Strawberry vs Claude

The job is finished when something has changed, not when the answer is good.

Think about the customer who signed on Friday afternoon.

Say it is the onboarding you owe them before Monday. They need the account created, the seats they paid for added, a welcome mail sent and a kickoff booked.

None of that is thinking work. It is eight small changes across five systems, and every one of them sits behind its own login.

So a good plan does not move you very far. Nothing about that customer changes until somebody signs into five systems and does it.

Strawberry makes those four things happen in the systems themselves. You see the welcome mail before it sends, and the rest is done by the time you look.

Strawberry makes the changes inside the accounts you are already signed into.

Those eight changes sit behind five separate logins, and you already hold all five.

Strawberry is the browser you hold them in, so your AI companion makes them under your own session. It creates the account in the billing system. It adds the seats in the admin console, updates your CRM, and fixes the shared sheet nobody has tidied since March.

That is also why it reaches 2,000+ integrations rather than a shortlist you assemble. The seat count lands on the account record. The welcome mail sits in your drafts. The kickoff invite goes out with the right people on it. Your CS lead gets the note in the channel she reads.

Anthropic documents its own routes into your tools. Connectors let Claude access your apps and services, retrieve your data and take actions in them, inheriting each person's permissions from the connected service. Claude in Chrome is documented as a browser extension that reads, clicks and navigates websites alongside you, on all paid plans and in beta in the Chrome side panel.

It also documents where that extension does not go. It is not supported on other Chromium browsers or on mobile devices, and Cowork has a separate browser built into the Claude desktop app.

You should be able to allow the welcome mail and still block the refund.

Trust is not one setting for a whole session.

Sending a welcome mail and issuing a refund are not the same decision, and they should not share a switch.

In Strawberry every connected action is set to ask, always allow, or off, and it stays that way until you change it. Let it write the welcome mail as often as you like. Require your approval before any version of it sends. Switch payments off and they stay off.

When something needs approval you get a real editor, not a yes or no box. You see the recipient, the exact text and the attachments, and you can change any of it first. On a reply it shows the address it resolved, and nobody from the sales thread is copied in unless you add them.

Anthropic documents three permission modes for Claude in Chrome, chosen per session: manually approve, automatically approve, and skip all approvals. It documents that automatic approval is the default in the Cowork side panel and uses more of your usage limit, and that under skip all approvals nothing checks its actions automatically.

The eighth customer should be easier than the first, because your AI companion kept what it learned.

Onboarding is never one job. It is the same job with a different name on it, roughly forever.

So the value is in what survives between customers. Which seat counts need finance to sign off. How your welcome mail sounds when you write it yourself. The step your CS lead always adds and nobody wrote down.

In Strawberry that knowledge sits with the job. The AI companion that onboards customers holds the onboarding memory, the onboarding skills, and only the accounts onboarding needs. You correct it in passing and the correction is kept.

Anthropic documents memory across chat and Cowork in the cloud, listed under Topics in settings where you can edit or delete any item. It is on by default for Free, Pro and Max plans, and off by default for Team and Enterprise organisations.

Friday's signature should start Monday's onboarding without you remembering.

The trigger is already there.

Somebody signs, and the same eight changes fall due whether or not you have the afternoon.

Save it once and Strawberry keeps how you like it done. The order the systems get touched. The wording of the welcome mail. The point where you want to look before anything goes out.

Then give it a schedule, or a trigger such as a new email arriving. Every run is recorded with what it did and what it cost, so an onboarding that quietly failed is visible rather than discovered by the customer. Anthropic documents scheduled tasks for Claude in Chrome, set to run daily, weekly, monthly or annually.

Then hand the whole thing over. An AI companion carries its memory, its skills and its connected accounts. Your next CS hire inherits the onboarding itself, not a document describing it. Your team workspace holds the shared files, and a shared credit pool pays for whoever is onboarding this week.

Frequently asked questions