Earnings Call - VARONIS SYSTEMS INC

AI agents aren't just answering questions anymore. They're calling tools, reaching sensitive data, and leveraging models at a pace that's hard to control. That's why Varonis Atlas is enhancing its AI gateway. The Varoni AI gateway already protects the data AI systems can reach, but now it's expanding to help teams govern the tools agents can call with the new MCP gateway and control model spend and token usage with the new LLM gateway. Three use cases, one gateway. Let's start with governing the tools your agents can call. The AI ecosystem can be complex and hard to visualize and control. But now with the MCP gateway, it's easy to see how AI systems are connecting with tools, APIs, and MCP servers, all in one place. Here's a centralized view showcasing activity, including tool calls, request outcomes, and trends over time. With Atlas, teams can easily publish virtual MCP gateways. These gateways can be scoped to different groups or use cases. This makes it easy to apply unique guardrails and approve specific tools for different uses. Here we see an inventory of some published virtual MCPs. Diving into this marketing vmcp, we can see any connected servers. Let's take a look at this MongoDB server. Here we can see all the available tools to the server and their capabilities. This is where teams can enforce granular controls for each available tool. By navigating into the inventory, teams can get a graphical view of how the server connects to other resources. Gone are the days of complex, hard-to-visualize toolchains. Now let's see how this looks like from the user side. Here's a user in Cloud Code. We can see they're authenticated into Acme MCP Gateway. This is one of the user's approved MCP servers. At the bottom of the Cloud console, you can see the approved tools for the server. This user prompts Cloud Code for the contact details of a lead in Salesforce. The agent successfully calls Salesforce through the controlled gateway under the credentials of the user making the request. The Varonis MCP gateway authenticated the connection, brokered credentials, and evaluated the tool call for approval, with no impact to the user's workflow. With MCP Gateway, teams can confidently govern any AI-to-tool interaction. So, Varonis MCP Gateway lets teams centrally control what their MCP servers and agents can do. But what about model spend and token usage? AI applications, agents, co-pilots, and employees are calling LLMs directly all the time, with little visibility or control over what models they're using and how much it's costing. As part of the Varoni AI Gateway, the LLM Gateway gives teams a single centralized doorway to govern and enforce controls around LL model spend in token usage. here is a central view of all the model and token usage including activity trends like any spikes in usage or total costs speaking of cost let's dive in there the llm gateway provides teams a detailed view of cost breakdowns tied to endpoints identities and specific models teams can easily see cost spikes over time based on this information organizations can enable cost controls here we see a nice overview of various budget policies let's walk through creating a policy of our own we're naming this one marketing policy users can assign the scope of the budget to apply to endpoints users or groups we're selecting the marketing group next we establish a budget amount in this case it's ten dollars and we set the limit to a daily time frame with this new budget in place let's see the llm gateway at work here's a marketing manager interacting with an ai chat tool we can see they've sent a basic greeting without a problem when the user prompts the model for help with writing a blog on the history of Varonis, the gateway blocks the request and notifies the user that the marketing budget policy has reached its limit. Back in Atlas Investigations, the full interaction is available, including the content and users involved. Here we're clicking into the prompt that exceeded the budget for a more detailed breakdown of what policy fired. Now, with the addition of both the MCP and LLM gateway, the Atlas AI Gateway gives you one central control point for AI in action. You get to decide what tools and models agents can use, monitor interactions, and control model spending. Your organization can confidently leverage the power of AI with the governance and protection of Varonis Atlas.