By David,
A ticket comes in: "My desktop is really slow this morning."
The traditional response is a round of questions. Which machine? What were you doing? Can you open Task Manager and read me what's at the top? Then a remote-control tool, a second login, and a lot of waiting while the user describes their screen.
With Atlas Reach, every enrolled Windows device has its own page that already knows the answers. This walkthrough follows one machine, avd-design-0, a Windows 11 multi-session host in a GPU design pool, through the four views a technician reaches for first: Performance, Software, Processes and Services. Then we click Connect and take over the desktop.
About the screenshots: every image is the real Atlas AVD console running in demo mode with fictional data. Demo mode has no real machine to stream, so the desktop in the final screenshot is a simulated Windows desktop played into the real Atlas Reach console.
Opening the device
From Atlas Reach, click any device in the fleet list. The device page opens with everything you can do to the machine in its header (Terminal, Toolbox, Run script, Quick actions, Export .pdf, Ask Advisor and Connect) and every view of the machine down the left: Overview, Performance, Inventory, Software, Device Manager, Scheduled Tasks, Services, Updates, Security, Registry, Event Viewer and more.
The header also answers the first question before you ask it: the device is Online, its security posture is OK, and its agent checked in just now. You're looking at the machine as it is, not as it was at the last nightly sync.
Why it matters: one page per device means one place to start every ticket. No tool-switching, no "which console was that in?"
Performance: is it the machine, or the network?
The Performance tab opens with a verdict, not a wall of charts. Device health: Healthy, all counters within thresholds, with a status chip for each area that can make a desktop feel slow: CPU, Memory, Storage, Network and User experience.
Below the verdict, Key metrics lays out CPU, memory, free disk, round-trip time, active sessions and frame rate, each with its own trend line and average and peak for the window. Pick 1h, 6h, 24h, 7d, 30d or a custom range, leave Auto refresh on to follow the device live, switch between Charts and Table, or Export the samples.
For our slow-desktop ticket this view already narrows things down. CPU and memory are healthy, the round-trip time is 24 ms and the session is drawing 27 frames a second. The machine isn't overloaded and the connection is fine, so the problem is more likely something on the machine: an application, a process or a service.
Why it matters: you rule out whole categories of cause in seconds, with numbers you can paste into the ticket, instead of asking the user how slow "slow" is.
Software: what's installed, and what's out of date
The Software tab is the device's installed-software inventory, kept current by the agent and refreshable on demand with Refresh from device.
Four tiles summarise the machine: 18 installed applications, 16 of them managed through winget, 2 updates available and 2 security issues. Every row shows the publisher, version, install date and source, plus a status that tells you whether the app is current, recently updated, has an update waiting, is excluded from automatic updates, or isn't managed at all.
Here, 7-Zip has an update available, and Adobe Acrobat Reader's last automatic update failed. That's worth fixing whether or not it's the cause of today's ticket, and you don't have to leave the page to do it: an app's Upgrade to action runs winget on the device right now, and Deploy software installs anything that's missing. The Winget History and Software changes tabs show what was installed, updated or removed and when, which is often the fastest answer to "it was fine yesterday."
Why it matters: you can see whether a device is patched without remoting in, and fix it from the same place you found the problem.
Processes: Task Manager, without interrupting anyone
Open Toolbox from the device header and the first tab is Processes: a live task manager for the remote machine.
The list is read straight off the machine and sorted by memory, so the heaviest processes rise to the top. Each row shows its PID, its session, its share of CPU and its memory, and has an end-task button. On a multi-session host the session column matters: it tells you which user's session each process runs in, so you can stop one user's runaway process without touching anyone else's.
On avd-design-0, Photoshop in session 2 and Blender in session 3 are the big consumers. That's expected on a design host, but it's exactly the kind of fact you'd otherwise have to ask the user to read out.
Crucially, none of this needs a screen session. The user keeps working while you look. As the toolbox itself says, actions run as the agent's service account and are recorded in the audit log, so ending a process is never an untracked event.
Why it matters: you diagnose the busy machine without taking it over, and without asking the user to stop what they're doing.
Services: the Services console, for every device
Back on the device page, the Services tab is a full-page version of the Windows Services console for the remote machine.
Every service is listed with its status, startup type (including Delayed Start and Trigger Start) and the account it logs on as. The tiles across the top count Running, Stopped, Transitioning or paused, Automatic, Manual and Disabled services, and each tile is also a one-click filter. Search by name, narrow by startup type, then Start, Stop or Restart a service from its row, or select several and act on them together. Export gives you the list for documentation or a change record.
On our device, one service is transitioning: Windows Search is stuck starting. A stuck indexer is a classic cause of a sluggish desktop and a slow Start menu, and a Restart from this row is a reasonable first fix. Like the toolbox, the list is read live from the machine, each action is audited, and only technicians with control permission can use it.
Why it matters: the most common "turn it off and on again" fixes take one click, on any device, with a record of who did it.
Connect: one click to the desktop
Sometimes you need to see the screen. Click Connect in the device header and the Atlas Reach console opens in its own browser tab, live on the user's desktop.
There's no VPN to start, no port to open and nothing for the user to install or read out: the agent is already on the machine. The console header shows that you're Live, the link quality and frame rate, and the host's own CPU, memory, disk queue and network while you work, so you can tell a slow machine from a slow connection without leaving the session.
The toolbar covers the rest of a support call:
- Paste to device and Copy from device for the clipboard, and Files to browse the device and transfer files (or just drop a file onto the picture)
- Processes, the same live process list, as a panel beside the picture
- Send Ctrl+Alt+Del, and Keys for the shortcuts a browser can't forward (Win+L, Alt+Tab, Alt+F4), plus an on-screen Keyboard for tablets
- Reboot & reconnect, which restarts the machine and reconnects your session automatically
- Privacy screen, Block local input and Lock on end for work the user shouldn't see or interrupt
- Listen for the device's audio, Full screen, Fit to window, and a quality setting from Auto to Always best quality
- Session chat and drawing tools, so you can talk the user through it and point at what they should click
On the device itself, the user always knows. A red banner across the top of their screen reads "operator is controlling this screen", with a Stop sharing button that ends the session from their side. Your policy decides whether a technician needs the user's consent before connecting (Consent required), simply tells them (Notify only), or connects to unattended machines (Silent). On a multi-user host where someone is also signed in over Remote Desktop, Atlas asks which session you want, and you can switch between them after connecting.
Why it matters: when you do need the screen, you're on it in one click, with the tools for the whole job in one window and a clear, visible session for the user.
The whole ticket, on one page
Back to the ticket. In a few minutes, from one device page, you:
- Checked Performance and ruled out an overloaded host or a bad connection.
- Checked Software and found an app whose last update failed.
- Checked Processes and saw who was using what, without interrupting anyone.
- Checked Services, found Windows Search stuck starting, and restarted it.
- Clicked Connect to confirm with the user, on their own screen, that the desktop is responsive again.
No VPN, no second tool, no reading the screen down the phone, and an audit trail for every action you took.
That's the idea behind Atlas Reach: everything about a Windows device on one page, one click from its desktop. See what else it does on the Atlas Reach page, or read how Winget Deployment keeps the software tab green across your whole fleet. If you're comparing RMMs, how to choose an RMM as an Azure-first MSP covers the criteria and the alternatives.