Dashboards

A dashboard is a saved grid of widgets. Each widget shows telemetry variables, KPI values or alert counts from your organization, and re-reads them on its own schedule. You can open any dashboard full screen, and put it on a control-room wall screen that shows the widgets and nothing else.

Where to find dashboards

• In the sidebar: switch to the Insight space, then open Analyze → Dashboards. The page lists every dashboard, most recently updated first, with its widget count and the date it was last updated. Renaming a dashboard and adding, changing or removing a widget all count as updates.
• On the Insight home: open Home in the Insight sidebar. The Dashboards section lists up to five dashboards, most recently updated first, and links to All dashboards. When your organization has none yet, it says No dashboards yet and, if your role can create dashboards, offers Describe a dashboard and Create Dashboard.

Permissions

Dashboard actions use the Dashboards permissions from User Management:

Each of these buttons shows only when your role has its permission. Without Write, a dashboard opens read-only: you can read its widgets and open it full screen, and nothing on it changes the dashboard.

Create a dashboard

1. On the Dashboards page, click Create Dashboard.
2. Enter a Name. Description (optional) can stay empty. Each dashboard in your organization needs its own name. Capital letters, in any language, and spaces at either end don't make a name different. If the name is taken, the form says so under Name and keeps what you typed.
3. Click Create. The new, empty dashboard opens.

Describe a dashboard

Instead of adding widgets one by one, you can describe the dashboard you want in plain language. The assistant drafts it, you see it with your live data, and it is created only when you have answered each of its assumptions and approved it. The assistant proposes; a person approves. It cannot create or change a dashboard on its own.

Click Describe a dashboard on the Dashboards page. The button is also on the empty Dashboards page and in the empty Dashboards section of the Insight home. It shows only when your role has the Write permission for dashboards.

1. Pick the population. In Population, on the left, choose the assets the dashboard is about. The assistant binds widgets to these assets only.
2. Describe it. In the Agent panel, say what you want to watch, for example "Power, availability and state of charge for these assets over the last day", then click Start with the assistant. Narration and each lookup the assistant makes appear as it works. Stop ends the turn and keeps what was written.
3. Watch the draft form. On the right, the draft outline fills in as the assistant writes: the name, the description, and the widgets. The outline is not validated yet and cannot be approved.
4. Review the proposal. When the draft passes validation, its name, description, Why and What this does appear, followed by the Live preview: the dashboard approval creates, drawn with your live data. Nothing exists until you approve it.
5. Answer each assumption. Under What the assistant assumed, each guess the assistant made is listed, most consequential first, with the data it is based on. Click This is right to accept it, or This needs changing to dispute it. There is no way to accept them all at once.
6. Approve. Approve and create says how many assumptions still need an answer, for example "1 of 3 still need an answer", and does nothing until every one has one. When all are answered, click it. The dashboard is created and opens.

When an assumption is wrong

Approving creates the dashboard exactly as drafted, so a disputed assumption blocks approval. After This needs changing, write what needs changing, then either:

• click Send to the assistant: your note is sent as your next message, and the assistant drafts again. A new draft starts with no answers, because its assumptions are new;
• click Reject with this note to reject the draft with your note as the reason.

To reject a draft for any other reason, click Reject, say why, and click Reject draft.

When a draft can no longer be approved

If the assets or variables a draft is bound to changed after it was drafted, or someone else decided it first, approving is refused and the page says why. A draft that can no longer be approved stays on screen marked This draft is superseded, with Ask the assistant to draft it again. If someone else approved it, the page says who and links to the dashboard with Open the dashboard. If it was rejected, the page says by whom and why.

If another dashboard in your organization already has the draft's name, nothing is created and the draft stays open. The page links to that dashboard and offers Ask the assistant for another name. You can also rename or delete the other dashboard and approve again.

If no answer comes back, it is not known whether the dashboard was created, and a message offers Retry. Retrying opens your dashboard if it was created, and never creates a second one.

Reloading the page resumes the conversation, the latest draft and its status. Start over in the Agent panel begins a new conversation.

Limits

• A draft has at most 8 widgets.
• The assistant does not add an Alert Summary. Add one yourself in edit mode.
• Widgets are bound to variables and KPIs on the assets in the population you picked, and nothing else.

Agent-drafted dashboards

A dashboard created this way shows Agent-drafted and Approved by followed by the name of the person who approved it, next to its name and on its card on the Dashboards page. After that it is an ordinary dashboard: rename it, edit its widgets or delete it like any other.

Rename, clone or delete a dashboard

• Rename: open the dashboard and click its name. Type the new name, then press Enter or click the check mark. Escape cancels. If another dashboard already has the name, the message under the field says so.
• Clone: on the Dashboards page, hover over a dashboard, open its Actions menu and click Clone. The copy is named after the original with (copy) added, or (copy 2), (copy 3) and so on when that name is taken, and carries its description and all of its widgets. If the original holds a widget of a type this version no longer supports, the clone leaves it out and the confirmation says how many were left out. A dashboard holds at most 24 widgets, so the copy of a dashboard made before that limit keeps its first 24 widgets, and the confirmation says how many were left out. If a widget uses more data sources than a widget can show today, the copy keeps as many as fit, and the confirmation says how many widgets were cut. A name can be up to 255 characters, so when the original's name is too long to take (copy), its end is cut to make room.
• Delete: open the Actions menu, click Delete, then confirm with Delete in the Delete Dashboard dialog. This cannot be undone.

Add a widget

Open the dashboard and click Add Widget. On an empty dashboard the button also sits in the middle of the page. The Add Widget dialog walks through its steps with Back and Next. When a step is incomplete, the dialog says what is missing beside Next. The last button is Add widget, and the new widget is added at the end of the grid. A dashboard holds at most 24 widgets. On a full dashboard, Add widget adds nothing and a message says the dashboard is full, so remove a widget first.

Most widget types walk Type, Data sources, Time range and Details. An Alert Summary walks Type, Scope and Details.

Widget types

A widget's type can't be changed later. To change it, remove the widget and add a new one.

Data sources

A data source is a telemetry variable (raw or derived) or a KPI value on one of your assets. Pick sources from your sites and assets, or search for them. A Gauge, Stat Card or Status Indicator shows one source, and choosing another replaces it. A Time Series Chart or Data Table shows several, and the picker shows how many you have selected against how many it allows.

An Alert Summary binds no data source. Its Scope step asks what to Count alerts for: The whole fleet or one site. Alerts raised by control rules are not counted.

Time range

Choose 1h, 6h, 24h, 7d or 30d. The default is 24h. A chart or table covers that window. A single value shows the newest value inside it.

A Time Series Chart or Data Table also takes an Aggregation (Average, Minimum, Maximum, Sum, Count, Median (p50), p95, p99 or Std dev) and a Bucket (1 min, 5 min, 15 min, 1 hour or 1 day). Long ranges can't be read at the finest buckets. The dialog says which buckets a range allows and blocks Next until the bucket fits.

Details

• Title (optional): without one, the widget is titled with its type.
• Refresh every: 30 seconds, 1 minute, 5 minutes or 15 minutes. The default is 1 minute. A widget never re-reads faster than every 30 seconds. An Alert Summary has no refresh setting.
• Gauge scale, for a Gauge only: Minimum, Maximum, Warning at (optional) and Critical at (optional). Values use the source's own unit. The arc fills from minimum to maximum, and the thresholds colour it amber, then red.

Edit mode

Click Edit to change the widgets. Each widget then shows its controls in its top right corner:

• Configure widget reopens the dialog on its Data sources step (or Scope for an Alert Summary). Save with Save changes.
• Decrease width and Increase width change how many columns the widget spans, from one to four. At either limit the control explains why it does nothing.
• Remove widget asks for confirmation in the Remove Widget dialog. Click Remove to remove it.

Click Done to leave edit mode.

Outside edit mode, a widget with data sources keeps one control: Open in Explorer, which opens the same sources and time window in the Explorer.

Save a view from the Explorer

In the Explorer (Insight space, Analyze → Explorer), select the variables you want, then click Save to dashboard in the toolbar. With nothing selected, the button is off and says Select a variable to save this view.

The Save to dashboard dialog asks for:

• Dashboard: choose one, or click New to create a dashboard without leaving the Explorer.
• Widget type: Time Series Chart or Data Table, the two types that hold several sources.
• Title (optional).

Before you save, the dialog states what the widget will hold, for example "Saved as 2 series over the last 7 days, at 1 hour buckets (avg)." When the view can't be stored exactly, it says what it will store instead, before you save:

• A fixed window is saved as a rolling window, because a widget always shows the last hours or days. It gets the shortest widget time range that covers the window, or 30 days for a longer one.
• Raw resolution is saved at a bucket, averaged.
• A bucket a widget can't use for that range is saved at one it can.
• Sources beyond what the widget type holds are named and left out.

Click Save widget to add it to the end of the chosen dashboard.

Reading a widget

A widget shows each value with its state. A value with nothing to doubt carries no marker. The markers are:

• uncertain, beside a KPI value: the value was computed on part of its expected inputs. Hover over it to see what share of them.
• completeness not recorded, beside a KPI value: the engine could not measure input completeness for that window.
• On a chart or table, a KPI series with doubtful points is summarised in the widget header, for example 3 uncertain points or completeness not recorded on 2 points. definition is on rev followed by a number means the KPI's definition has changed since these values were computed.
• Uncertain, Bad or Unknown, beside a single telemetry value: the source flagged the reading, or recorded no quality with it. The value is shown as read.
• Agent-drafted: the KPI's definition was drafted by the authoring agent and saved by a person. The agent proposes, a person approves.

KPI quality markers explain the readings shown; they do not remove uncertain or completeness-not-recorded points from a chart or widget. A visible value is not a promise that only good points contributed.

A telemetry series on a chart or table is graded on its newest reading only, because the chart's data carries no quality. When that reading is not good, the header names the series, followed by newest reading and Uncertain, Bad or Unknown.

The value and its age

A Gauge, Stat Card or Status Indicator shows the newest value in the widget's time range, with its unit. Under it, the widget says how old the value is, for example "2m ago". A value more than a day old shows its date and time instead. Hover over the age to see the exact time. The age keeps counting while the page is open, also between refreshes.

When the newest value is older than the time range, the widget does not show it as its value. It says, for example, No reading for Turbine 1 / Active power in the last 24h (No value for a KPI), and below that Last seen with the value's age. When no reading of a telemetry variable can be found at all, the widget says No recent reading found for followed by the source.

Stale values

A value is stale when it is older than its limit:

• Telemetry: more than 5 minutes old.
• KPI: older than its cadence allows, as in the table below. If the widget can't read the KPI's cadence, it uses the daily limit.

A stale value stays on screen, and its age turns red with a warning icon. A Stat Card also shows STALE. A Status Indicator reads Stale, except for a true/false value, which reads Active or Offline.

How current a chart or table is

Below a Time Series Chart or Data Table, the widget says how current its data is, using the same limits:

• When no source is stale, it shows Latest data with the age of the source that lags most.
• When a source is stale, it lists each stale source by name with latest and its age. A telemetry variable that has sent nothing for 10 minutes or more shows no reading in the last 10 minutes.

When a KPI has more values in the time range than the widget can read, the widget draws the newest ones and says when they start, for example only points from 25 Sep 00:50 on are drawn. Older values are not drawn. Choose a shorter range to see them.

When the browser is offline

While your browser is offline, a widget can't read, and it does not report that as an empty result:

• A widget that already shows data keeps it and adds Waiting for connection. This is the last value read. (This is the last data read. on a chart or table). The age keeps counting, so the value turns stale on time.
• A widget with nothing to show yet says Waiting for connection. and that its sources will be read when the connection is back.

When a read fails

When the browser is online but a widget's read fails, for example while a service is down:

• A widget that already shows data keeps it and adds Could not update. This is the last value read. (This is the last data read. on a chart or table). On a chart where only some sources failed, the note names them. The age keeps counting. On the dashboard page, Retry next to the note reads again at once; in full screen the widget reads again on its own schedule.
• A widget with nothing to show yet says Could not read and names its sources, with Retry.

Full screen

Open a dashboard and click Full screen in its toolbar. The button shows when the dashboard has widgets and you are not in edit mode. The browser goes full screen and shows the widgets under a thin bar with the dashboard's name: no sidebar, app header or dashboard controls, and nothing on a widget opens another page. If the browser does not allow full screen, the same view opens in the browser window.

In full screen the dashboard keeps the rows, order and widget widths you gave it. The layout scales to fit, with spare space around it; widgets keep their own height. A dashboard with more rows than fit scrolls. Text and figures grow with the screen width. Each value has its unit beside it and its source and age below, and values in the same row share one size. A stat shows its measured trend over the labelled time range when that series is not already drawn in a chart on the dashboard.

Power and other measurements whose units support prefixes use three significant figures with an SI prefix. Percentages, temperatures, counts and wind speed keep their unit. Chart axes and legends use the same unit. The first series follows a solid stroke and comparison series use dashed strokes; latest values appear at the line ends on the wall screen when they fit. If a plot has too many series to keep those labels separate, each latest measured reading appears beside its full series name and unit in the legend. Time spacing follows elapsed time, including unequal intervals; the chart names its requested window and the interval shown, with dates when it crosses midnight. A stat source name wraps in full above its age.

A value that is not current says so on its widget: the widget gets an amber outline, and Stale (the value is older than its source normally allows) or Held (the latest read failed or the connection is down, so this is the last value read) appears beside the unit. A widget with no value shows a dash and says what it looked for.

The bar across the top shows, from left to right:

• The dashboard's name.
• Switch dashboard, which lists your organization's dashboards with the one on screen selected. Choose another to show it instead; the view stays full screen. It works with the keyboard too: press Tab to reach it, Enter to open it, the arrow keys to move, Enter to choose and Esc to close.
• The theme: Light, Dark or System. It is the same setting as Theme in your account menu, so a choice made in either place shows in both.
• Exit full screen.
• The status, which says how current the data is. See What the screen says about its data.

After 3 seconds without input, Switch dashboard, the theme and Exit full screen hide, together with the pointer. Moving the pointer or pressing a key brings them back. They stay reachable with the Tab key, and stay visible while one has focus, the pointer rests on them or the dashboard list is open. The name and the status never hide.

To leave, click Exit full screen, or press Esc: when you entered with the Full screen button, Esc leaves full screen and returns to the dashboard, including after you switched to another dashboard. Pressing Esc to close the Switch dashboard list does not return you to the dashboard. Leaving the view any other way, for example with the browser's Back button, also ends full screen.

Put a dashboard on a wall screen

A control-room screen can show one dashboard permanently, with no one at the keyboard.

1. Find the display address. Open the dashboard and copy its address from the browser's address bar, then add /display to the end. It has the form https://<your Fleetera address>/insight/dashboards/<dashboard id>/display.
2. Start the browser in kiosk mode on the screen's computer, for example Chrome or Edge started with --kiosk followed by the display address. Use a normal browser profile that keeps its data. A guest or private window forgets the sign-in when it closes.
3. Sign in once, with an account that has the Viewer or Operator role (see below). If the browser isn't signed in, the display opens the sign-in page first. After you sign in, the browser returns to the display.
4. Set the theme to Dark. Move the pointer so the controls appear in the bar across the top, and choose Dark in the theme control. System follows the computer's light or dark setting, which on many computers changes with the time of day, so a wall set to System can switch on its own.

Opened this way, Esc does not leave the display.

What the screen says about its data

The right end of the bar across the top shows how current the data is, and stays visible when the controls hide. When data is stale or a connection or widget read fails, the status is shown larger and the bar gets an amber line along its bottom edge. Before the first data update, Waiting for data means the dashboard is waiting for its readings; an empty dashboard shows No widgets. Live with the time since the last update means every widget is updating. A count such as 2 widgets could not update means Fleetera answers but that many widgets could not read their data, and No connection means the screen cannot reach Fleetera; both give the time of the last update, and the widgets keep showing the last values they read until the data comes back. After a new release of Fleetera the screen reloads itself onto it within about 15 minutes, and never while it has no connection.

Which account to sign in with

The screen stays signed in, so anyone at its keyboard or touch screen can do what its account can do. Sign it in with an account that has one of these roles:

Recommendations

• On a 4K screen, set the operating system's display scaling to 200%.
• Give each screen its own dashboard. A display shows one dashboard.

Known limits

• Changes to the dashboard itself, such as an added or removed widget, reach a display within 5 minutes. If the dashboard is deleted, the display says This dashboard no longer exists.
• If the screen's sign-in ends, the display opens the sign-in page. After someone signs in, the browser returns to the display. This also applies when the sign-in page appears after reloading the display.
• Where the browser allows it, the display keeps the screen from going to sleep. Otherwise the computer's own power settings decide.