What it looks like

Two screens and you know how your network is doing

One panel for technical operations and another for the money. Click any screenshot to see it larger.

tuempresa.fishernetworks.net/welcome/dashboard

NOC dashboard. Overall network status, OLTs, Mikrotik and ONTs with low signal, on a single screen.

Network health on loginDevices online, in maintenance and down, with the latest reading of each.
ONTs with low signalThe system tells you which customers are below the optical threshold before they report.
NOC quick accessManagement of OLTs, Mikrotik, ONTs and customers with a balance due, one click away.

The month's money, in viewCollections today and for the month, overdue invoices, what falls due in 7 days and the collection rate.
Portfolio in numbersActive customers, suspended, active PPPoE and contracted services.
Top debtorsWho owes, how much and how many invoices, with direct access to collect.

tuempresa.fishernetworks.net/dashboard2

Operations dashboard. Collections, portfolio and open cash registers for the day, with a 12-month trend. It is a freshly created demo environment: that is why the month's collected amounts show as zero.

Module by module

What each part of the system does

Each block is a real CRM screen, not a designer's mockup.

Customers and portfolio

The whole customer record in one place

Contract, services, balance, billing cutoff day, documents and complete history. You search by name, phone, contract, PPPoE user or address.

Own contract number (FN000123) to identify the customer on any channel
Balance in green or red right in the list, without opening the record
Guided sign-up with a wizard, or bulk import from Excel
Customer documents: contract, ID and receipts
Change audit: who modified what and when

See which plan it comes in

Collections and cash register

Know whom to collect from today, and collect

The system builds the list of pending payments and from there you collect. The cash register balances the shift on closing, with a count and a breakdown by payment method.

Automatic invoicing by cron and scheduled payment reminders
Cash register with shift opening, the day's collections and count on closing
Cash, transfer, OXXO and complimentary kept separate, because cash is counted and a transfer is checked at the bank
Point of sale for equipment and accessories sales
Automatic suspension and reactivation of the service according to payment
Collection reports by month, by employee and by service

The shift's cash register. What is in the drawer, the total collected and the breakdown by payment method, with the latest collections live.

The count. On closing, the system fills in the expected amount and the person types what they counted: the difference is saved exactly as seen.

Whom to collect from today. The list of pending payments, with a collect button on each row.

Technical support

A console designed for the technician in the field

A screen separate from the CRM, light and mobile-first: unassigned tickets, "my day", customers and network status. When you open a customer you see their service, their ONT and their NAP box.

PPPoE username and password, assigned equipment and service status
ONT with serial number, PON port and OLT, with live check
NAP box, port and splitter, so you never have to guess again where it hangs from
Suspend, reactivate or cancel the service from the same screen
Tickets with priority, area, owner and response times

Network and provisioning

The network connected to the CRM, not in a separate spreadsheet

Huawei and VSOL OLTs, Mikrotik routers via API and ONTs via TR-069 with an ACS built into the system. Creating a service creates the PPPoE and leaves the equipment ready, without entering the equipment console.

Inventory of OLTs, Mikrotik, NAP boxes, zones and ONT models
PPPoE creation and removal on Mikrotik from the CRM itself
TR-069: reboot the ONT, change the WiFi or upload an image remotely
Outages per PON port with the list of affected customers
Device status with last reading and Telegram alert

NOC monitor FTTx plant Native TR-069

Usage and capacity

Who is eating your bandwidth

Daily usage per PPPoE customer, with download and upload separate and a ranking of the heaviest users. Useful for capacity planning and for justifying a plan upgrade.

Chart of the whole network's daily usage by date range
Top 20 customers by usage, with their PPPoE user
Per-customer detail to see their usage pattern
Abuse detection before it degrades the node's neighbors

WhatsApp with artificial intelligence

The channel where people really write to you

Bot connected to the official WhatsApp API: it answers questions, receives fault reports, collects payment and escalates to a human when needed. And it learns who each number belongs to.

Configurable conversation flows, no code
Reads the payment receipt with AI vision and proposes to apply it
Automatic technical diagnosis: optical signal, ONT down, PON with outage
Inbox of numbers to identify, with what each one said
Chat inbox with agent assignment and response metrics

More than half of the numbers that write in don't match any record. This inbox turns them into identified customers.

NOC monitor

Your network watched on its own, even at dawn

The monitor polls each device every minute, knows when something stopped responding and alerts the NOC group before the first phone rings. It is not a separate Zabbix: it lives inside the CRM, so every outage already knows which customers it affects.

tuempresa.fishernetworks.net/noc/monitor

Network monitor

Live · polling every 60 s

128Devices monitored

124Responding fine

3Over threshold

1No response

Incidents per hour · last 24 h Peak at 2:31 pm

Status per device

Device	Type	Status	Latency	24 h availability
OLT-CENTRO10.0.0.11	OLT Huawei MA5608T	Online	2 ms	99.98 %
OLT-NORTE10.0.0.12	OLT VSOL V1600D	Online	4 ms	99.91 %
MK-BORDE-0110.0.0.1	Mikrotik CCR2004	Online	1 ms	100 %
MK-NODO-SUR10.0.0.21	Mikrotik RB5009	CPU 87 %	6 ms	99.62 %
RADIO-CERRO10.0.0.34	5 GHz point-to-point link	Down 8 min	—	96.40 %

Latest events

RADIO-CERRO stopped responding 5 consecutive missed polls. Alert sent to the NOC group on Telegram. 14:32
PON 0/1/3 of OLT-CENTRO: 14 ONTs down at once A single outage was opened with the 14 affected customers, not 14 tickets. 14:31
12 ONTs below −27 dBm They go to the low-signal inbox to be checked before they report. 13:58
MK-NODO-SUR recovered It responded again after 2 min 14 s. The incident closed itself with the downtime recorded. 11:20

Network operations center monitor view rebuilt on this page with sample data, so that it reads well on mobile. The screenshots with the browser frame and the enlarge button are from the real system.

It doesn't alert at the first missOne lost ping is not an outage. The monitor requires several consecutive failures and checks another path before raising the alarm.
Own threshold per deviceThe CPU of an edge CCR and that of a node RB aren't measured the same way. Each device carries its own limits for CPU, memory, temperature and latency.
Every outage knows whom it affectsBecause the monitor lives inside the CRM, when a PON or a node goes down the list of customers hanging from it appears, with their phone numbers, so you can warn them before they call.

What it watches, and how it asks

Each device family is queried over the protocol it really supports, not everything by ping.

OLTs over SNMP

Temperature, usage of each PON port, registered ONTs, ONTs down and the optical power each one receives. Huawei and VSOL with their own OIDs.

Mikrotik over the RouterOS API

CPU, memory, uptime, active PPPoE sessions and traffic per interface. No agents or scripts to maintain on the router.

ONTs over TR-069 and OMCI

Whether the ONT answers, with what signal and since when. What the OLT sees is cross-checked with what the ONT itself says, which don't always match.

Links and servers over ICMP

Latency, packet loss and availability of radio links, towers, DNS and internet exits, with history to claim from the carrier.

The system itself

That the billing cron ran, that the WhatsApp queue didn't clog and that last night's backup completed. A CRM that is down silently is worse than a router that is down.

History and availability

Every reading is stored. You can pull the real uptime of a site from last month without depending on anyone's memory.

1. Detects The probe fails several times in a row. A stray latency spike wakes nobody. Silently

2. Confirms It checks again and looks at the upstream device: if the whole node went down, you are not alerted five times. Groups the outage

3. Alerts Telegram to the NOC group and email to the owner, with the device, since when and how many customers it affects. Telegram and email

4. Closes When the device returns, the incident closes itself and the downtime is kept for the month's report. Closes itself

The alert, exactly as it arrives

To the NOC group, not a dashboard nobody watches

Most outages happen when nobody is in front of the screen. That is why the alert goes out on Telegram with what is needed to decide whether someone gets up or waits until tomorrow.

Which device it is, its IP and since what time it has been down
How many customers hang from it, to gauge the urgency
A second message when it recovers, with the total time
Different groups per zone, so you don't wake the one in the other city
Silenced by a maintenance window, so the group isn't flooded with noise when the outage is yours

NOC · Your Company14 members

🔴 DOWN · RADIO-CERRO
5 GHz PtP link · 10.0.0.34
No response since 2:24 pm (8 min)
Customers hanging from this link: 63 14:32

🟠 PON OUTAGE · OLT-CENTRO 0/1/3
14 ONTs down on the same port in under 60 s
Outage #218 opened with the 14 affected customers 2:32 pm

🟢 RECOVERED · MK-NODO-SUR
Back at 11:20 am · downtime 2 min 14 s 11:20

Alerts come out in this format: device, since when and how many people it affects.

FTTx plant

Your fiber plant documented, not in the installer's head

Each customer hangs from a specific thread: an OLT, a PON port, a splitter, a NAP box and a port on that box. When that is written down, an outage is solved by phone; when it isn't, it is solved by opening boxes to see which one it was.

OLT Huawei MA5608T
Make, model and site SNMP + Telnet

PON port 0/1/3
With its real occupancy 8 ports

Splitter 1:8, 1:16 or 1:32
With its loss in dB Cascaded

NAP box NAP-CTR-01
Location and coordinates Port 6 of 8

Customer's ONT HG8145V5
Serial and optical signal −21.4 dBm

From the OLT to the customer's wall outlet. The CRM stores the complete chain, and from the customer record it can be traversed in both directions.

Plant inventory

Know where you have a free port before promising an installation

The salesperson promises Thursday's installation and the technician arrives at a full NAP. With occupancy on screen that stops happening: coverage is checked by address and the system tells you which box has a free port.

Inventory of OLTs, cards, PON ports, splitters and NAP boxes with their coordinates
Occupancy per PON port and per box, updated with what is actually connected
Port reservation when the contract is signed, so another installation doesn't take it
Cascaded splitters: first level on the pole, second in the box
Zones and sectors, to distribute capacity and see where to expand
Feasibility check by address before selling the service

And the customer's equipment configures itself via TR-069

tuempresa.fishernetworks.net/fttx/planta

FTTx plant · OLT-CENTRO

Huawei MA5608T

8PON ports

512Registered ONTs

71 %Plant occupancy

12NAP boxes

Occupancy per PON port

Port	Splitter	NAP box	Occupancy	Status
0/1/0	1:8	NAP-CTR-01	7 of 8	Normal
0/1/1	1:16	NAP-CTR-02	11 of 16	Normal
0/1/2	1:16	NAP-SUR-04	16 of 16	No ports
0/1/3	1:8	NAP-SUR-07	8 of 8 · 14 ONTs down	Outage #218
0/1/4	1:32	NAP-PON-11	13 of 32	With headroom

NAP-CTR-01 box · 8 ports

19.594, −100.389

1234 5678

Occupied Signal warning Free

Optical budget · port 6 (customer FN000412)

−21.4 dBm

−8 to −18Optimal −18 to −24Warning −24 to −30Critical

Today's reading: −21.4 dBm. 30 days ago it was −19.8 dBm: the system flags the progressive drop before the customer is left without service.

View of the plant module rebuilt on this page with sample data.

ONT autofind

The OLT sees a new unauthorized ONT and the CRM shows it with its serial number and its PON port. You pick the customer and authorize from there, without opening a console.

Sign-up that leaves the service running

When the ONT is authorized, the PPPoE is created on the Mikrotik, the NAP port is occupied and its speed profile is applied. One operation, not four screens.

Outage per PON, not per customer

If several ONTs on the same port go down at once, it is fiber, not their routers. One outage is opened with the list of affected customers and the alert goes out only once.

Optical history per customer

Each ONT's signal is stored day by day. A degrading fiber shows up in the curve weeks before the customer is left without internet.

Plant map

NAPs, poles and customers geo-referenced. The technician opens the map on their phone and sees which box the customer they are visiting hangs from.

Transfers and box changes

When a customer moves or a port is rebalanced, the NAP and port are changed in one step, and the occupancy of both boxes adjusts itself.

GPON with Huawei OLT

MA5608T and MA5683T over SNMP and command line: autofind, authorization, signal per ONT and reboot.

GPON and EPON with VSOL

V1600D and equivalents, with their own command set and line profiles.

FTTH over PPPoE on Mikrotik

The OLT in bridge mode and the Mikrotik as concentrator: it is the most common setup and is supported out of the box.

And the wireless side

If you also have radio customers, they coexist in the same inventory with their sector and tower.

Native TR-069

An ACS inside the CRM, with no separate server to maintain

The CWMP server runs inside the system itself. The ONT sends its Inform to your domain and appears in the customer record without anyone registering it by hand. No GenieACS, no Node, no extra database to back up. And if you already have GenieACS set up, it can still be used through the API.

tuempresa.fishernetworks.net/tr069/dispositivo/FN000412

ONT HG8145V5 · contract FN000412

Online Last Inform 42 s ago

V5R020C10S115Current firmware

−21.4 dBmOptical receive signal

11 d 04 hUptime

300 sInform interval

Parameter tree

InternetGatewayDevice.
DeviceInfo.SerialNumberHWTC5A2B1C4D
DeviceInfo.SoftwareVersionV5R020C10S115
LANDevice.1.WLANConfiguration.1.SSIDFamilia_Ramirez
LANDevice.1.WLANConfiguration.1.KeyPassphrase••••••••
LANDevice.1.WLANConfiguration.5.Channel36 (5 GHz)
WANDevice.1.…WANPPPConnection.1.Usernamefn000412
WANDevice.1.…WANPPPConnection.1.StatusConnected
X_HW_…OpticalDiagnosis.RXPower−21.4 dBm
ManagementServer.PeriodicInformInterval300

The parameters marked in blue are the ones the customer just changed from their portal.

Device task queue

SetParameterValues WiFi name and password, requested by the customer from the portal. Applied in 3 s. 42 s ago
Download · Firmware Upgrade Image V5R020C10S120 · batch of 184 ONTs, window from 2:00 to 5:00 am in progress
Reboot Remote reboot launched from the support console. Back in 51 s. yesterday 6:12 pm
GetParameterValues Reading of optical signal and WAN status. Scheduled every 5 minutes. scheduled

CWMP session

21:04:11 Inform 1 BOOTSTRAP, 4 VALUE CHANGE oui=00259E sn=HWTC5A2B1C4D

21:04:11 InformResponse 200 session open, "GPON residential" profile applied

21:04:12 GetParameterValues InternetGatewayDevice.DeviceInfo. OK

21:04:12 SetParameterValues …WLANConfiguration.1.SSID OK 0

21:04:13 Download 1 Firmware Upgrade Image in progress

21:04:14 ConnectionRequest accepted http digest, source ACL verified

21:04:15 TransferComplete OK session closed in 4.1 s

View of the TR-069 console rebuilt on this page with sample data. The parameter names and RPCs are the real ones from the standard.

What talks to the device

The standard's RPCs, and what the CRM uses them for

TR-069 has a reputation for being complicated because almost nobody explains what each thing does. This is what the system actually uses, and what it does for you.

Supports Huawei, VSOL and ZTE ONTs and generic routers with TR-098 or TR-181
Each model carries its own parameter template, because the WiFi branch is not in the same place on all of them
If one of your models isn't there, its template is added without touching the rest of the system

RPC	What it is used for
Inform	The device introduces itself and is linked to the customer by its serial number.
GetParameterValues	Optical signal, firmware, current SSID, WAN status and connected devices.
SetParameterValues	Change WiFi, channel, VLAN, DNS or the Inform interval itself.
AddObject / DeleteObject	Create or remove the WAN connection, a guest SSID or a port mapping.
Download	Firmware or configuration file, one by one or in batches with a time window.
Reboot	Remote reboot from the record, from support or from the WhatsApp bot.
FactoryReset	Leave the device blank before removing it or reinstalling it at another address.
ConnectionRequest	Talk to the ONT right now, without waiting for the next Inform.
TR-143	Ping, traceroute and download test launched from the customer's device.

Installation without touching the device

The ONT comes out of the box, sends its first Inform and receives its profile: WiFi, PPPoE, VLAN and management password. The technician mounts it and leaves.

Profiles per model

One profile for the Huawei units, another for the VSOL ones, another for business routers. You change the profile and it applies to all devices in that group.

The customer changes their own WiFi

From the portal or the app they set the name and password they want. Underneath it is a SetParameterValues, and that call to support ceases to exist.

Firmware in batches

You select the devices of a model, choose the overnight window and the system updates them with retries and a report of which ones were left out.

Diagnosis before sending a technician

Optical signal, WAN status, devices connected to the WiFi and a ping test from the customer's own router. Many visits are cancelled right here.

With security in place

Different CWMP username and password per device, HTTPS, ConnectionRequest with digest and a list of allowed origins. An open ACS is a door into your entire plant.

And all of this, connected

The customer writes "I have no internet" and the system already knows the answer

These three pieces don't work separately. When someone reports a fault on WhatsApp, the bot checks the TR-069 of their ONT, checks the PON port it hangs from in the FTTx plant and checks whether the NOC monitor has an open outage on that node.

If there is an outage on their PON, they are told the truth and no duplicate ticket is opened
If their ONT is down and the neighbors' are not, it is their home: they are guided or a visit is scheduled
If the optical signal is in warning, the ticket already comes out with that reading written in
If it was only the WiFi, they are offered a remote restart of the device, right there

See the rest of the modules Request a demo of this part

Your Company Internetonline

Hi, my internet went down 2:33 pm

I already checked, Mr. Ramírez. We have a fiber outage affecting your area since 2:31 pm and a crew is on the way. It is not your equipment and you don't need to do anything. 2:33 pm

I will let you know here as soon as it is restored. Your report was recorded with ticket number #218. 2:33 pm

That answer comes from crossing the NOC monitor with the FTTx plant. Without that, the bot can only say "we will look into it".

The technician is on the way

The customer knows what is going on without calling anyone

A failed visit is two trips and an annoyed customer. When the technician marks "On the way", the customer gets a WhatsApp and can follow the visit from their portal.

tuempresa.fishernetworks.net/portal/ticket/53

This is how your customer sees it. The visit tracking in their portal, with the time window and what they have to do.

Your Company Internetonline

Your visit is scheduled for today between 4:00 and 6:00 pm. We will notify you again when the technician heads to your address. 11:32

Our technician is on the way to your address for report #53. Please make sure an adult is home. You can follow it here: tuempresa.fishernetworks.net/portal 15:48

The texts are the ones the system sends, word for word.

Scheduled "Your visit is scheduled." Sent with the date and the time window. Notifies by WhatsApp

On the way "The technician is on the way. Please make sure an adult is at the address." Notifies by WhatsApp

Working Already at the address. Nothing is sent here: the technician is in front of the customer. Doesn't notify, on purpose

Finished "We finished the visit. If anything wasn't right, reply to us right here and we will come back." Notifies by WhatsApp

A button for the technician

From the phone they mark the status with one tap. They don't write messages or copy phone numbers: the alert goes out on its own.

A time window, not an exact time

It says "between 4:00 and 6:00 pm" because that is what can be met. Promising an exact time is promising an annoyed customer.

Fewer failed visits

The "make sure an adult is home" notice is the only thing between a visit that works the first time and a "nobody was there".

Online payments

Get paid without anyone leaving home

Payment link through WhatsApp or from the customer portal. The payment comes in, is applied to the invoice and the service reactivates itself.

Stripe Card, Apple Pay and confirmation webhooks

Mercado Pago The most used method in Mexico, with webhook

Conekta Card, SPEI and cash at stores

Receipt by WhatsApp AI reads it and leaves it ready to reconcile

Customer portal and app

Your customers serve themselves

Each customer signs in with their email and sees their own, without calling the office. That lowers calls and raises collections.

Their invoices and payment history, with PDF download
Online payment of the month's invoice
Payment methods always in view, without asking on WhatsApp
Report a fault and follow their ticket, even see whether the technician is on the way
My WiFi: change the name and password of their network from the portal
Mobile app for Android, and the portal works the same on iPhone

See the app View plans

tuempresa.fishernetworks.net/portal

The customer's home page. What they owe, their services with their fee, the invoices to pay and how much internet they have used in 30 days.

Their invoices. Issued, due date, amount, what has already been paid and the pay button on the one that is due.

Day-to-day operation

The things you'll be grateful for after six months

They don't show in a five-minute demo, but they are the ones that prevent scares.

Importer with undo

Upload your Excel, it shows you how it understood each column, you can simulate without writing anything and, if something goes wrong, you undo the whole batch.

Backups outside the server

Automatic copies that leave the server itself, so that a hardware problem doesn't take your backups with it.

Roles and permissions per menu

Superadmin, admin, purchasing, supplier, cash and customer. Each role sees only what is theirs, checked on every request.

Truly multi-company

Each company with its own database and its own files. Neither a customer nor a document gets mixed between companies.

Automatic tasks

The month's invoicing, payment reminders, suspensions, device reading and notifications run on their own and leave a record.

Audit and log

Changes to customer records, logins and sensitive actions are recorded with user and date.

Telegram alerts

A device that goes down or a new outage reaches the NOC group without anyone watching the screen.

REST API

Endpoints for PPPoE, TR-069, FTTx plant and whatever you already have set up (GenieACS included), with keys and a call log.

Demo mode

The system can be opened to an outsider with creations and deletions blocked, with no fear that they touch anything.

And the administrative side

Purchasing, inventory and employees

If you also want to control spending and the warehouse, the ERP lives in the same system with the same catalogs.

Requisitions and approvals

Who requests, who authorizes and what step each request is at.

Purchasing and suppliers

Purchase orders, comparison, payment complements and supplier files.

Inventory and warehouses

Stock, movements and equipment assigned to each technician.

Employees

Payments, materials and work clothing, openings and rewards.

Budgets and expenses

Expense accounts, budgets per area and control of what has been spent.

Reports

Sales, collections by month and by employee, services and customers by zone.

See the ERP features in detail

Why trust this system

It is not a showcase product: it is our own work tool

Fisher Networks is an ISP with a Single Concession from the IFT. This CRM is the one we use every day to invoice, collect, handle WhatsApp and watch our own network. Everything you see here exists because we needed it first, and if something breaks we fix it the same day, because it breaks us too.

Technical questions

What other ISPs ask

Does it work with my equipment?

With Huawei (MA5608T, MA5683T) and VSOL OLTs, with Mikrotik through its RouterOS API and with ONTs through TR-069 using the ACS built into the system. If you have another manufacturer, tell us the model and we will check the integration before you sign up.

Is TR-069 GenieACS or is it yours?

It is ours: the CWMP server runs inside the same system, so you don't have to set up or back up a separate GenieACS with its Node and its MongoDB. The ONTs send their Inform to your own domain and are linked to the customer by serial number. If you already have a GenieACS running and prefer not to move it, it can still be used through the API. See the TR-069 console.

What can be done remotely on the customer's ONT?

Read its optical signal and WAN status, change the WiFi name and password, change channel or VLAN, reboot it, factory-reset it, upload firmware in batches in an overnight window and launch a ping or a traceroute from the customer's own device (TR-143). The customer can also change their WiFi themselves from the portal.

Does the NOC monitor need another server, a Zabbix or a LibreNMS?

No. The monitor lives inside the CRM and polls the devices every minute over SNMP, ICMP and the RouterOS API. The advantage of it being inside is that every outage already knows which customers it affects: when a PON port goes down, a single outage is opened with the list of those affected instead of fourteen loose tickets. See the monitor.

My fiber plant is already built and isn't documented, does it still work?

Yes, and it is the normal case. You start by importing OLTs, PON ports and NAP boxes, and from there the system itself fills in the rest: each ONT the OLT already has registered is hung from its port, and the NAP port is completed at the technician's first visit. After a few weeks of normal operation the plant is documented without stopping anyone to do it. See the plant module.

I already have my customers in Excel or in another system, how do I move them?

With the importer: it generates a template, shows you how it understood each column, lets you simulate the load without writing anything and apply it when it looks right. If something goes wrong, the whole batch is undone.

Is my data kept separate from other ISPs' data?

Yes. Each company has its own database and its own file folder: you share no tables with anyone. It is one of the things we test explicitly in every update.

Does the WhatsApp bot use the official API?

Yes, Meta's official WhatsApp Cloud API, with your own number. No unofficial apps that break or get your number blocked.

How does reading receipts with AI work?

When the customer sends the photo or PDF of their transfer, the system interprets it, extracts amount, date and reference, and leaves it in an inbox for the cashier to confirm with one click. The final decision is always made by a human.

Can I request changes or new modules?

Yes, the development is ours. Custom changes are quoted separately according to scope, and stay on your installation.

Can I try it before paying?

Yes. The Free plan comes with a 7-day demo, and we can set you up a demo with sample data to walk through with your team. Write to proyectos@fishernetworks.net.

Try it with your own data

We set up a demo with sample data and walk you through it. If it works for you, you choose a plan.

See plans and pricing Request a demo

The system that runs a complete ISP