The desktop guardrail for AI work

AI moved out of the browser. Your policy follows it

ChatGPT, Claude, and Copilot now run as desktop apps, outside the browser. dME Guard runs on Mac and Windows, watches every app including the AI ones, enforces the same rules you set in dME, and records every decision as evidence.

Available · Mac · Windows

The shift

AI left the browser. A browser policy cannot see it.

For years the browser was where work and AI met, so a work browser was enough. That changed. AI tools now ship as desktop apps, and people paste client data into a window no browser control can reach.

01

AI is a desktop app now

The fastest-growing AI tools run as native apps on Mac and Windows. A browser policy never sees them.

02

Banning it backfires

Block an AI app and people switch to a personal account on their phone. The work still happens, just without a record.

03

You need one boundary

The same rules should hold whether data is in a browser tab or a desktop app. Two policies leave a gap.

What Guard does

Watch. Decide. Log.

dME Guard lives at the desktop layer, between every action and the work itself, AI apps included. Same policy, one evidence trail, across the whole machine.

01 · Watch

Sees every app, including the AI ones

It watches the desktop, not just the browser. The ChatGPT app, Claude, and Copilot, plus Slack, Teams, Notion, and Figma. Copy, paste, uploads, and screenshots, wherever your team works.

02 · Decide

Enforces one policy, OS-wide

The rules you wrote in dME Browser apply across the desktop. Personal AI versus work AI, what data can leave, which app can receive it. Decided the same way in a browser tab or a native app.

03 · Log

Turns every decision into evidence

Each action becomes a clear record: what moved, which app, which account, allowed or blocked, and why. It streams into the same evidence trail as dME Browser. Nothing hidden, nothing guessed.

Evidence

Every AI decision leaves a record.

A block is only useful if you can show what was blocked, when, and why. dME Guard writes one plain line for every decision it makes across the desktop, so a question about an AI tool has an answer instead of a guess.

Every record shows

  • What was about to move, and which app or AI tool it was headed to
  • Which account was in use, personal or work
  • The decision, allowed or blocked, with the rule that applied

It streams where your records already live: Datadog, Splunk, Sentinel, or your own S3. One pipeline for the browser and the desktop.

dME EvidenceLive activity
Chrome · Browser open · Google ChromeBlocked
ChatGPT · Prompt · OpenAI API keyRedacted
ChatGPT · Copy · Acme contract, personal accountBlocked
ChatGPT · Copy · Acme contract, work accountAllowed
Datadog · Splunk · Sentinel · S3Export

What ships in v1

Three things you can use today.

dME Guard ships on Mac and Windows with the same policy engine as dME Browser. Three capabilities are live in the first release.

01

Mac and Windows agent

Native install on macOS 13+ and Windows 10+. A single binary, deployed through the MDM you already run (Jamf, Intune, Kandji). No new agent infrastructure.

02

One policy across browser and desktop

Write your data rules once in dME. Browser enforces them in the browser, Guard enforces them across desktop apps and AI tools. One set of rules, one audit.

03

One evidence stream

Every Guard decision lands in the same destination as dME Browser. Datadog, Splunk, Sentinel, or raw S3. One pipeline, two surfaces.

Pricing

Ships standalone or with dME Browser.

dME Guard runs on Mac and Windows with the same install, policy, and evidence stream as dME Browser. Buy them separately or together. The boundary holds either way.

Good fit if

  • Your team uses AI desktop apps like ChatGPT, Claude, or Copilot
  • You need the same data policy across the browser and the desktop
  • You are already running dME Browser or evaluating it

dME Guard is the desktop side of dME. dME Browser is the browser side. Together they hold one policy everywhere work and AI happen.

See dME Browser