Back Issues/Search Home → Calendar → Archive → RSS → Subscribe → Services → Vendors → Blog → Current Issue → Popular → Newsletters (LinkedIn, Substack)

All issues › Volume 344, Issue 2 › IT Vendor News › Varonis

AI Gateway Security: How to Control Models, Tokens, and Tool Calls

Varonis, Wednesday, November 11th, 2026

Varonis explains what a complete AI gateway should control: prompt and response guardrails, model and token spend, and MCP tool access.

Varonis says organizations struggle to know what tool calls AI is making and what it costs, citing a Cloudera survey in which 89% of enterprises are limiting or redesigning AI use cases due to rising token consumption.

Most AI gateways focus on routing, authentication, rate limiting and logging, leaving security, cost and tool control as separate problems.

A complete gateway should inspect prompts and responses against policy, decide which models can be used and cap their cost, and govern which tools agents can access.

Varonis says its Atlas LLM Gateway spans 95+ providers and 3,000+ models, and its Atlas MCP Gateway ties every agent tool call to a real identity, and the post outlines six steps toward a complete gateway, starting with discovery.

more →  ·  More from Varonis →