt injections, and malicious queries in near real time by scanning LLM logs, helping to remediate threats before they cause harm.
• AI Model SecurityStress-tests commercial and open-source LLMs to expose weaknesses and helps in selecting the most secure, use case-aligned models with confidence.
• Agentic RadarThe first open-source security scanner for agentic AI, visualizing workflows, surfacing attack patterns and tool-level vulnerabilities, and providing hardening recommendations for each agent.

Key capabilities

• AI Asset Management
• Automated AI Red Teaming
• AI Runtime Protection
• AI Governance & Compliance
• Dynamic Remediation
• AI Runtime Threat Inspection
• AI Model Security
• Open-Source SAST tool for agentic workflows (Agentic Radar)
• Real-time threat detection & response
• Compliance mapping
• Continuous Red-Teaming
• GenAI Risk Mitigation
• GenAI Guardrails
• Regulatory Compliance
• On-Topic Moderation
• AI Safety
• AI Risk
• GenAI Application Security
• Pentesting
• Chatbot Security
• Prompt Injection prevention
• Hallucination detection
• Multi-Modal Prompt Injection protection
• Security Framework Mapping
• Automated insights and audit-ready reporting
• Unified platform for full lifecycle AI security
• Multi-Modal Support (voice and image)
• Target Connection via Platforms (Slack, WhatsApp, Microsoft Teams)
• Remediation Steps
• System Prompt Hardening
• Organization Management
• 99.9% Uptime SLA
• Regional Data Center (US or EU)
• Compliance Framework Check (MITRE ATLAS, OWASP LLM Top 10)
• Continuous Threat Monitoring
• On-Prem Deployment
• Hybrid / VPC Deployment
• SSO / SAML
• Advanced Workflow Integrations (Jira, ServiceNow)
• Platform Training & Onboarding
• Premium Support
• Probe Credits
• Custom Probes
• Multi-Language Support
• Advanced Visualization
• PDF Reports
• Tailored Remediation Steps
• Organization with Multiple Users
• Target Connection via LLMs
• Target Connection via REST API
• CI/CD Integration
• Public API
• Cloud Deployment
• Community Support
• Customer Success Program
• Technical Account Manager
• Full Data Privacy
• LLM discovery & benchmarks
• AI workflow scanning
• Agent-level risk analysis
• MCP server detection
• AI compliance readiness
• Enterprise-wide visibility
• AI model discovery
• AI workflow analysis
• MCP Server scanning
• Input & Output Filtering
• Custom Policy Enforcement
• Attack Telemetry
• Inline AI Security

Use cases

• Securing LLM-powered systems across the entire lifecycle
• Uncovering vulnerabilities in AI systems
• Blocking live threats in AI systems
• Enforcing AI policies at scale
• Adopting LLMs, chatbots, and agents with confidence
• Protecting against prompt injection
• Protecting against jailbreaks
• Protecting against data leakage
• Protecting against off-topic responses
• Protecting against privilege escalation
• Protecting against evolving AI threats
• Deploying internal copilots securely
• Deploying external-facing assistants securely
• Gaining visibility into AI risks
• Controlling AI risks
• Automating AI risk management
• Testing AI applications with advanced features
• Scaling AI deployments
• Assessing AI apps with advanced features
• Deploying trusted AI systems at scale
• Managing AI security posture
• Continuous risk analysis
• Compliance reporting
• Security posture management
• Identifying unsafe models
• Visualizing AI workflows
• Surfacing attack patterns and tool-level vulnerabilities in agentic AI
• Providing hardening recommendations for agentic AI
• Accelerating secure AI deployments
• Reducing security overhead for AI
• Preventing high-impact AI incidents proactively and in real-time
• Running scalable, continuous testing for AI workflows
• Monitoring entire LLM stack (prompts, agents, runtime behavior)
• Tracking AI security standards with automated insights
• Centralizing AI security operations
• Assessing AI apps with advanced features
• Deploying trusted AI systems at scale
• Securing conversational AI
• Stress-testing AI systems for security & safety risks
• Exposing weaknesses in AI systems before they cause harm
• Aligning AI applications with business objectives
• Staying ahead of emerging AI threats
• Running automated AI risk assessments
• Tracking AI systems' risk exposure
• Testing malicious prompts from adversarial and regular user personas
• Evaluating AI filter mechanisms
• Refining AI filter mechanisms
• Integrating AI security into CI/CD pipelines
• Protecting against Prompt Injections, Off-Topic, Hallucinations, and Social Engineering
• Simulating domain-specific attacks
• Testing with multiple input methods (text, voice, images, documents)
• Tracking and fixing uncovered vulnerabilities
• Automated policy mapping for AI security frameworks
• Monitoring and preventing prompt-based threats in production
• Preventing sensitive data exposure in production
• Preventing unsafe outputs in production
• Enforcing real-time guardrails on inputs and outputs
• Blocking jailbreaks, data leaks, and off-topic behavior in real-time
• Creating custom AI policies
• Defining off-topic rules and risk boundaries
• Analyzing policy violations and emerging attack patterns
• Fine-tuning filter sensitivity and feedback loops
• Reviewing inputs & outputs tied to policy violations
• Tracing detection logic with similarity scores

AI approach

SPLX provides an AI security platform that uses automated red teaming, real-time threat detection, and compliance mapping to secure LLM-powered systems. They build their own tools and use a comprehensive AI attack database, continuously evaluating leading commercial and open-source LLMs.

Tech named: LLM Security, Continuous Red-Teaming, GenAI Risk Mitigation, GenAI Guardrails, Regulatory Compliance, On-Topic Moderation, AI chatbots, Conversational AI, AI Safety, AI Risk, GenAI Application Security, Pentesting, Chatbot Security, Large Language Models, Prompt Injection, Hallucination, Multi-Modal Prompt Injection, Security Framework Mapping, AI red teaming, runtime security, AI Asset Management, Automated AI Red Teaming, AI Runtime Protection, AI Governance & Compliance, Dynamic Remediation, AI Runtime Threat Inspection, AI Model Security, Agentic Radar (open-source SAST tool for agentic AI), Probes, Generative AI, LLM CTFs, AI Benchmarks, AI workflow scanning, Agent-level risk analysis, MCP server detection, Input & Output Filtering, Custom Policy Enforcement, Attack Telemetry, Inline AI Security

Industries served

• Computer and Network Security
• IT & telecommunications
• Highly regulated industries

What it says sets it apart

• Unified platform to test, protect & govern AI at scale
• Built by world-class AI red teamers for enterprise scalability
• Purpose-built to secure every stage of the AI lifecycle
• World's most extensive AI attack database
• Dynamic remediation minimizes attack surface by up to 95%
• First-ever SAST tool for agentic workflows (Agentic Radar) which is free and open-source
• Accelerates AI deployments and reduces security overhead
• Provides real-time AI risk surface visibility
• Streamlined compliance & policy alignment with automated insights
• Centralizes AI security operations in one purpose-built platform
• Unmatched red teaming coverage with 25+ prebuilt probes
• Customizable probes for specific risk scenarios
• Ability to upload custom datasets for targeted evaluations
• Multi-modal testing (text, voice, images, documents)
• Dynamic system prompt hardening
• Near-zero latency for runtime protection
• Ability to define custom AI policies through natural language
• Access to full telemetry for flagged prompts and responses
• Adherence to highest standards of security and trust (ISO 27001 Certified, SOC 2 Compliant, GDPR, CCPA)
• Invent what doesn’t exist (yet) - constantly pushing the frontier of security for AI
• Push performance boundaries - engineered to be faster, smarter, and more adaptive
• Build to have a real impact - mission to make AI safer for everyone