What an agent does
An agent is a small program installed on each computer. It runs all the time with system rights, reports health and inventory to the RMM over the internet, and carries out scripts and patches when told to.
Because it talks out to the RMM itself, an agent keeps working wherever the device has an internet connection: at the office, at home or on hotel Wi-Fi.
How agentless monitoring works
Agentless monitoring uses a collector: one small server or virtual machine at each customer site. The collector connects to devices with the management protocols they already have:
- Windows: WinRM (PowerShell remoting) and WMI.
- Linux and macOS: SSH.
- Switches, firewalls, printers, UPSs and NAS units: SNMP.
- Hypervisors and clouds: their APIs (vCenter, Hyper-V, Proxmox, AWS, Azure and others).
It needs two things: network reach to each device, and credentials with the right to read what it checks.
Where agentless is stronger
- Nothing to roll out, update or repair on every device; adding a site means placing one collector.
- It covers devices that cannot run an agent at all: switches, firewalls, printers, UPSs, storage and hypervisor hosts.
- Fewer moving parts on endpoints, and nothing to conflict with security software or slow down older machines.
- Onboarding a new customer is mostly discovery, not a deployment project.
Where agents are stronger
- Laptops that rarely sit on the customer's network, since an agent reports from anywhere.
- Sites too small to justify a collector, such as a single home worker.
- Features that need constant presence on the device, such as instant remote control without a VPN.
The security trade-off
An agent is software with system rights on every device, signed and updated by a vendor; if the agent or its vendor is compromised, every device running it is exposed. Agentless monitoring removes that, but concentrates risk in the collector and the credentials it holds.
So with agentless monitoring, protect those credentials: use dedicated accounts with only the rights the checks need, store them encrypted, keep the collector patched and off the internet, and change the passwords when a technician leaves.
How to choose
- List what you manage per customer: servers, desktops, laptops, network gear, hypervisors, cloud.
- Mark which devices spend most of their time on the customer's network.
- If most of the estate is on site, agentless covers it with less work, and it is the only option for the network gear.
- If most of it is roaming laptops, agents (or a light check-in helper) cover them better.
- Many MSPs mix the two: agentless for sites, something on the laptops that leave.
How TechSavve RMM does it
TechSavve RMM is agentless: one collector per customer site monitors and patches Windows, Linux and macOS over WinRM and SSH, network gear over SNMP, and hypervisors and clouds through their APIs. For laptops that leave the network there is an optional check-in helper. See how it works or compare it with agent-based RMMs.