Claude for small business: what it does well, and what it doesn't

Short answer: Claude is strong at long-document work, first-draft writing and code. It is weaker at integrations, spreadsheet work and anything that needs real-time information. For a small Australian business, the right question is not "is Claude the best AI?" but "which tool fits the two or three jobs I actually need done?"

What Claude is, in plain terms

Claude is a large language model built by Anthropic, an AI safety company based in San Francisco. You interact with it through a chat interface at claude.ai, through desktop and mobile apps, or through an API if you have a developer on staff. It reads text, writes text, analyses documents and generates code. It does not connect to your business systems out of the box, and it does not make phone calls.

Anthropic sells Claude on four tiers. The free plan gives you access to Sonnet and Haiku models with a 200,000-token context window and a rolling usage cap. The Pro plan costs US$20 per month (about AU$31) and adds access to Opus and Fable models, higher usage limits, priority access during peak times, and Claude Code for development work. The Max plan starts at US$100 per month with five to twenty times the Pro usage allowance. The Team plan, which is the one that matters for a business with more than one person, costs US$25 per seat per month on monthly billing or US$20 per seat per month billed annually for a standard seat, with premium seats at US$125 monthly or US$100 annually¹.

Enterprise plans start at US$20 per seat per month plus usage billed at API rates, with SSO, SCIM provisioning, audit logs, custom data retention, and a 500,000-token default context window¹.

Five jobs Claude does well for a small business

Not every AI tool is equally good at everything. Claude's architecture gives it genuine advantages in a handful of areas that matter to a business with a small team and no dedicated IT person.

1. Reading and summarising long documents

Claude's 200,000-token context window on all plans (roughly 150,000 words, or a 500-page book) means you can paste an entire lease agreement, a supplier contract or a compliance document and ask it to pull out the clauses that matter. Most competing tools have shorter effective context or lose accuracy in the middle of long inputs. If your business deals with contracts, tenders or regulatory documents, this is where Claude earns its subscription cost fastest.

2. Drafting written communication

Proposals, client emails, policy documents, board papers. If you run an accounting practice, that includes BAS summaries, engagement letters and client update emails. Claude produces clean, structured first drafts that read like a person wrote them, not like a template filled in by a machine. The output still needs a human pass, always. But the time from blank page to reviewable draft drops from 30 minutes to five.

3. Analysing data you paste in

Claude can work through a table of figures, find patterns, flag anomalies and produce a plain-language summary. It handles CSV data, JSON exports and plain text tables. For a business that does not have a data analyst on staff, this covers the gap between "I have the numbers" and "I know what they mean."

4. Writing and reviewing code

If your business has a website, an internal tool or a spreadsheet with macros, Claude is strong at writing, explaining and debugging code. Claude Code, available on Pro and above, operates directly in a terminal and can work across an entire codebase. For a business that occasionally needs a developer but cannot justify a full-time hire, this is a genuine capability.

5. Research synthesis

Give Claude five competitor websites, three industry reports and a set of questions, and it will produce a structured comparison that would take a person half a day. The output is a starting point, not a finished analysis. But for a business owner who needs to make a decision by Friday, half a day recovered is worth more than the subscription.

Three jobs Claude is not good at

1. Anything that needs live information

Claude's web search feature is available but limited. If your workflow depends on checking stock levels, current pricing, live regulations or today's news, you will hit a wall. ChatGPT's browsing and Microsoft Copilot's integration with Bing are stronger here. Claude works best when you bring the information to it, not when it needs to go and find it.

2. Working inside your existing tools

Claude does not natively connect to Xero, MYOB, Google Sheets, your CRM or your project management system. You can copy and paste data in, but that is manual work. Microsoft Copilot, by contrast, sits inside Word, Excel, Outlook and Teams if your business already runs on Microsoft 365. If tight integration with your existing stack is the priority, Copilot has a structural advantage that Claude cannot match today.

3. Image generation, design and visual work

Claude does not generate images. ChatGPT has DALL-E built in. If your business creates social media content, product images or marketing collateral, Claude will not help with the visual side. It can write the caption, plan the content calendar and draft the copy, but someone else has to make the picture.

Claude vs ChatGPT vs Microsoft Copilot: what a business owner actually cares about

Most comparison articles list features. A business owner making a purchasing decision cares about four things: what happens to their data, who controls access, what the tool plugs into, and what it costs.

The honest summary: if your business lives in Microsoft 365, Copilot is the path of least resistance. If your team needs a general-purpose AI for writing, analysis and code, Claude and ChatGPT are close on price and capability, with Claude stronger on long documents and careful reasoning, and ChatGPT stronger on browsing, image generation and third-party integrations. Most businesses with more than a handful of people will end up using at least two of these, because none of them does everything well.

The Australian data residency question

If your business handles personal information under the Privacy Act 1988 (Cth), you need to know where that data goes when you paste it into an AI tool.⁷ Australian Privacy Principle 8 requires you to take reasonable steps to ensure an overseas recipient handles personal information in line with the APPs, or to get informed consent for the cross-border transfer.⁸

Here is what each vendor offers today.

Claude: Anthropic processes data in the United States by default. There is no Australian-region option for the direct Claude chat product. If your organisation uses the API through AWS Bedrock, you can route inference through the AWS Sydney region (ap-southeast-2), which keeps prompts and responses in Australia. This requires a developer to set up and is not available through the standard claude.ai interface³.

ChatGPT: OpenAI processes data in the United States. Enterprise customers can negotiate data residency terms, but there is no self-serve Australian region.

Microsoft Copilot: If your Microsoft 365 tenant is provisioned in Australia, your data stays in the Australian datacentre region. For businesses already on Microsoft 365, this is the simplest path to keeping AI-processed data onshore.

The practical takeaway for most small businesses: if you are pasting customer names, financial figures or health information into any of these tools, check the vendor's data processing agreement against your obligations under APP 6 and APP 8. If you are using the tool for internal work only, with no personal information in the prompts, the residency question is less pressing but still worth understanding. An AI policy that names your approved tools and sets rules for what data goes in is the minimum responsible step.

How to decide which tool to start with

Do not start with the tool. Start with the job. It is the same order we work in at Bulletproof: the workflow first, the tool second.

Write down the three tasks that eat the most time in your week. For each one, ask: does this task involve reading long documents (Claude), generating images or browsing the web (ChatGPT), or working inside Office applications (Copilot)? The answer usually points you to the right first subscription.

If the answer is "writing emails and drafting documents," all three will do the job. In that case, sign up for the free tiers of Claude and ChatGPT, test both on your actual work for a week, and upgrade the one your team prefers. The Pro tiers cost the same (US$20 per month), so the deciding factor is which interface and output style your people will actually use, not which benchmark score is higher.

For a practical framework on choosing a first AI project and measuring whether it is working, the guide to starting with AI covers the method we recommend to every business we work with.

Frequently asked questions

Is Claude free for small business use?

Claude has a free tier with access to Sonnet and Haiku models, a 200,000-token context window and rolling usage limits. It is enough to test whether Claude suits your workflows. For regular business use, the Pro plan at US$20 per month or the Team plan at US$20 to US$25 per seat per month removes the tight usage caps and adds admin controls¹.

Does Anthropic train its models on my business data?

On Team and Enterprise plans, Anthropic does not use your inputs or outputs for model training by default. On the free, Pro and Max plans, training is opt-in: you choose whether to allow it, and it is off by default since mid-2026. If you submit feedback through the thumbs-up or thumbs-down buttons, that conversation may be used for improvement regardless of plan².

Is Claude better than ChatGPT for business?

Neither is categorically better. Claude handles long documents and nuanced writing more reliably. ChatGPT has stronger web browsing, image generation and a larger ecosystem of plugins and integrations. For most small businesses, the deciding factor is which tool fits the specific jobs you need done and which interface your team will actually adopt.

Can I keep my data in Australia when using Claude?

Not through the standard claude.ai interface. If your organisation accesses Claude through the API on AWS Bedrock, you can route inference to the AWS Sydney region. This keeps prompts and responses in Australia but requires technical setup. Microsoft Copilot is the simplest option for Australian data residency if your business already uses Microsoft 365.

How much does Claude cost per user for a team?

The Team plan standard seat costs US$25 per user per month on monthly billing, or US$20 per user per month billed annually. Premium seats are US$125 monthly or US$100 annually. Enterprise plans start at US$20 per seat per month plus usage billed at API rates. All prices are in US dollars and exclude GST¹.

What is the difference between Claude Pro and Claude Team?

Pro is an individual plan. Team adds multi-user admin controls, SSO, central billing, a shared workspace and a default policy that Anthropic will not train on your data. If more than one person in your business will use Claude, Team is the right starting point. If it is just you, Pro is sufficient and US$5 per month cheaper on monthly billing.