Measurement data as a report with key figures and a chart.
Drop the CSV, press Analyse now, download the report. Key figures for each channel, one chart and the observations in one file.
As of 15 October 2026
Upload measurement data and analyse it
View the task for Biwak
Analyse the measurement data and create a standalone HTML analysis. Read the columns and units from the header row. For each channel, calculate the minimum, maximum, peak-to-peak value, mean, RMS and dominant frequency, with units. Plot all channels in one shared chart against time, with a legend, grid and axis labels. Show missing or invalid values as data errors and do not invent any measured values. Summarise the most important observations in three to five points and attach the raw data as a collapsible table. Without external libraries, save the result as auswertung.html.
100 one-time trial credits, with no payment details.
- Processing happens in the EU
- No training on your content
- Sign in with an email code, passkey or Google
To analyse measurement data as CSV with Biwak, drag the file into the window and press Analyse now: a prepared task goes along. Biwak is an AI agent from Munich. It reads columns and units from the header row, calculates the minimum, maximum, peak-to-peak value, mean, RMS and dominant frequency for each channel, and creates a report as an HTML file, with a chart, observations and the raw data. It should show missing or invalid values as data errors rather than replace them. To try it out, there is a one-off grant of 100 trial credits. Processing happens in the EU.
How to analyse measurement data.
-
01
Drop the CSV
Drag the export from your oscilloscope or data logger into the window, or choose it. Up to eight files per task, each up to 100 MB.
-
02
Analyse now
A prepared task goes along with the file: key figures for each channel, one shared chart, three to five observations, raw data to expand. You watch Biwak at work.
-
03
Check, adjust, download
If you need other key figures, a limit value or a second chart, just add it in the chat. The report appears as a card in the chat; one click opens the preview, where you download it.
501 rows of measured values become one report.
- Zeit_s,Kanal_A_V,Kanal_B_V
- 0.00000000,1.03601,1.66859
- 0.00001000,1.12512,1.70438
- 0.00002000,1.31250,1.67988
- 0.00003000,1.34088,1.72122
- 0.00004000,1.55133,1.74407
- 0.00005000,1.61420,1.68429
- 0.00006000,1.67914,1.73188
- 0.00007000,1.76350,1.62915
- …
- Channel A · peak-to-peak
- 3,67 VMean 1,00 V · RMS 1,62 V
- Channel B · peak-to-peak
- 2,29 VMean 0,65 V · RMS 1,17 V
- Dominant frequency
- 1,00 kHzMeasurement window 5,00 ms
Channel AChannel B0 to 5,00 ms · −2 to 4 V
Recalculated when this page was built: 5 rising crossings through the mean of channel A at an interval of 1.00 ms, so 1.00 kHz
- Channel A is almost sinusoidal, with a small harmonic and noise content.
- Channel B is square-wave in shape, phase-shifted relative to channel A, with brief overshoots on the edges.
- Five periods lie entirely within the measurement window; the key figures are estimates from the data.
- The analysis is a single HTML file without external libraries: it opens in any browser and can be shared.
Invented measurement data (501 samples, two channels) and its expected result. What you see is the expected key figures and the data itself, not a model run. To check for yourself: Measurement data (CSV) · Expected analysis (HTML)
Why the report takes longer than the measurement.
The measurement is in the bag within seconds. Then come spreadsheet work, formulas, a chart you have to adjust, and a text that says what you can see.
Every set of measurement data is similar and yet never quite the same: other columns, other units, one more channel. A fixed template rarely fits, and building a new one is not worth it for a single measurement. Biwak writes the analysis for exactly this file and explains how it did the calculations.
How Biwak goes about it
- Biwak reads the header row and recognises the time column, channels and units.
- It calculates the key figures from all rows, not from a sample.
- It is instructed to show missing or invalid values as data errors and not to invent any measured values.
- It describes noise and phase relationship cautiously as estimates, and adds the observations.
The key figures are calculations from your data. They do not replace a calibrated measurement or test software.
Then keep working.
The same chat can do more. Sentences you can write right afterwards:
- “Mark every place where channel B is above 1.5 V, and give the duration.”
- “Compare messreihe-vorher.csv with messreihe-nachher.csv and show the differences.”
- “Put the key figures of all measurement files in this task into one Excel table.”
- “Write two paragraphs for the test report: set-up, result, anything noteworthy.”
Where your measurement data stays.
- In the browser
- You upload the files to your workspace on the Biwak server in Paris; tasks and answers are stored in our database in Frankfurt am Main.
- In the app
- With Biwak for macOS and Windows, the files stay in your folder on your computer. The task, conversation context and the file contents used go to the model.
- Processing
- Processing happens in the EU: on Microsoft Azure in the EU Data Zone.
- Training
- None on your content.
- Access
- Only for a stated reason, with every access logged. While the browser workspace is a preview feature, quality review is one such reason; you can object at any time.
- DPA
- For businesses, Biwak’s data processing agreement applies; there is then no quality check. Even so, measurement data that allows conclusions about people should only be used after your own assessment.
What it costs.
After email confirmation, you get 100 one-time trial credits, with no payment details. After that, a subscription applies: Base Camp for €19 a month with 1,500 credits or Rope Team for €99 with 9,000 credits, cancellable monthly. Final prices, no VAT charged under § 19 UStG (German small-business VAT exemption).
How many credits an analysis needs depends on the file size and the performance level; you see the usage after every task.
When you do not need Biwak for this.
If your measuring instrument or laboratory software already performs the analysis, for example frequency and peak-to-peak directly on the oscilloscope, use those values. For calibrated tests and evidence, your laboratory’s procedure applies.
Biwak is worth it when a raw data file is to become a report that others read: with a chart, key figures and a few sentences on what stands out. And when the columns look a little different every time.
Frequently asked questions
Which files can I analyse?
Tables with one row per measuring point: CSV or TXT with comma, semicolon or tab separators, or an Excel file. Biwak reads columns and units from the header row; if they are missing, write in the task what each column is.
Does Biwak calculate the key figures itself, or does the model estimate them?
Biwak is instructed to calculate the key figures from all rows rather than estimate them, and not to invent any measured values. An instruction is not a guarantee: spot-check a key figure by recalculating it before you pass it on.
What happens with gaps or outliers?
The prepared task says: show missing or invalid values as data errors, and do not invent any measured values. If you want outliers handled differently, for example excluded or marked, just add it in the chat.
Why HTML rather than PDF or Excel?
An HTML file carries the chart, key figures, text and raw data in one file, opens in any browser and can be printed. If you also want the key figures as an Excel table, say so in the task.
How large can the measurement data be?
Up to 100 MB per file, up to eight files per task. With very long series, a good chart does not plot every point anyway; add in the chat whether Biwak should thin out the data or show an excerpt.
Do I need an account?
Yes. You sign in with an email code, passkey or Google account; the chat then starts right on this page. You can already drop the file in before signing in. After an hour the page no longer uses it and deletes it on the next visit; if you leave the page or reload it, it is deleted – except on the way to sign-in.
What does it cost?
To try it out, 100 one-time trial credits with no payment details, then a subscription from €19 a month. You can see the usage after every task.
Responsible: Hermann Hampel · Created with AI assistance