Inform Pillar | Procurement Knowledge for SMB Buyers
5 min read
The Australian AI Provider Comparison
Sylvia Luchian
Founder & Head of Procurement Practice
Everyone has an opinion on which AI tool you should be using. Opinions are like bellybuttons.
What most people sharing those opinions are not doing is reading the terms and conditions, checking where your data actually goes, or comparing what you pay against what you get. That is the job. So let's do it properly.
This is Part 1 of the D1 Advisory Australian AI Provider Comparison. It covers Tier One: the four global consumer and enterprise platforms that most Australian small to medium businesses will encounter first. Tier Two (specialised sector and function-specific platforms)and Tier Three (Australian and regional providers) will follow in subsequent posts.
By the end of this article you will know what each Tier One platform costs, where your data lives, what the free tier will and will not protect, and which platform is most likely to suit your business depending on the tools you already use.
Most Australian small business owners choose an AI tool the same way they choose a coffee order. They go with what someone else recommended, they pick the familiar name, or they just try the free one and see what happens.
That approach is fine for coffee. It is less fine when the tool you are using is processing your client communications, your financial data, your contracts, and your internal strategy documents.
The AI market is designed to get you in on a free tier and move you up. That is not a criticism. It is just the commercial reality. The problem is that the free tier and the paid tier are often not the same product when it comes to data handling. Your data on the free tier of some platforms can be used to train the model. On certain paid tiers, that risk remains unless you actively opt out. On business and enterprise plans, the protections change significantly. Most people do not read far enough to know the difference.
Procurement, at its most useful, is the discipline of understanding exactly what you are buying before you commit. It applies to AI subscriptions just as much as it applies to software contracts, fit-outs, and service agreements. The questions are the same: What am I actually getting? Where does my data go? What am I locked into? What does it cost when I scale?
Let's answer those questions for each of the four Tier One platforms.
Origin: USA. OpenAI, San Francisco.
ChatGPT operates across four consumer tiers and a separate business track.
• Free: Access to GPT-4o mini. Limited usage. Consumer data protections apply.
• Plus: USD $20/month. Full GPT-4o access.
• Pro: USD $200/month. Unlimited usage across all models.
• Go: Available in select markets. Positioned between Plus and Pro.
• Team: USD $30 per user per month. Business-grade protections.
• Enterprise: Custom pricing via OpenAI's sales team. Annual and multi-year contracts available.
This is the part most people skip. The answer depends entirely on which tier you are on.
On the free and Plus tiers, OpenAI's consumer terms permit your data to be used to train models unless you manually opt out in Settings under Data Controls.¹ This is not hidden. It is in the terms. Most people do not read the terms.
On Team and Enterprise plans, OpenAI does not train on your data by default.¹ That is a meaningful distinction.
OpenAI has made Australian data residency available for ChatGPT Enterprise and API customers. Your data can be stored at rest in Australia. The important caveat is that data at rest and inference are two different things. Inference is what happens when the model is processing your prompt. Even with Australian data residency, your prompts are temporarily processed on infrastructure before the result is returned to you.
Key Takeaway: Never use a free or Plus ChatGPT tier for client data, financial records, or anything sensitive. If you are on Team or Enterprise, you have meaningful data protections. If you are not, you do not. And if you are on Free or Plus, the opt-out is available, but you need to go and turn it off.
ChatGPT has the widest integration ecosystem of any platform on this list. It connects to more third-party tools, has the largest developer community, and covers the broadest range of use cases: content creation, coding assistance, customer communication, research, and general productivity.
If you do not have an existing software ecosystem pulling you towards a different platform, ChatGPT is a reasonable starting point, provided you are on a tier that actually protects your data.
Origin: USA. Anthropic, San Francisco.
• Free: Limited usage.
• Pro: USD $20/month (USD $17/month billed annually).
• Max: USD $100/month (5x usage limits) or USD $200/month (20x usage limits).
• Team: Approximately USD $30 per user per month.
• Enterprise: Custom pricing via Anthropic's sales team.
This is where the picture changed significantly in late 2025, and most businesses using Claude have not caught up.
In September 2025, Anthropic updated its Consumer Terms and Privacy Policy. Claude Free, Pro, and Max plans moved to an opt-out model for model training.² Unless you actively turned it off, your conversations on those plans may be used to improve Anthropic's models. The data-sharing toggle in the updated terms was pre-checked, making it straightforward to accept without realising the implications.²
If you are on Claude for Work, an Enterprise plan, or accessing Claude via the API (including through Amazon Bedrock), you are excluded from this. Commercial plans remain protected bydefault.³
Australian data residency for Claude is available, but with a catch for small businesses: accessing it requires routing through Amazon Bedrock using the AWS Sydney region. That is not a setting you flick on in a dashboard. It requires technical capability or a managed service provider to set up correctly.
Key Takeaway: Claude has strong data protections on commercial and enterprise plans. If you are on a consumer plan (Free, Pro, or Max) and have not actively opted out of model training in your Privacy Settings, your conversations may be in the training pipeline. Go and check your settings. Today.
Claude performs particularly well on long document analysis, structured reasoning, and tasks requiring writing quality and nuance. If your business works with contracts, complex reports, or detailed client communications, Claude is worth evaluating seriously. For businesses handling sensitive or regulated data, the commercial plan is the appropriate tier.
Origin: USA. Google DeepMind.
• Free: Gemini 1.5 Flash.
• AI Pro: USD $19.99/month.
• AI Ultra: USD$249.99/month.
• Gemini Enterprise: USD $30 per user per month, bundled with Google Workspace.
• Gemini Code Assist: From USD $19 per user per month for developers.
Google processes data across global infrastructure. Workspace Enterprise customers can configure data region settings, including Australia, for data at rest. The critical limitation is that data region settings are a Workspace Enterprise feature. They are not available on lower Workspace tiers.
On the consumer Gemini plans (Free and AI Pro), Google's terms treat your data as consumer data. If you are a Google Workspace business user, your data protections depend on your Workspace edition and settings.
On the compliance and certification front, Google Gemini holds ISO 42001 certification, the international standard for AI Management Systems, achieved in December 2024.⁴ Anthropic and Microsoft have since earned the same certification. ISO 42001 matters if you are evaluating AI tools against a compliance or governance framework.
Key Takeaway: If your business runs on Google Workspace, Gemini is the most natural fit. The native integration with Gmail, Docs, Sheets, and Drive is genuinely useful. If you are not on Google Workspace Enterprise, the data region controls are not available to you. Check your Workspace edition before assuming you have the data protections the platform can offer.
Businesses already operating inside the Google Workspace ecosystem will get the most from Gemini. The native integration is deep and genuinely reduces friction. Gemini also has strong multimodal capabilities, meaning it handles text, images, and data across formats more fluidly than some competitors.
Origin: USA. Microsoft, Redmond. Powered by Azure infrastructure.
• Free tier available.
• Pro: USD $10/month.
• Pro+: USD $39/month.
• Microsoft 365 Copilot: USD$30 per user per month. Requires an existing Microsoft 365 subscription.
• Copilot Studio: USD$200/month (25,000 credits for building custom AI agents).
This is where Microsoft Copilot has a meaningful structural advantage over some competitors. Azure OpenAI is not the same infrastructure as OpenAI's consumer product. When you use Microsoft Copilot through Azure, your data is processed in the Azure region associated with your tenant.⁵
Microsoft does not use business customers' data to train its models.⁵ Your prompts, outputs, and embeddings are not used to improve foundation models.
In November 2025, Microsoft announced in-country processing for Australian Microsoft 365 Copilot customers, available as an opt-in option.⁶ This means prompts and model responses can be processed within Australian borders, not just stored here. This option is designed primarily for government and regulated industries but is available to eligible commercial customers. It is not the default for all tenants. You need to enable it. Verify availability and the configuration process with Microsoft or your licensing partner before assuming it is active.⁶
Key Takeaway: For Microsoft 365 users, Copilot has a strong data sovereignty story for Australian businesses. In-country processing is available as an opt-in option. The separation from OpenAI's consumer infrastructure is a genuine compliance advantage. Understand your Workspace tier and whether the in-country processing option has been enabled before treating it as a given.
Businesses running on Microsoft365 will find the most value here. Copilot is embedded directly into Word, Excel, Outlook, and Teams, which means it works inside the tools your team is already using rather than sitting alongside them. The Microsoft 365 Copilot licence requires an existing Microsoft 365 subscription to make sense economically, so the total cost for smaller teams needs to be assessed carefully.
Before making a decision, compare these four platforms across the dimensions that matter most for an Australian small business.
This is the most practical starting point. If your business runs on Microsoft 365, start with Copilot. If you live in Google Workspace, start with Gemini. If you have no strong ecosystem tie and just want to explore what AI can do, ChatGPT on a Team plan is a reasonable entry point. The tool that works inside your existing workflow will get used. The tool that sits next to it probably will not.
If you are using AI for general content drafting, ideation, or non-confidential research, the platform matters less. If you are putting client data, financial records, legal documents, or anything commercially sensitive into the tool, the tier you are on and the data handling terms matter a great deal. Read the terms before you decide, not after you have been using the free version for six months.
Free tiers exist to get you started, not to run your business on. Work out what the tool will cost at the team size you are aiming for in twelve months, not the team size you have today. A USD $30 per user per month platform that ten people need to use is USD$3,600 a year before currency conversion. That is a procurement decision, not a free trial.
Key Takeaway: Match the platform to your existing tools. Understand the data handling terms before you commit. Price it at scale, not at entry.
Choosing an AI tool for your business is, at its core, a procurement decision. It involves understanding what you are buying, what it will actually cost, what risks you are taking on, and whether the thing you are subscribing to will do what it says it will do.
The AI market is moving fast. Pricing changes. Features are added and removed. Data handling terms are updated. The comparison in this article is accurate as at publication. Verify current terms directly with each vendor before signing anything.
Tiers Two and Three are coming. Part 2 will cover specialised platforms, including Salesforce Einstein, HubSpotAI, and Xero's AI capabilities. Part 3 will cover Australian and regional providers, because there are genuinely good local options that deserve more attention than they get.
For now: do not let the noise make this harder than it needs to be. Pick the tool that fits your existing stack. Use the tier that actually protects your data. Price it properly before you commit.
Book a discovery call with D1 Advisory. We will help you work out which platform actually suits your workflow, what tier makes sense for your data handling obligations, and what it will genuinely cost to run it at scale. We promise it will not involve a sales pitch, a deck full of buzzwords, or anyone saying 'synergy' with a straightface.
Anthropic 2025, 'Updates to our consumer terms and privacy policy', Anthropic, San Francisco, viewed 27 March 2026, https://www.anthropic.com/news/updates-to-our-consumer-terms.
Anthropic n.d., 'Is my data used for model training?', Anthropic Privacy Center, viewed 27 March 2026, https://privacy.claude.com/en/articles/7996868-is-my-data-used-for-model-training.
Google n.d., 'ISO/IEC 42001 compliance', Google Cloud, viewed 27 March 2026, https://cloud.google.com/security/compliance/iso-42001.
Microsoft 2025, 'Microsoft offers in-country data processing to 15 countries', Microsoft 365 Blog, 4 November, viewed 27 March 2026, https://www.microsoft.com/en-us/microsoft-365/blog/2025/11/04/microsoft-offers-in-country-data-processing-to-15-countries-to-strengthen-sovereign-controls-for-microsoft-365-copilot/.
Microsoft n.d., 'Copilot data movement across geographies', Microsoft Learn, viewed 27 March 2026, https://learn.microsoft.com/en-us/dynamics365/business-central/ai-copilot-data-movement.
OpenAI n.d., 'How your data is used to improve model performance', OpenAI, San Francisco, viewed 27 March 2026, https://openai.com/policies/how-your-data-is-used-to-improve-model-performance/.
Procurement is trust brokering in commercial form. When done properly, it creates the conditions under which buyers and suppliers can exchange value with confidence.
Small and specialist suppliers are operating in a more economically precarious environment than three years ago. Their costs have risen. Margins have compressed.
When a vendor relationship goes wrong, it rarely feels like a commercial dispute. It feels like a betrayal. You were promised something.
Book a Discovery Call
