Analyze

Start with the question you need to answer. In Analysis, you can ask the assistant to build a notebook using your platform data, then inspect and refine the investigation. A notebook keeps the analysis code, explanations, and results together so you can revisit how you reached a conclusion.

​

Debug a growth

Suppose a growth behaved differently than expected. You might have noticed a change during the run, or found an unexpected result in later characterization. Give the assistant the growth’s name and your observation:

Growth [name] behaved differently than expected: [describe what you observed]. Investigate the available data, identify changes that might explain it, and build a notebook showing the supporting evidence. Explain what remains uncertain and what I should check next.

Open Analysis, select New Notebook, enter your request, and click Create Notebook. See Creating a Notebook for the full instructions and other ways to start.

​

Inspect the evidence

The analysis depends on the question and the available data. It might include timelines, comparisons, summary tables, or an explanation of a gap that needs more information. Check that it uses the intended growth and sources. Look at the units, time alignment, selected intervals, and assumptions. A change in two signals at the same time can suggest a connection, but does not establish a cause. Ask the assistant to explain any step you cannot assess. You can inspect and edit the notebook’s Python cells as well as discuss the analysis in the Agent panel. See Working in a Notebook.

​

Refine the investigation

Follow up based on what you find. For example:

Does that change coincide with a recipe transition or instrument event?

Is this unusual compared with other growths under similar conditions? Explain which growths are suitable references and why.

What evidence would distinguish a process change from a measurement problem?

Summarize the strongest explanations, the evidence for each, and the next checks that would distinguish them.

You can narrow the time range, add context, challenge an assumption, or pursue another explanation. These are examples of how to investigate, rather than a required sequence for every growth.

​

Save and return

Click Save Changes to preserve your edits as a notebook version. Reopen the notebook from the Analysis sidebar to review saved results or continue editing. See Saving, Versions, and Access for the distinction between saved results and a live editing session.

​

Related tasks