Windows
Server 2016 and later, Windows 10 and 11. Runs as a service, installs with one PowerShell line, watches services, the event log and patch state.
What the agent watches on Windows →
Linux
Ubuntu, Debian, CentOS, Proxmox and friends. systemd unit, journal and failed-unit checks, ZFS pools on Proxmox.
What the agent watches on Linux →
macOS
Apple silicon and Intel. LaunchDaemon, one line to install.
What the agent watches on macOS →
OPNsense · pfSense
FreeBSD 13 and 14, amd64. The same agent on the firewall itself: services, the log, disk, memory, the WAN address and every custom check.
What the agent watches on OPNsense →
Teltonika · OpenWrt
RutOS on RUTX and similar ARM routers, OpenWrt 21+. Cellular signal, operator, SIM and data counters in every heartbeat. Survives sysupgrade.
What the agent watches on Teltonika →
VyOS
VyOS 1.3, 1.4 and rolling builds. Installed under /config so it survives an image upgrade, with the config-change time and the booted image on every heartbeat.
What the agent watches on VyOS →
VMware ESXi
Standalone ESXi hosts, no vCenter needed. Watched through any agent at the site over SSH: version, uptime, memory, datastores, virtual machines and maintenance mode, with a check that a named VM is powered on.
What the agent watches on VMware ESXi →
TrueNAS · Proxmox · ZFS
TrueNAS, Proxmox, any Linux or FreeBSD host with zpool. Pools as volumes with health. A degraded pool is a critical alert; a full pool is caught even when df says otherwise.
What the agent watches on TrueNAS →
the same everywhereWhat every platform gets.
The agent
One static binary, about 8 MB, no runtime. Outbound HTTPS only; works behind NAT, CGNAT and 4G. Signed self-update.
The data
Heartbeat every five minutes, events on change, custom checks every minute, network facts, two years of hourly history.
The investigator
BE AI uses the same read-only tools on every platform; the recipes know what a service, a log and a package look like on each.
More on BE Hub BE Hub overview Plans & pricing Service desk & tickets 38-tool MCP server Disk-full forecast guide The same alert, with and without BE AI. Autonomy AI RMM Classic RMM Multi-site Windows estates that need a named person on every change. When your customers will not put the estate in a tool you cannot place. A global product. Put the hub where your policy says. From a full disk at 07:12 to a fix approved at 07:19. Most outages announce themselves weeks ahead. The primary RMM. One rule: AI cannot write. Run BE Hub beside the incumbent. Keep one. Cut the other. The AI can read. Only a person can write. Internal IT that has to control the change, not just see the estate. Security checks Reports, digest and knowledge base Integrations Atera alternative RMM pricing in AUD Essential Eight & M365 posture NinjaOne alternative (AU) Datto & ConnectWise alternative CVE patching with 1-click rollback VMware ESXi & Proxmox ZFS monitoring Teltonika & OPNsense monitoring