Import your data

Supported sources at launch

CSV / TSV files

Upload one delimited-text file — .csv, .tsv, .tab or .txt — or paste its contents, and it becomes a versioned dataset in your workspace. Working in Excel? Save the sheet as CSV first; CollabHub doesn’t read binary .xlsx workbooks on the way in (it does write them on the way out).

One file per import job, one dataset out. For a quick single dataset you can skip the importer entirely and drop the file straight onto the Datasets → Upload dataset page — the importer is for when you want the tracked job, the dry-run preview, and the verification report.

Benchling

A live API connection that migrates your Benchling tenant — notebook entries, datasets (assay results), protocols, custom entities, and file blobs — page by page, with real file contents, not stubs. You’ll need a Benchling API key (in Benchling: Settings → API keys); the import won’t start without one, so placeholder data can never end up in a real workspace.

That’s the whole launch list. If you see another tool named anywhere else, it isn’t connected yet — tell us which one you need below.

The dry-run preview

Every import starts with a preview that runs the same extraction and mapping as the real import, but writes nothing: no records, no files, no job. You see:

• how many records would be imported, by type;
• what won’t map, with a sample of those records and the reason for each;
• how many records fall beyond your plan’s per-job record limit, if any (see below);
• a fidelity estimate and any warnings, up front.

The preview’s counts equal what the real run will store. Running an import again is safe too: records that already made it in are recognized and skipped, and the preview only counts what’s new.

How many records come in at once

Each import job has a per-job record limit that scales with your plan (the free tier imports 250 records per job; paid tiers run into the thousands — see pricing for the exact numbers). It’s a resource limit, not a paywall on your data.

Nothing is ever dropped silently: records beyond the limit are recorded as cap-skipped with their full raw payload retained, they’re counted in the preview before you commit, and re-running the import picks them up.

How an import runs

1. Open Import & Export (sign-in required) in the sidebar and choose the destination workspace.
2. Pick a source — CSV/TSV (attach or paste your file) or Benchling (paste your API key).
3. Run the preview and check the plan: counts, unmapped records, cap-skipped records, warnings.
4. Start the import. It runs in the background while you keep working.
5. Follow the job until it completes, then read the verification report: an honest account of what was imported and at what fidelity.

To run an import you need to be able to create content in the destination workspace (Owner or Editor). Every imported record gets a timestamped priority proof when it arrives, the same as work created in CollabHub. See Getting started.

Coming from another tool?

Most tools can export to CSV, which the importer takes today. Tell us which source you’d like a direct connection for; it helps us decide what to enable next. And whatever you bring in, you can always take back out: export is open, on every tier, forever.