Built-in Tools

The IronFlock AI assistant includes four built-in tools that are available in every project. These tools work with any data collected by any app — you don’t need to install anything special to use them.

Data Explorer

The Data Explorer gives the AI direct access to the time-series data stored in your project’s database. It can query live readings, historical trends, aggregations, and comparisons across devices.

What It Can Do

Query current and historical sensor values
Compare metrics across multiple devices or time periods
Calculate averages, min/max values, and trends
Filter data by device, time range, or data table
Save a calculation as a custom transform — a named, self-refreshing view that widgets and later questions can build on

Example Prompts

You: What was the average temperature across all devices last week?

IronFlock AI: The average temperature across all 12 devices from March 31 to April 7 was 22.4 °C. Device gateway-north had the highest average at 26.1 °C, while sensor-cold-room had the lowest at 4.2 °C.

You: Show me the power consumption for device plc-line-3 over the past 24 hours.

IronFlock AI: Here’s the power consumption for plc-line-3 over the last 24 hours:

Peak: 14.7 kW at 10:32 AM
Average: 8.3 kW
Minimum: 0.4 kW at 3:15 AM (idle period) Total energy consumed: 199.2 kWh.

You: Which devices have reported error codes in the past 3 days?

IronFlock AI: Three devices reported error codes in the past 3 days:

press-station-2 — Error E-1042 (hydraulic pressure warning) on April 5 at 14:22
conveyor-belt-7 — Error E-3001 (speed deviation) on April 6 at 09:15
press-station-2 — Error E-1042 again on April 7 at 11:03

press-station-2 has had a recurring hydraulic pressure warning — this may need physical inspection.

You: I need overall equipment effectiveness per shift for line 3. Nothing in the data holds it.

IronFlock AI: Right — availability, performance and quality live in three different tables, so I computed them together and saved the result as a custom transform called oee_per_shift, refreshed every 15 minutes. It returns one row per shift with availability, performance, quality and the OEE figure itself. You can now bind widgets to it like any table, and I can build them for you.

The Widget Generator creates live and historical chart widgets directly inside the AI conversation. Instead of just describing data in text, the AI can render interactive visualizations that you can inspect and interact with.

What It Can Do

Generate line charts, bar charts, and gauges from project data
Show live-updating widgets that reflect current device readings
Display historical data over custom time ranges
Combine multiple data sources into a single chart

Example Prompts

You: Show me a chart of humidity over the past week for all greenhouse sensors.

IronFlock AI: Here’s a line chart showing humidity readings from all 4 greenhouse sensors over the past 7 days.

[An interactive line chart appears in the conversation with one series per device, showing humidity values over time.]

Sensor gh-zone-2 shows a notable spike on April 4 — humidity reached 94%, which may indicate a ventilation issue.

You: Create a gauge showing the current battery level of device remote-sensor-5.

IronFlock AI: Here’s a live gauge for remote-sensor-5:

[A gauge widget appears showing 34% battery level, color-coded orange.]

The battery is at 34% — below the 40% threshold. You may want to schedule a replacement.

You: Can you chart the vibration data from motor-pump-1 for the last 48 hours?

IronFlock AI: [A line chart appears with vibration amplitude over the last 48 hours.]

There’s a clear increase in vibration amplitude starting around 6 AM yesterday. The baseline was 0.8 mm/s but it’s now consistently above 1.4 mm/s. This pattern often indicates bearing wear.

Dashboard Generator

The Dashboard Generator helps you create and configure full dashboards — complete boards with multiple widgets arranged in a layout — using natural language. Instead of manually dragging widgets and configuring each one, you describe what you want and the AI builds it.

What It Can Do

Create new dashboards from a description
Add and configure widgets on existing dashboards
Set up data sources, chart types, colors, and labels
Lay out widgets in a grid arrangement

Example Prompts

You: Create a dashboard for monitoring our cold storage facility. I need temperature and humidity for each room, plus a table showing the last 10 alarm events.

IronFlock AI: I’ve created a new dashboard called Cold Storage Monitor with:

4 temperature gauges — one for each cold room (Room A through D), showing live readings with color thresholds at -18 °C (green), -15 °C (yellow), and -10 °C (red)
4 humidity line charts — one per room, showing the last 24 hours
1 alarm table — displaying the 10 most recent alarm events across all rooms

The dashboard is now available in your project. Would you like me to adjust the layout or add anything else?

You: Add a line chart to the production overview dashboard that shows OEE percentage for each production line over the past month.

IronFlock AI: Done. I’ve added a multi-series line chart to the Production Overview dashboard showing OEE for all 3 production lines (Line A, Line B, Line C) over the past 30 days. Each line is color-coded and the chart includes a legend.

You: I want a simple dashboard with just two big numbers: total units produced today and current scrap rate.

IronFlock AI: Created a dashboard called Production KPIs with two large value widgets:

Units Produced Today: 1,247 (updates live)
Current Scrap Rate: 2.3% (updates live)

Both are centered on the dashboard with large font sizes for visibility on shop floor screens.

App Files

Apps store files as well as readings: reports they generate, images they capture, exports they write out. The assistant can browse that storage, find a file by its path, and read what is inside it.

What It Can Do

List the files any app in the project has stored
Find a file by part of its path, or by several words that appear anywhere in it
Read text files — CSV exports, JSON, YAML, logs, configuration
Open a PDF and read it — a generated report, a datasheet, a signed-off procedure, even a scan
Look at images an app has captured, and show them to you in the chat
Report a file’s size, type, and when it last changed

Two things are worth knowing. Searching matches the file’s path, not its contents — asking for a word that only appears inside a document finds nothing. Several words are matched as all of these appear somewhere in the path, in any order, so “shift report” finds reports/2026-08-shift-a.csv. And a PDF is opened whole rather than converted to text: the assistant reads the actual pages, so charts, tables and scanned paperwork come through the way they look, and the document appears in the chat for you to open as well. PDFs above about 6 MB are not opened — the assistant tells you the size instead. Archives, spreadsheets and other binary formats still cannot be read in the chat; the assistant will tell you what the file is and point you to the data browser to open it.

Files stay subject to your own permissions: the assistant sees exactly what you would see yourself, and nothing more.

Example Prompts

You: What files has the quality-inspection app stored?

IronFlock AI: The Quality Inspection app has 143 files across two storage areas:

reports/ — 118 CSV files, one per shift, the newest from today at 06:00
frames/ — 25 JPEG images captured on failed inspections

You: Open the newest shift report and tell me the scrap rate.

IronFlock AI: From reports/2026-01-31-shift-a.csv (412 KiB, written today at 06:00): 1,204 units inspected, 31 rejected — a scrap rate of 2.6%. The most common reject reason was surface-scratch (19 of 31).

You: Show me one of the failed inspection images.

IronFlock AI: Here’s frames/2026-01-31T05-42-11.jpg, captured at 05:42 this morning:

[the image appears in the chat]

The scratch is visible along the upper-left edge of the part.

You: The maintenance app filed a PDF service report last night. What does it say about the pumps?

IronFlock AI: From service-reports/2026-01-30-pump-station.pdf: the technician replaced the mechanical seal on pump 2 and noted that pump 1’s bearing housing was running warm, with a re-check recommended after 500 operating hours. The vibration trend on page 3 is flat for pump 2 and rising slowly for pump 1, which matches the note.

You: And the report from the week before?

IronFlock AI: service-reports/2026-01-23-pump-station.pdf is a scan of a signed paper form, but it reads fine: same technician, routine inspection, no parts replaced. The handwritten note at the bottom says “check seal next visit” — which is what happened on the 30th.

You: Find anything mentioning calibration.

IronFlock AI: I can only search file paths, not what’s inside the files, and no path in this app contains “calibration”. There is no content search, so here is what the app actually has: 118 shift reports under reports/ and 25 inspection images under frames/ — nothing else. If the calibration values are written into a shift report, tell me which shift and I’ll read it.