Endpoint Central on the server manages nothing by itself. Every patch, remote-control session, inventory scan and software deployment reaches a PC through one small program — the agent. This guide installs that agent properly: the download-and-install route shown in the video, and every serious rollout route after it.
In the video, Anshu follows the route every Endpoint Central setup starts with: log in to the Endpoint Central console, open the Agent tab, go to Computers, click Download Agent for the office you are setting up, and install that package on the client machine. He also points out that the console offers agent downloads for Mac and Linux, not just Windows. That manual route is perfect for the first few machines and for testing. An office of fifty PCs needs more — so this guide continues with the console push install, the three GPO flavours, SCCM and OS imaging, and then the parts most setup videos skip: verifying the agent is truly alive, removing it when it is protected, and understanding what this small program can actually do on a PC.
Open the Endpoint Central console and sign in as an administrator. Go to Agent → Computers, select the Remote Office the PC belongs to, and click Download Agent. Copy the downloaded package to the client PC — pen drive, shared folder or any way you like — and run it as an administrator. The installer is quiet and quick; when it finishes, the agent service starts and the PC checks in with the server. Within a short while the computer appears in the console under Agent → Computers (the Scope of Management list). That check-in is the real finish line, not the installer's last screen.
Two honest notes. First, the package is built for your server and office — it already knows where to report, which is why downloads are separated per remote office. Second, download a fresh package when you need one instead of reusing a months-old copy from a downloads folder; an old package can carry an old agent version that then has to update itself anyway.
The doctrine of this whole product is one line: no agent, no management. The Endpoint Central server cannot patch a PC, take a remote session, read its inventory or push software to it until the agent is installed and reporting. The agent is what receives the server's instructions — patch scans and deployments, configurations, remote-control requests — and reports back results, hardware and software data, and status. So an agent rollout is never a side task you finish later; it is the project. A server with agents on 30 of 50 PCs is managing 30 PCs, and the other 20 are simply not in the system, whatever the licence count says.
For PCs already visible on the network, the console can push the agent itself: from Agent → Computers you add the target computers and start the installation, and the server installs the agent remotely. ManageEngine lists this automatic console installation among its standard methods, and it is the fastest way to cover a known list of machines in one office. The honest requirements: the target PC must be switched on and reachable, you need administrative credentials for it that the console can use, and the office firewall rules must not block the server from reaching it. Machines that are off, off-network or credential-blocked simply fail — note them and finish them with another route instead of retrying forever.
On an Active Directory domain, Group Policy is the workhorse, and Endpoint Central supports three flavours. The GPO lightweight tool is the simplest: it builds the policy for you. The GPO scheduler is the fastest, installing the agent in the next policy refresh cycle through a scheduled task — ManageEngine's own docs use the agent file renamed to LocalOffice_Agent.exe with a PowerShell script placed in the SYSVOL scripts folder. The classic startup script is the most common method of all, but it honestly costs a restart before anything happens, and a laptop that restarts at home before its VPN connects can miss the startup script entirely. Pick per office: scheduler for speed, startup script for simplicity, lightweight tool when you want the fewest manual steps.
If your office already runs SCCM (Microsoft Configuration Manager), Endpoint Central provides an SCCM route: download the agent zip from the SCCM tab under Agent Installation, place the executable in a network share every target can read, and deploy it as an SCCM package — the silent command documented for the package is LocalOffice_Agent.exe /silent. The agent can also ride inside your OS deployment images, so a freshly imaged PC arrives already managed, and ManageEngine documents automatic installation through the Scope of Management policy, which installs the agent on new machines when they are added to Active Directory. None of these is 'the best method' in the abstract — the best method is the one your office's infrastructure already supports.
As the video mentions while downloading, Endpoint Central also ships agents for Mac and Linux — the Linux agent arrives as a .bin installer package rather than an .exe. The management idea is identical: small agent, installed once, reporting to the same server and appearing in the same Computers list. The honest differences are in coverage and handling — a Linux server fleet and a Mac design team are managed through the same console as the Windows PCs, but each platform's agent package is downloaded separately for its remote office, and Linux installs are typically run from the terminal with administrator rights. Do not let the Windows majority make you forget these: an unmanaged Mac is exactly as unpatched as an unmanaged PC.
An installer that 'finished' is not proof of a managed PC. The proof is in the console: the computer listed under Scope of Management with a current last-contact time and an agent version. Give a fresh install a little time — agents work on a refresh cycle (about 90 minutes by default) for many settings, though the first check-in normally shows much sooner. If a machine never appears, work the failure ladder in the companion case guide below instead of reinstalling blindly. And once your first machines report in, turn on the agent protection settings described in the uninstall guide, so the agent you just installed cannot be casually removed by the PC's user.
The console route from this guide — Agent tab, Computers, Download Agent, and the install on a real client PC — shown live, step by step.
▶️ Watch Full VideoNo. The server alone manages nothing until the agent is installed on each PC and the PC checks in. No agent, no management — patching, remote control, inventory and deployment all travel through the agent.
The one for the remote office the PC belongs to, from Agent, then Computers, then Download Agent in your console. Each package carries your server's address for that office, so packages are not interchangeable between offices.
No. The agent installs as a system service, and silent installation is supported (for example LocalOffice_Agent.exe /silent). The user may see the agent icon in the system tray afterwards — in an office you manage, it is good practice that staff know the management agent is there and why.