Leave Airtable without leaving your data

Tables, fields and records come across with their types mapped — text, numbers, currency, percent, ratings, checkboxes, selects with their options, and dates including their times. Linked-record fields are rebuilt as real RowFold relations by matching Airtable's own record ids, so 'this order belongs to that customer' survives the move exactly. Attachments are downloaded and stored in RowFold's own storage during the import, which matters because the file links in an Airtable export expire within hours. Formula, rollup and lookup columns import as the values they last computed, clearly marked — rebuilding the computation is a few minutes in RowFold's own formula and rollup editors, and the import report lists every column that needs it.

Yes — that is the point of the sync. After the first import you can put the base on an hourly or daily mirror: new records appear in RowFold, changed records update, and the source wins on the columns the import mapped. Columns you add on the RowFold side are never touched, records you post through an approval are sealed against it, and deletions never propagate — a machine is not allowed to remove your data here. Your team migrates one workflow at a time while the numbers stay current in both places, and you turn the mirror off when nobody has opened Airtable in a month.

A few thousand records land in a couple of minutes. Larger bases run in the background — the import commits as it goes and shows live progress, so a base with tens of thousands of records and a few gigabytes of attachments finishes on its own while you do something else, and a network hiccup partway costs a retry rather than the whole run. Airtable's API is rate-limited per base, and the import paces itself under that limit deliberately: a slightly slower import that arrives complete beats a fast one that got throttled into silence half-way through.

Automations don't port — no migration tool can honestly claim otherwise, because they are scripts against Airtable's own model. The practical path is the one that works: move the data first, run both tools through one overlap period, and rebuild the handful of automations that matter in RowFold's own editor, which covers the common shapes — when a record is created or changes, on a schedule, when a webhook arrives, then update records, send email, post to a webhook. Most teams discover they rebuilt five of their forty automations and never missed the rest; the overlap period is what makes that discovery safe.

RowFold is one number: a flat monthly price per editor, with unlimited free read-only viewers, no record tiers, no plan ladder, and a metered AI allowance included with every editor seat. There is nothing else on the invoice. We deliberately don't print Airtable's prices on this page — they change, and your own bill is the honest source — so the calculator above takes your seat count and per-seat price from your invoice and shows the two side by side, including the case where a small all-editor team would genuinely pay less where they are. Whether the number is smaller matters less than being able to compute it in your head.

There are no per-plan record tiers to upgrade through — a table holds your data, not a quota. One honest cap exists on the importer itself: a single import run brings across up to a hundred thousand records per table and says so plainly in the import report if a table was larger, rather than trailing off silently. Day-to-day use — filtering, sorting, grouping, rollups across linked tables — is engineered against hundreds of thousands of records, and the pricing page and trial make no distinction by volume. If your base is big enough to test that claim, the thirty-day trial is the honest way to test it.