Skip to content

Getting Started — AI Assistant

Opening the Assistant

The AI Assistant is accessible from the main navigation. Click the AI Assistant icon to open the chat panel. Each session starts a new conversation, and the assistant remembers the full context of your conversation as you go.

Managing Your Conversations

Past conversations are listed in the sidebar on the left of the chat panel, most recent first. From here you can:

  • Start a new chat — click New Chat to begin a fresh conversation without losing your previous ones
  • Return to a past conversation — click any conversation in the list to reopen it with its full history
  • Rename a conversation — hover over it and click the pencil icon to give it a more memorable title
  • Delete a conversation — hover over it and click the trash icon to permanently remove it and its messages

Collapse the sidebar using the panel icon at the top if you want more room for the chat itself.


What to Ask

The assistant can answer questions about your live plant data, historical records, energy metrics, and organisational structure. You do not need to use specific commands or syntax — ask naturally.

Plant & Asset Queries

Example Question What You Get
"How many machines do we have?" Count and list of machines across all workspaces
"List all workspaces" Full workspace list for your organisation
"What parameters does the Compressor machine track?" Parameter list for that machine
"Show me all production lines in Process A" Lines and their details

Asset Health, Maintenance & Reliability

Example Question What You Get
"What's the health score for Compressor 1?" Current health score, band, and the factors behind it (VFD assets)
"Which assets are in critical condition?" Fleet health overview, worst first
"What maintenance is overdue?" Every overdue maintenance plan, with how far past due
"Show me the maintenance history for Pump 3" That asset's plans and its logged service history
"What's the MTBF for Compressor 1?" Reliability KPIs — MTBF, MTTR, availability, PM compliance — over a chosen window
"Which assets fail the most?" Fleet reliability overview, worst first, with open-breakdown counts
"What inspections are overdue?" Every overdue inspection round

VFD Alarms & Troubleshooting

Example Question What You Get
"What VFD alarms are active right now?" Every currently-open drive alarm across your fleet, worst severity first
"Does Compressor 1 have any active alarms?" Active alarms for that specific machine
"How many times has this drive tripped this month?" Alarm history for the machine, filterable by type, severity, or date
"What's the SOP for this alarm?" / "How do I fix an overvoltage trip on this drive?" Step-by-step SOP for that alarm
"What drive model is this?" Drive configuration, ratings, and escalation matrix for the machine
"What does alarm code A32 mean?" Alarm code reference for the machine's drive family

See VFD Machines for how alarms and SOPs are set up.

CNC Machine Monitoring

Example Question What You Get
"What CNC machines do we have?" List of CNC machines, controller type, and hub
"Was this machine running yesterday afternoon?" Historical run-state timeline for the machine
"How many cycles did this machine complete today? What's the average cycle time?" Cycle log with counts and durations
"What's this CNC's OEE this week?" Computed Availability, Performance, Quality, and OEE for the period
"What alarms has this CNC had this month?" Alarm history, or currently-open alarms only
"Has tool wear been drifting on this machine?" Logged tool offset drift events
"Why was this machine down yesterday?" Operator-logged downtime with reasons

See CNC Machines for setup and the Historian for the underlying data.

Production, Work Orders & OEE

Example Question What You Get
"What is running right now?" / "How far along is production?" The live Execution board — every in-flight work order with its progress and current step
"What work orders are open?" Work orders with status, product, quantity, and scheduled dates
"Which step is this order stuck on?" The order's job cards — one per routing step — with stage, machine, and dates
"Why is this step late?" / "How long did it actually take?" Planned versus actual duration for that step, and by how much it varied
"What was produced on each shift for this step?" Per-shift produced, rejected, and scrap quantities, with the downtime recorded against them
"What was OEE on the press yesterday?" OEE for that machine, shift, and date, with the minutes and counts behind it
"What was OEE for this order?" OEE for the work order — measured per routing step where the order is routed
"How did the bottling line run this shift?" Line OEE, built from the runs and stoppages the line itself recorded
"Is OEE improving?" OEE per day across the period you ask for
"What is causing our downtime?" Downtime grouped by reason, longest first, for one machine or plant-wide
"Why is Performance blank on the press?" Which products have a cycle time standard set for that machine or line, so you can see which one is missing
"When did the cycle time for this product change?" Every version of that standard and the date each one took effect
"When did the press stop yesterday, and for how long?" Each logged stoppage with its times, duration, reason, and operator notes — including any still running
"What downtime reasons do we have?" The reasons operators choose from, grouped by category

There are two kinds of OEE, and the assistant will tell you which it used

A CNC machine works out its own OEE from cycle data (see the section above). The figures here are built instead from logged stoppages and production counts — and for a routed order, from its individual job cards rather than one figure spread across every machine it touched. Name a machine, an order, or a line and the assistant will say which basis it used.

A blank figure is not a zero

If the assistant reports a figure as unavailable rather than 0%, something needed to work it out is missing — usually nothing was logged, or no cycle time standard is set for that machine or line. A genuine 0% only happens when a machine was logged as stopped for an entire shift. Note too that rework lowers Quality, not Performance: the machine still made the unit, it just wasn't right first time.

See OEE and Execution for the underlying features.

Parameter Compliance & Deviations

Example Question What You Get
"Which parameters are performing worst?" / "What's our compliance on critical control points?" Fleet-wide parameter health overview, worst first
"Why did the cold store go out of range last week?" The excursion record for that parameter — reading, limits breached, and the recorded reason
"How many genuine deviations did we have last month?" Deviation counts filtered by cause
"Which batches were affected by that temperature excursion?" Affected batches, for traceability

See Parameter Health — Monitoring & Reports for the full feature.

Fleet-wide Alerts

Example Question What You Get
"What issues need attention right now?" A single, chronological list of open issues across triggers, VFD alarms, CNC alarms, overdue maintenance, and overdue inspections — the same view behind the notification bell and Machine Issues dashboard
"Are there any critical alerts?" Open issues filtered by severity
"What's wrong with Compressor 1?" Every open issue for that machine, across all sources

Energy Management

Example Question What You Get
"What is our energy consumption this month?" Current-month kWh total with source breakdown
"How does this month compare to last month?" Side-by-side comparison with percentage change
"How much solar energy did we use in May?" Solar incomer consumption for the date range
"Which section uses the most energy?" Section-wise consumption breakdown
"What was our peak demand last quarter?" Peak kVA per incomer type
"Is our power factor at risk of a penalty?" Power factor trend with penalty threshold flagging
"How close are we to our sanctioned MD limit?" Demand utilisation percentage and headroom

Reports

You can ask the assistant to generate a formatted report. Reports can include tables, charts, and multi-section summaries.

Example requests:

  • "Give me a full energy report for this month with charts"
  • "Generate a monthly consumption comparison for the last 6 months"
  • "Summarise our peak demand events for Q2 with a table"

Reports can run to several pages. The assistant collects the data it needs automatically before writing the report — you may see it working through multiple steps before the final output appears.


Complex and Multi-Step Analysis

The assistant is designed for in-depth analysis, not just quick lookups. You can ask layered questions that require data from multiple sources:

"Compare energy consumption across our three production lines, identify which one is most efficient per unit of output, and flag any anomalies in peak demand."

The assistant will call the relevant data tools, collect the results, synthesise them, and present a single coherent answer. You do not need to break this into separate questions.

Tips for complex queries:

  • Be specific about the time period: "last month", "Q1 2025", "between 1 Jan and 31 Mar"
  • Name the machine, workspace, or incomer if you want data scoped to a specific asset
  • Ask for charts when comparing values across categories or trends over time — the assistant will include them automatically where relevant
  • Follow-up questions within the same conversation build on what was already retrieved — you do not need to repeat context

How the Assistant Remembers Context

Within a conversation, the assistant keeps track of:

  • What you have asked and what data it has already retrieved
  • Decisions or conclusions reached earlier in the conversation
  • Your stated goals or constraints (e.g., "we are investigating high peak demand in June")

This means follow-up questions work naturally:

You: "Show me energy consumption for all incomers in June."

Assistant: (retrieves and shows data)

You: "Now break that down by week."

Assistant: (uses the same incomers and date range — no need to repeat them)

The assistant retains this context across a long session. If a conversation grows very long, older exchanges are summarised rather than dropped entirely, so key facts and data from earlier in the session remain accessible.


Charts

When the assistant presents numerical or comparative data, it will automatically include a chart where one adds clarity:

  • Bar charts for comparing values across categories (e.g., consumption per section)
  • Line charts for trends over time (e.g., monthly demand)
  • Pie charts for proportional breakdowns (e.g., energy share by incomer type)

You can also explicitly request a chart type: "Show me that as a bar chart" or "Give me a line graph of daily consumption for the week."


Adjusting the Assistant for This Chat

Click the settings icon in the chat header to adjust how the assistant behaves for the current conversation:

  • Use Knowledge Base (RAG) — turn off if you don't want answers grounded in your organisation's uploaded documents
  • Enable Tools — turn off to have a plain conversation without the assistant querying platform data (useful for general questions)
  • Temperature — lower makes answers more focused and consistent; higher makes them more varied and exploratory

These settings apply only to your current conversation and reset for new ones.

Tips for Best Results

  • Start with a clear goal: "I want to understand why our peak demand increased in June" gives the assistant more to work with than "show me demand data"
  • Ask for reports when you need something shareable: the assistant can produce a structured, multi-page document you can download
  • Use follow-ups freely: the assistant is designed for back-and-forth — refine, drill down, or change direction at any point
  • Mention constraints: if you only want data for a specific workspace, shift, or date range, say so upfront