The short answer
If operators already write readings into a spreadsheet each shift, or someone exports a historian to Excel, that history is the fastest start: InduQT learns each machine’s normal from it before any sensor is fitted. In Claude with the InduQT connector you ask for an upload link, drop the file in, and say which columns are the time and which machine they belong to. New sensors (see the other guides) then continue the same signals.
- Months of history analysed in minutes, no sensors needed
- Past breakdowns found in the data and used to recognise the next one
- A baseline for the sensors you add later
What to buy
One row per reading time, one column per value. Several machines can be in one file with a machine column. Maker: —.
Prices are typical street prices in US dollars when we wrote this, not quotes; buy from the maker or a local distributor. InduQT does not sell sensors and is not paid by any maker named here.
Get the file ready
- One row per reading, a time column (date and time, or separate date and time columns) and one column per value.
- Numbers as numbers: “12.4”, not “12.4 bar”. Put units in the column header instead.
- Several machines? Add a column with the machine name, or one sheet per machine.
- Write down the dates of any breakdowns you remember; they help InduQT learn what bad looks like.
| Date | Time | Machine | Discharge pressure (bar) | Oil temp (°C) | Load hours |
|---|---|---|---|---|---|
| 2026-09-01 | 06:00 | compressor-1 | 7.4 | 68 | 15210 |
| 2026-09-01 | 14:00 | compressor-1 | 7.2 | 71 | 15218 |
| 2026-09-01 | 22:00 | compressor-1 | 7.3 | 70 | 15226 |
Check it and load it with Claude
I’ve attached our compressor shift log. Before I load it anywhere: tell me the date range, which columns are numeric, any obvious typos (values 10× too big, dates out of order, blanks), and summarise how each value has trended.Using InduQT, give me an upload link for an Excel file. After I upload it, import it into Plant 1: the time is the Date and Time columns, the machine is the Machine column, every numeric column is a signal. Then tell me what InduQT found in the history.No Claude? In the InduQT app, open the machine and choose Upload history (CSV or Excel), or use the free troubleshooter to analyse a trend file once. Scheduled imports from a shared drive or Google Sheets are described in the Excel and Google Sheets connector docs.
What InduQT does with it
- Learns normal ranges and operating modes for each machine from the history.
- Finds past events in the data (sudden changes, slow drifts) and asks you whether they were real problems; your answers make the next warning better.
- Keeps going when you add a sensor that publishes the same signal names.
Frequently asked questions
How much history do I need?
A few weeks gives a first baseline; six to twelve months including a breakdown or two is better.
We have a home-grown system with a database. Can InduQT read it?
Yes: export to CSV on a schedule, or have the system post to the InduQT ingest API (one HTTPS call). We also connect in-house systems for customers on annual plans; see free setup.