Platform

What the platform is for

Bridge plant-floor OT and enterprise IT — without becoming another control system.

Visibility Bots is meant for operations that already run PLC, SCADA, ERP, and MES — and still need one shared record for production, energy, evidence, knowledge, and follow-up work.

Visibility Bots platform — PLC, SCADA, MES, and ERP connected into one shared operational record — Product view

Architecture

Six layers, one operational record.

Each layer adds context to the one beneath it. Signals move up from the plant floor, and decisions move back down as alerts, tasks, and workflows.

Visibility Neo Agent

Operations assistant in Visibility Live — the model writes the sentence, code owns the number, every figure is sourced.

LiveAnalysisAnomalySteamReportsInventory

Application Products

Specialized interfaces onto one shared architecture rather than separate tools.

LiveVisionDocsComplianceGrowth Engn

Intelligence Layer

Where raw records become operational context, reasoning, and recommendations.

OntologyKnowledge graphRAGAnalyticsRulesML

Unified Data Layer

One operational record spanning machines, media, documents, and work.

Time seriesEventsVideo metadataDocumentsTasksEvidence

Connectivity Layer

Edge gateways and industrial protocols — including LoRaWAN — for machines and meters that do not already emit usable data.

IoTERPMESSCADAAPIsCamerasBLELoRaWANModbus

Industrial World

The physical operation the platform is there to explain.

MachinesSensorsCamerasPeopleProcessesAssets

Visibility Neo Agent

The model writes the sentence. The code owns the number.

The operations assistant in Visibility Live — floating chat on dashboards and the full Command Center, sharing one conversation per signed-in user.

Python computes totals, averages, fill and drain events, cooling cycles, shift windows, unit conversions, and tables. The language model classifies the question, picks which attribute to look at, and writes prose around figures that already exist.

Floating widget

Present on operational dashboards — tanks, steam, energy, flow, cold rooms, AQI, weather, solar, filling, inventory, and more. Hidden where a full chat already exists.

Command Center

The full Neo Agent experience: session history, project focus, charts when an analysis produced one, and the account-wide “what needs attention” scan.

Shared thread

The widget and Command Center read and write the same privacy-scoped history. A chat started on a tank page is the same thread in Command Center.

Operations LeadHow much oil came into ST7 last week?

Fill events for ST7 last week are reconstructed from level history and tank calibration — not from a lifetime counter. The computed intake total is listed with the window used, and the Source line names the asset, attribute, and time range that were fetched.

Source

ST7 · level history · last week Asia/Karachifetched context

Visibility Livefetched context

What Neo can answer

Skills for plant questions — not a free-floating chatbot.

Suggested prompts come from the solutions this user actually has. Answers stay inside the assets and stock that user is allowed to see.

Live

Current reading of a named asset or group — never history.

Example What’s the current level of ST7?

Analysis

Anything over a past period: trends, averages, min/max, fill and drain, cooling cycles, machine runs, shift counters.

Example How much oil came into ST7 last week?

Anomaly

Rule-based scans across many assets — smoke, offline devices, which machines or cold rooms are running. Not a machine-learning model.

Example Which cold rooms are running right now?

Steam summary

Production, consumption, and loss for the steam system, including daily breakdown from totalizer history differences.

Example How much steam did we produce today, by department?

Report

The same computed data as chat, rendered as Excel or PDF with a download link.

Example Generate a PDF for Cold Room 2 last week.

Inventory

Stock, batches, expiry, locations, vendors, and receiving or dispatch history. Writes are drafted and sent only after explicit confirmation.

Example How many bags of urea are in stock?

General

Small talk, reformatting the last reply, and textbook explanations. If a real asset matches the words, it says so.

Example What is a flow totalizer?

Plant clock

Asia/Karachi. Production day starts at 6:00 AM.

“Today,” “yesterday,” and “this week” follow the 6:00 AM production-day boundary unless the question clearly wants a calendar day.

Shift 16:00 AM – 2:00 PM
Shift 22:00 PM – 10:00 PM
Shift 310:00 PM – 6:00 AM

Domain knowledge

Asset kinds Neo already understands.

Kinds are recognized from the attribute names an asset reports. A new client asset of a known kind needs no code change.

Level tanksConverts level to tonnes from tank calibration, and reconstructs fill and drain events from level history.
Cold roomsInfers on/off cycles from temperature, and reports which rooms are running.
Filling machinesTreats machine status as running or stopped, keeps shift packet counts on the 6:00 AM day, and refuses to report instantaneous speed as a per-shift figure.
Steam traps and steam systemTrap health from live attributes; production, consumption, and loss from server-side totalizer differences — never by summing a lifetime counter.
Energy meters and solarKnows lifetime kWh versus counters that reset at 6:00 AM, and today’s solar production.
SCADA tagsUnwraps readings nested inside JSON data attributes so the tag is answerable.
OccupancyCurrent occupied or vacant state. Previous occupancy over time is not available yet.
InventoryStock and ledger with locations, vendors, and confirmed receive/dispatch.

Trust boundaries

Grounded answers — or a visible note that a figure could not be verified.

A general chatbot would read telemetry into a prompt and hope the model adds it up. Neo Agent does the opposite.

Session-scoped assets

Neo only sees assets registered for this signed-in user, re-checked on every read.

Code owns the number

Totals, tables, and events are computed in code. The model narrates figures that already exist.

Grounding

Every figure must appear in fetched data or follow by simple arithmetic. Unsupported numbers are retried, then flagged if still unverified.

Ask, don’t guess

Ambiguous asset names or unclear “now vs past” questions get a clarification instead of a silent pick.

Source line

Every data answer ends with a Source line built from what was actually fetched — asset, attribute, time window, or live reading time.

Lifetime-counter guard

“How much over a period” never sums a counter that never resets.

Confirmed inventory writes

Receive and dispatch are drafted first. Nothing is recorded until the user confirms.

Not yet

Honest limits of the current Live Neo Agent

Occupancy history over a past period
Period totals for water, gas, and power the way steam totals are built
Weather Station, computer vision, and smart water management as Neo skills
Solar totals beyond today
Machine-learning anomaly detection — current scans are coded rules

How it works

From connection to coordinated action

The same five steps apply whether the starting point is a machine, a camera, a document, or a spreadsheet.

Connect

Bring industrial and enterprise sources into one place.

Unify

Normalize signals, media, documents, and work into a shared record.

Understand

Build the operational context that explains what the data means.

Reason

Apply AI, analytics, and rules to find variance and risk.

Ask Neo

In Visibility Live, Neo Agent computes the figures in code, narrates them, and ends with a Source line.

Integration philosophy

Connect what you already have.

Visibility Bots is designed to sit alongside the systems an operation already depends on.

Keep what already runs

Visibility Bots does not require enterprises to replace SCADA, MES, ERP, or other systems they already use.

Connect the existing stack

Connect ERP, MES, SCADA, PLCs, databases, cameras, sensors, meters, and existing software.

Add intelligence in steps

Start with the operation that needs answers first, then extend across the platform.

Differentiation

Not a bolt-on chatbot. Not a standalone IIoT brochure.

Live, Vision, Docs, Compliance, and Growth Engn share one operational record on Visibility Bots. Visibility Neo Agent is the grounded operations assistant inside Live — code owns the number, the model writes the sentence.