Microsoft taught Copilot to act on a folder of files: attach it, describe the job, watch it run. Your Airtable base is not a folder, so the cleanup Copilot Actions could finish in one pass has nowhere to run. The only bulk routes write straight to a base your views, automations, and interfaces react to the instant a cell changes.
Scratch pulls the base down as that folder. Copilot edits every record on your laptop, and Scratch holds each change as a diff. The last 1%, what writes back, stays yours. Nothing reaches the live base until you approve it, row by row.
How it works
- Scratch pulls your base into files. A base or a single table becomes a folder on your laptop, one file per record, every field laid out to edit.
- Copilot edits the records. Open the Copilot app on Windows and attach your Scratch folder to Copilot Actions. Describe the cleanup in plain English and it works through the table in its own agent workspace while you watch. Normalize every company name and fill the empty category cells from the description. Copilot works the files, never the live base.
- You review every diff and publish. Back in the Scratch app, each changed field sits beside the original, word by word. Approve what holds up, and Scratch pushes only those records through the Airtable API.
What people use it for
The cleanup that never happens because doing it by hand means opening every row:
- Normalize inconsistent company or contact names across the table.
- Fill empty category, status, or tag cells by reading each record's own notes.
- Cut every long summary down to a single line.
- Turn freehand text into the exact single-select your views expect, without inventing new options.
- Backfill the linked-record references someone left blank.
Run it on a few records to feel the loop, then let it take the other 5,000.
Why not an MCP server?
An Airtable MCP server or extension wires an agent straight to your live base. One pass rewrites every record at once, and your views, automations, and interfaces all react before you have seen a thing.
Scratch gives Copilot the same reach against a local copy instead. The write-back is lifted out and handed to you. Copilot can change anything in the folder; only you can commit it. On a base your team runs on, that gap is the whole point.
What Copilot edits in Airtable
- Long-text and rich-text fields
- Single-line text, including titles
- Single-select and multi-select tags
- Date, currency, and number fields
- Attachment captions and linked-record references
Formulas, rollups, lookups, autonumber fields, and the created and last-modified timestamps pass through untouched. Validators read your option sets, so Copilot cannot coin a select value that would break a view. The full list lives on Scratch for Airtable.
Questions people ask
Is this a Copilot plugin or an MCP server?
Neither. A plugin or an MCP would hand the agent the write button. Scratch keeps it. Copilot Actions only ever sees files in the Scratch folder, and writing back is a separate step you approve, one record at a time.
Will it break my formulas, views, or automations?
No. Formulas, rollups, lookups, and autonumber fields are never editable, automations are never exposed, and validators stop Copilot from inventing a select value that would knock out a view.
Can I roll a change back after it writes?
Yes. Scratch keeps the original beside the rewrite, so every written record reverts per row. You decide which version stays.
How is this different from a CSV import or a script?
A CSV import and a script both write straight to the live base with no diff and no per-record approval, and a find-and-replace does only what you spelled out. Copilot handles the records a rule cannot, and Scratch still holds every change for review before it lands.
Can it run across a 5,000-row table?
Yes, that is the use case. Run it on a handful of records to feel the flow, then describe the cleanup for the whole table.
Do I need to be technical?
No. Copilot Actions takes a plain-English instruction, so there is no terminal and no editor. You attach the Scratch folder, describe the change, and approve the diffs in the Scratch app. You do need the Copilot app on Windows.
See it on your own base
The fastest way to trust it is to watch it run on your data. See it run on your Airtable base →, or download Scratch free and take the first pass yourself.