Forge is now OpalSpan, and Whisper is now Mech. The demonstration below shows the current OpalSpan/Mech interface.
A support log can contain the answer you need, buried among rows of timings or repeated events. Mech can now use Python to work through that data, calculate a finding and return a useful file in the same conversation.
You supply the file and the question. Mech handles the code and the calculation, so you can stay with the support case.
From a timing log to a summary you can use
In the fictional example below, a log contains six runs of two scheduled jobs. The request is simple:
“Use Python to compare average duration by job in job-timings.log. Return a summary CSV.”
Source
Current shared Mech UI, generated 13 October 2026 with fictional timing rows and scripted calculations. No real model/device work or hosted Python execution. The supplied six runs support the stated averages, not a diagnosis of their cause. Generation provenance.
Earlier published recordings: desktop · phone. Historical versions: desktop · phone.
The result: Backup averages 20 minutes; Scan averages 5 minutes. Mech returns the comparison and a job-summary.csv file in the conversation’s Files panel. You have a concrete lead to investigate and a small result you can download.
Useful when the next step needs a calculation
Ask Mech to count recurring entries in a log, compare durations or turn structured evidence into a summary. A focused question makes the result easier to check against the original file.
Python runs in an isolated hosted environment, with the selected conversation file as its input. You do not need to install Python on your workstation or the customer’s device. Findings and generated files return to the private conversation; scratch files follow OpalSpan’s normal expiry rules.
Try it with a small text or log file you are allowed to use. Attach it in Mech, name the calculation you want and ask for a downloadable summary.