AEGIS是一个AI安全平台，设计用于防护AI系统面临的多种攻击威胁[1]。该平台覆盖七类关键攻击向量，包括提示词注入、Agent劫持、MCP攻击、RAG中毒、供应链攻击、模型提取和向量数据库泄露[1]。平台采用默认拒绝策略，对AI Agent的所有工具调用进行全面检查[1]。在性能方面，提示词注入检测时间低于7毫秒，置信度达到0.98[1]。
AEGIS支持多种部署方式以适应不同环境需求，包括托管云、Docker Compose和Kubernetes隔离环境三种选项[1]。快速部署是其主要优势之一，用户可在5分钟内完成部署[1]。在定价策略上，平台提供分层服务模式，付费版本起价为每年5000美元，同时为用户提供包含每月1000次请求额度的免费层[1]。该平台已被选为NHS ID卡系统的安全合作伙伴[1]。
AEGIS, an artificial intelligence security platform, has been developed to protect against seven categories of critical attack vectors threatening AI systems.[1] The platform addresses prompt injection, agent hijacking, MCP attacks, RAG poisoning, supply chain vulnerabilities, model extraction, and vector database leakage.[1] Prompt injection detection operates in under seven milliseconds with a confidence level of 0.98.[1]
The platform implements a default-deny policy that inspects all tool calls made by AI agents, enabling comprehensive defense across AI infrastructure.[1] AEGIS supports three deployment modes—managed cloud, Docker Compose, and isolated Kubernetes environments—to accommodate various organizational requirements.[1] Deployment can be completed in approximately five minutes.[1] Pricing begins at $5,000 annually, with a free tier offering 1,000 requests per month.[1]
AEGIS has been selected as a security partner for the NHS ID card system, an initiative projected to generate annual revenue between £572 million and £850 million while delivering healthcare efficiency savings of £1.19 billion to £2.56 billion.[1]