Claude for business in Australia

For most businesses the choice is Team or Enterprise. The API is a separate thing: it is how a developer puts Claude inside a system, billed by usage.

The deciding question: does your IT team provision every app through Entra ID or Okta, and does someone expect logs of who did what? If yes, Enterprise. If your ops lead adds new starters by hand, Team covers you. The full comparison is in rolling Claude out to a 100 person business.

It depends which door you use. Through Amazon Bedrock in the Sydney region, the request is processed on infrastructure in Australia. Through the Claude apps or the direct API, it is processed offshore, and Anthropic does not present the direct API as a regional residency option. Anthropic’s commercial plans do not train on your content by default.

Sending personal information to a model processed offshore is a cross-border disclosure under Australian Privacy Principle 8, so your privacy policy should say so. Anthropic sets out its own commitments in its commercial terms and privacy centre, and AWS lists which Bedrock models run in which region in its model and region reference.

When we build on Claude for a client, each tool gets a dedicated API key scoped to that client, no client’s data is used to train or fine-tune any model, and any working copy of records sits on a Neighbourhood server in Sydney. Our builds call Anthropic’s API directly, so requests are processed offshore. If you need processing in Australia, we build on Amazon Bedrock in Sydney instead. The detail is in where your CRM data goes when Claude reads it and the Australian answer on AI and your data.

Most businesses we talk to already use AI, just not on purpose. One described it as scattered: personal ChatGPT, Gemini and Claude accounts, with no agreed line on what data can go where. Another was close to banning Claude from sales documents because people were prompting badly and not checking what came back. Both are fixed by the same 30 days.

Start from our one-page AI policy, then set up a Claude Project for your sales team. If you would rather run it with us, tell us about your team and we will scope it.

HubSpot’s connector for Claude works with every HubSpot plan and any paid Claude plan, once a Super Admin approves it. It reads almost every object in your portal, creates and updates a shorter list, and deletes nothing. It also skips your custom validation rules when it writes, which is the line worth reading twice: the rules that stop a deal being saved without an amount are not standing between Claude and your data.

That is why we switch it on read-first, and why the process your team follows has to be written where Claude can read it. More in the Claude HubSpot integration, what it does, the limits we set when we connected HubSpot to Claude and making Claude follow your HubSpot process every time.

Systems that do a job a team can name, running in production with a review step and a log of every run.

• Plungie: briefings written from its own HubSpot and sales data, for teams in Australia and the US. Sales gets a short brief on each buyer before the call.
• NHFA: a regular briefing on the business, written from a data warehouse we built and run for them.
• Flint: a fact-find that reads the documents a borrower uploads, fills the loan record and flags mismatches for a broker to check.

We use Claude ourselves too. Our engineers rebuilt this website with Claude Code, and we wrote up what happened when we gave our engineers Claude Code. For a custom build, see AI consulting in Australia and custom AI systems built on Claude.

A construction business we spoke to this year had narrowed it to Microsoft Copilot, Claude Team and ChatGPT Business, and did not want to get stuck in analysis paralysis. That is the right instinct. Choose on three things rather than on a benchmark:

• Where your work lives. If everything runs through Microsoft 365, Copilot sits closest to it. If your customer records live in HubSpot, check which assistant connects to it, and what it can write.
• What your policy needs. SSO, audit logs, retention controls and where requests are processed. Compare the plan you would actually buy, not the free tier.
• What the pilot shows. Run the same weekly tasks through your shortlist for a fortnight and keep the one your people keep opening.

We build on Claude because it does the work we ask of it inside the systems our clients run, and we will tell you if a different tool fits your business better.

Guides from the team on using Claude at work, with HubSpot and without it.