Cleaning, filling and analysing data with AI in Grid
With your own AI account, Grid by Zaweba can clean up data, analyse a selection, suggest a chart, fill down from examples, translate cells and make sample data.
Tidying and making sense of data takes up a large part of spreadsheet time. If you've connected your own AI account, Grid by Zaweba has a set of AI actions aimed squarely at that work. Each is careful about what it changes: replies that would fill cells are previewed first, every write is a single undo step, and nothing ever overwrites a locked cell on a protected sheet.
Clean up data
Clean up data handles the jobs that eat an afternoon: trimming stray spaces, putting names into proper case, standardising dates as dd/mm/yyyy, and splitting two-part columns such as a full name into first and last. Select the messy column, run the action and look over the preview before writing it into the sheet.
It's a good first step after pasting from a website or importing a CSV from another system, before you build lookups or pivot tables on top of the data.
Analyse selection
Analyse selection gives a written summary of the selected data, the trends in it, and any outliers with their cell addresses. The addresses are the useful part: you can go straight to the odd values and decide whether they're mistakes or genuine. Treat the summary as a starting point for your own look, then confirm anything important with a formula or a chart.
A related action, Suggest a chart, helps when you're not sure whether a line, column or scatter chart suits your data, by recommending one for the selection. You still make the chart yourself, and F11 is the quick way to try it.
Fill down from examples
Fill down from examples learns a pattern from a few rows you've filled by hand and continues it down the column. Type the first two or three results, such as initials from names or a code built from two fields, and let the AI follow the pattern. Check the preview, because a pattern can be read more than one way.
Translate cells and generate sample data
Translate cells... translates the selected text into another language, which suits product lists, menus and short labels. Have a native speaker check anything customer-facing. Generate sample data makes realistic test rows, handy for trying out a template, a pivot table or a chart layout before real data arrives, or for teaching without using anyone's personal details.
Safe by design
Every action that would write cells shows a preview of the table first. Only when you choose Write into cells does anything change, and the whole write is one undo step, so a result you don't like is gone with one Ctrl+Z. If the target cells are locked on a protected sheet, they aren't touched.
Only the context the request needs is sent, and only when you ask: the selection as text up to a set size, plus details such as the sheet name and header row. Select just the columns you need.
When to use a formula instead
For a job you'll repeat every week on fresh data, a formula is often better than an AI action, because it updates itself. TRIM, the case functions, TEXTSPLIT and Remove Duplicates handle much routine cleaning without any account. The AI actions shine on one-off, irregular jobs where writing the formula would take longer than checking a preview. Both approaches live side by side in Grid by Zaweba, and you choose per job.
