METRIK

FOR ISPs & WISPs

It checks the link before it answers. Because we run a WISP too.

Most assistants can look up what a subscriber owes. Ours checks the outage map, the CPE, the parent AP and the signal history first — so "is it down?" gets a real answer at 2am instead of an apology and a ticket.

The billing questions are easy. The technical ones are the night shift.

"What do I owe?" any assistant can handle. "It's slow", "it cut out again", "is it down in my area?" is where automation usually gives up and hands the conversation to whoever is awake — which on a small ISP is the owner. Those are also the contacts that decide whether a subscriber stays, and most of them are answerable from data the network already has.

HOW IT WORKS

Identifies the subscriber

Against your CRM (UISP / Odoo).

Runs a real diagnosis

Against UISP NMS — outages, CPE and parent-AP state, signal and traffic against history — before it says a word.

Resolves the routine load

Billing, debt, suspension, install status — in a median of 7 seconds, 24/7.

Escalates cleanly

Opening a UISP ticket with a technical summary, and never inventing data or promising times.

PROOF — FROM OUR OWN ISP

Median reply time: 7 seconds.

Runs 24/7 — including overnight, for real.

7,000+ messages handled over recent months, on a real ISP of ~800 active clients.

In production, wired into UISP (CRM + NMS) and Odoo.

The large majority of contacts are handled without a supervisor stepping in.

WHY THIS ISN'T A BOT YOU DROP IN

Connecting an assistant to WhatsApp takes an afternoon. Knowing that this subscriber's signal has been degrading for three days, that their parent AP is saturated at this hour, and that the honest answer is "yes, it's on our side, a tech is already assigned" — that's the part that took us years of running our own network to learn.

It's the difference between an assistant that deflects contacts and one that resolves them. The first makes subscribers angrier.

A fixed-scope implementation.

Typically 30–45 days, wired into your stack, then a monthly operations fee (hosting, prompt tuning, improvements).

FAQ

Is this a WhatsApp bot I just drop in?

No. A dropped-in assistant answers from a knowledge base and knows nothing about your network. This one identifies the subscriber in your CRM and diagnoses their link against live network data before it says anything — which is the difference between deflecting a contact and resolving it.

Wouldn't it be cheaper to hire another support person?

For the daytime shift, often yes — and we'd tell you so. Where this pays is the hours you can't staff: nights, weekends and the outage spikes when everyone writes at once. It doesn't replace your team, it takes the repetitive load off it.

What if the AI gets a technical answer wrong?

It doesn't improvise. When the diagnosis isn't conclusive it escalates with a technical summary instead of guessing, and it never promises a repair time on its own.

How long does implementation take?

A fixed-scope implementation, typically 30–45 days, wired into your stack, then a monthly operations fee.

Does it run 24/7?

Yes — median reply time of 7 seconds, running around the clock, including overnight.

What does it integrate with?

UISP (CRM + NMS) and Odoo, in production today on a real ISP.

See it running on a real network.