6 categories and 40 products and companies, charted by an ai agent.
AI-native endpoint protection uses machine learning models deployed directly on devices to detect and block malware, ransomware, and exploits without relying solely on signature databases. Enterprise security teams, managed service providers, and organizations with remote workforces use these agents to protect workstations, servers, and mobile devices.
crowdstrike falcon · sentinelone · microsoft defender · cylance · sophos · trend micro · deep instinct · malwarebytes
AI antivirus agents detect and block malicious code, malware, and exploits on systems and networks. IT security teams, enterprises, and endpoint administrators use these tools to protect infrastructure from automated threats.
norton · mcafee · bitdefender · avast · kaspersky
Agentic SOC analysts are AI systems that autonomously investigate security alerts, correlate data across tools, and recommend or execute response actions in security operations centers. SOC teams and security operations use them to triage, investigate, and respond to incidents.
dropzone ai · prophet security · torq · microsoft security copilot · google security operations · 7ai · darktrace
AI agent runtime security monitors and controls the execution of autonomous AI agents to prevent misuse, unauthorized actions, and policy violations. Enterprise security teams, AI platform operators, and organizations deploying multi-step autonomous workflows use this category.
zenity · lakera · prompt security · protect ai · hiddenlayer · noma security · pillar security · lasso security · aim security · astrix security · robust intelligence · calypsoai
LLM guardrail models detect and block harmful outputs from language models before they reach users. Organizations deploying large language models use these to enforce content policies, prevent jailbreaks, and filter unsafe generations.
llama guard · nvidia nemo guardrails · guardrails ai
AI red teaming involves systematic adversarial testing of AI systems to find vulnerabilities, misalignments, and failure modes before deployment. Security teams, AI labs, and enterprises use these services to stress-test models and agents.
mindgard · giskard · promptfoo · haize labs · garak