ISP and WISP operations without scaling manual troubleshooting
An ISP does not look after one router. It looks after a ticket queue, and most tickets ask for the same few checks. Is the customer reachable, where does the path break, is the link fast enough, what changed on the device. Each check is quick. Repeated across a day and across a distributed network, they are what keeps a small operations team busy.
Take the repeated checks off the queue
ARIA runs those checks from a plain-language request. Reachability and path are tested from the device itself, and loss, latency and the hop-by-hop path come back interpreted, not dumped. A full health baseline combines ping, path and speed into one assessment, so every ticket starts from the same picture.
First-line staff get an answer they can act on, or escalate with the evidence attached. RouterOS expertise that used to sit with a few people reaches the whole queue.
One view of the fleet
The fleet view covers every customer and device, with live status and a diagnostic history for each device.
A technician who inherits a customer's setup can paste the configuration into ARIA and have it explained line by line, with the risks flagged. The question gets asked before the change, not after the outage.
Provisioning through vetted templates
ARIA keeps a catalogue of vetted, typed templates for configuration changes: firewall, DHCP, VLAN, DNS, VPN, QoS, routing and system. The model chooses the template and fills in the values. Free-form writes are possible too, inside the permitted namespaces; a template-only mode exists but is optional and off by default. Either way, a person approves the exact change before anything runs.
That is what keeps provisioning consistent when many hands do it. When a template is used, the structure of the change comes from the template and only the values vary. When no template fits a request, the gap is logged for an administrator, so the catalogue grows from real work.
Why a large fleet is not a larger risk
Scale is where a mistake gets expensive, so the limits sit in code and not in a prompt. Every command is classified as a read or a write by a classifier validated on more than 1,600 real commands. Free-form writes stay inside 16 permitted namespaces, and 10 are never writable.
A session makes no more than 15 writes. Past that, the engineer opens a new chat. Each approval is single use and expires if nobody acts on it, and an audit record is written before the device is touched. The model can ask. Only code decides.
Priced per user
ARIA costs per user per month. Base is EUR 49 and covers inventory, reads, chat and suggestions, with no live writes. Pro is EUR 99 and adds templates and approved writes. Pro Max is EUR 199 and adds model choice and more capacity. The cost follows the engineers who use it, not the size of the fleet.
Bring a real queue and we will run it through ARIA. See it live