Shadow AI Discovery

Shows which AI tools your devices are actually reaching, including browser-based services and local LLMs, and in which departments and how often.

The Shadow AI Discovery skill uses AI-detection telemetry on Lansweeper inventory, joined for owner and department context. It normalizes the detected AI tools and services, then shows which devices and departments touch them, how often and over what protocol. It catches web-based AI and locally run models that leave no install footprint, and it can produce an AI-usage register for governance and acceptable-use policy.

image 223

Why the Shadow AI Discovery skill Matters Here

  • IT perspective

    “Which AI tools should we standardize on, and who is using something else?”

    • Usage by department and frequency shows what people actually adopt.

    • Devices using a tool other than the standardized one, such as Copilot, are visible per device.

  • Security perspective

    “Is company data flowing into AI tools we have not sanctioned?”

    • Public AI services and local LLMs are attributed to a device and an owner.

    • Devices sending AI traffic over unencrypted HTTP are identified.

Beta

Lens Workbench

Ask your asset data a question. Get a straight answer.

Lens Workbench is a dedicated space where you can ask questions about your asset data and get straight answers, build live reports that stay current, connect your existing tools through MCP connectors, and take action directly, like adding or retiring an asset.

Sign up

Requirements

Lansweeper MCP

https://mcp.lansweeper.com/mcp

AI Governance (Beta)

Lansweeper plan

Starter, Pro, Enterprise

LLM

Claude 3.5+, ChatGPT, Gemini (Antigravity), Copilot

Step by step setup guide

Step 1: Join the AI Governance beta
This skill needs AI Governance (Beta) enabled on your Lansweeper site. Sign up at Lansweeper Beta Program.

Step 2: Download
Download the skill package for your platform. Lens Workbench users can skip this step.

Step 3: Install

  • Claude: install as a skill in Claude Desktop or Claude Code.
  • ChatGPT: add the skill to your ChatGPT (Codex) skills folder.
  • Gemini (Antigravity): load the skill into your Antigravity agent’s skill folder.
  • Copilot: add it as a skill in GitHub Copilot, or import it into Copilot Studio.
  • Lens Workbench: request access to the closed beta at Lansweeper Beta Program. Once you have access, the skill is already in the workbench.

Step 4: Connect Lansweeper
The skill needs your Lansweeper data to return results.

  • Claude: connect the Lansweeper MCP connector and sign in.
  • ChatGPT: connect Lansweeper MCP, EU or US endpoint based on your tenant.
  • Gemini (Antigravity): connect a Lansweeper MCP connector separately; Gemini doesn’t ship one built in.
  • Copilot: connect the Lansweeper MCP connector to your agent.
  • Lens Workbench: select the site(s) you want to run the skill on.

Step 5: Ask

  • Claude: “Which of our machines are using ChatGPT or other public AI tools?” or “Is anyone running local LLMs such as Ollama or LM Studio?”
  • ChatGPT: trigger with $shadow-ai-discovery or “Is anyone running local LLMs such as Ollama or LM Studio?”
  • Gemini (Antigravity): “Which departments use AI the most, and how often?”
  • Copilot: “Give me an AI-usage register for our acceptable-use policy.”
  • Lens Workbench: type /shadow-ai-discovery in the prompt box.
Related Skills
See all skills