This page will help you understand which components are needed for each requirement
Below are the main components of the solution:
Proxy- inspect network traffic
Browser extension- Interspets Browser traffic
Shadow AI detector – Inspects network traffic to pick up any AI usage – both chats and agents
Prompt Guardian - Analysis prompts to identify sensitive info, AI attacks like prompt injection, and usage intent. Returns action based on policies set
MCP Gateway- secures Enterprise systems that are exposing API via MCP
Prompts Such as data privacy and prompt injection attacks.AI Gateway - Controls LLM usage of internal applications, third-party, and coding assistant agents. In addition, the AI Gateway helps with discovering internal agents, monitoring and optimizing costs, and securing API Key access to LLMs.
Endpoint Defender - Controls what Local agents are doing on the endpoint. Controls third-party agents and coding assistants.
Models Guardian- Inspects the level of security of the models developers are downloading to use internally, using on company policies that are based on multiple conditions like
Which component does the customer need
Sedurity Risk/ Requirements | Proxy | Browser extension | Shadow AI detector | Prompt Guardian | MCP Gateway | AI Gateway | Endpoint Defender |
|---|---|---|---|---|---|---|---|
Sensetive data exposure or prompt injection attacks while using public AI services such as ChatGPT in the browser | V | V | |||||
Sensetive data exposure or prompt injection attacks while using public AI services such as ChatGPT in the browser, as well as apps like the Coplit app in Teams / windows | V | V | |||||
My local agents are exposed to public LLM | V | V | |||||
My agents might perform risky activities like changing permissions, executing code, accessing mass data | V | ||||||
Don’t know which AI services my users are using. | V | ||||||
Want to conrol which AI services are used by my users (need to chose proxy or browser extention) For wider coverage, better to use a proxy | V | V (soon) | V | ||||
Agents might do risky operation with my enterprise system via MCP | V |
Please specify:
Do you have applications/ agents that use AI providers (OpenAI/ AWS Bedrock/ Anthropic)? If yes, you need to use the AI Gateway
Do you have an application that exposes MCP for agents to use? If yes, you need the MCP Gateway. Please get help from support to add your app
Do you have Agents in the endpoint (such as •GitHub Copilot •Cursor •Codex •Antigravity •Claude Code •Open Claw •OpenCode) - if so, you need the Endpoint defender
Do your users use public AI services (such as ChatGPT, Claude, Gemini)? If so, you need the prompt Guardian using the Proxy
Do you have a local LLM installed? If so,? you need the AI Gateway
Do you have a pac file in use and by which proxy
Do you have a DLP engine and which one. If so need to configure data classification rule to work with using ICAP.
Which local agents does your company use
Which SaaS agents does your company use (ChatGPT, Claude, CoPilot), And to which systems do they connect?