Twenty CRM Strawberry

Automate Twenty CRM, custom objects included

A companion lists every object and field in your workspace before it writes, so “add these six renewals and link them to the right companies” works on an object that exists nowhere except your instance.

Your workspace has a Partners object and a Renewals object you invented on a Tuesday. Every connector you try lists People, Companies and Deals and stops there.

A companion reads your schema first: every object type, every field, the data types and the select options, custom ones included. Then it queries and creates in the vocabulary of your own instance. Twenty is a native integration, connected with an API key, and it accepts your self-hosted URL.

Custom objects a connector has never heard of

The discovery step is the whole trick.

A companion lists every object type in your workspace, standard and custom, then every field on any of them: label, data type, the options on a select field, and whether you added it.

Only then does it compose a query or a create, and it shows you what it intends to write in your own vocabulary before it runs.

  • Pipeline stages read from the workspace rather than assumed, because admins rename and reorder them.
  • A fan-out search across people, companies, opportunities, notes and tasks when the name you were given is ambiguous.
  • Attachments, favourites and webhooks in the same surface, so the tidy-up jobs are not a separate project.

How do you import 200 rows without creating 200 duplicates?

By checking first. A dedicated duplicate check looks up each candidate before it is created (people by primary email, companies by domain or name, opportunities by name) and a combined tool creates only what is genuinely missing, optionally attaching a note in the same call.

Then by batching. Records create up to sixty per call, so a 200-row import is four round trips rather than two hundred, and every new id comes back in one response for the linking work that follows. You see the plan before the batch runs.

Undo a bad write in Twenty

Deletes in Twenty are soft by default, and the integration exposes a restore tool for any object type. The worst realistic outcome of a bad instruction is an afternoon of restores.

It has a friendlier use too. When an old contact re-enters the funnel, reviving the archived record keeps the history attached instead of starting a clean row.

What can a companion do that a sync script cannot?

Look at the world.

“Add the companies from this newsletter that actually sell to hospitals.” “Find which of these forty accounts changed their pricing page since we quoted them.” Neither is a filter, and both are the step that costs you an afternoon.

A companion reads all forty pages, applies your bar, and writes the result into your instance with the reason attached. It is data extraction wired into a CRM instead of a spreadsheet, scheduled like the rest of your operations work.

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide

Frequently asked questions

Yes. The connection takes an API key and accepts your own instance URL, so a self-hosted Twenty is a first-class case. The tools then run against your instance.

Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed · Canonical facts for AI agents

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide