Running tests

Simulate vs your bench

Two run modes with genuinely different meanings. Knowing which question each one answers saves you from trusting the wrong result.

Simulate

Answers: is this sequence coherent?

Runs in the browser. No hardware, no desktop app, no cost.
Compiles the real workflow and plays it: loops iterate, variables sweep, each measurement records a synthetic value.
Gives you a progress bar, the step rail, a streaming log, a loop counter with live values, and a results table.
Catches a loop that never terminates, a measurement outside its loop, a missing settle, a step in the wrong order.

Run on my bench

Answers: does the part meet spec?

Runs the generated Python on your machine, through the desktop app, over VISA.
Real instrument traffic, real readings, real timing.
Limits evaluated against actual measurements at record time.
The only mode whose result means anything about your hardware.

What simulation deliberately does not do

It does not model your part. The values are synthetic. A simulated reading of 3.30 V is not a prediction.
It does not model your instruments. Timing, ranging, and settling behaviour are real instrument properties that only a real instrument has.
It does not prove a connection. An offline instrument simulates perfectly.

When to use which

Situation	Mode	Why
You just built a new test	Simulate first	Cheapest way to find a structural mistake before energising anything.
Hardware has not arrived	Simulate	The whole workflow, plan, and schematic can be finished and reviewed in advance.
You changed a set-point	Straight to bench	The structure is already proven. You want the reading.
You are demonstrating the flow	Simulate	It is repeatable and it cannot damage a DUT.
You need a number for a report	Bench	A report is built from real rows. There is no other source.

Asking for a run in chat

Ask the agent to run the test and it runs it on the bench, immediately, and says so in the present tense. It only simulates if you use the word yourself: simulate, preview, dry run.