What LLMs does your organisation use?

Help AshLab understand which language models are in use. This builds your LLM inventory — the foundation of your AI posture and security deep dive. You can provide a repository URL (auto-refreshed daily) or simply paste model names manually.

Every interaction is logged with the exact model name — your inventory lets AshLab flag when agents use undeclared models, a key security signal.

Do you use any known MCP servers?

Model Context Protocol (MCP) servers extend your AI agents with tools, knowledge, and external connections (file system, browser, databases, APIs). Declaring known MCPs gives AshLab the context to detect unknown or unauthorised MCP connections — a key security posture signal.

Security Deep Dive: AshLab captures mcp_server_id, tool_calls, and connection_id for every MCP interaction. Undeclared MCPs appear in Vendor Management → Discovered Items and can be individually alerted or blocked — without blocking the underlying LLM.

You can also add custom MCPs — AshLab will discover additional ones automatically from network traffic and agent telemetry.

Do you use any known AI connectors?

AI connectors bridge your AI agents to business systems — SharePoint, Dynamics, SAP, Salesforce, databases, and APIs. Knowing which connectors are sanctioned helps AshLab flag when agents access unexpected data sources.

Security Deep Dive: AshLab discovers connectors from Copilot Studio agent definitions (Dataverse) and live telemetry metadata. Connectors not in this declared list are flagged as undeclared data access — a key signal for data exfiltration detection in the coming Security Deep Dive tab.

Anything agents are allowed to read from / write to — files, APIs, databases, email — should be listed here.