The four options Australian businesses actually have
When someone says 'I want AI for my business' in 2025/2026, there are four genuine options. They're often confused with each other; they're very different in cost, capability, and risk.
1. ChatGPT Plus ($20/user/month): Personal subscription. Your team uses it ad hoc. Great for individual productivity; risky for confidential business data because OpenAI's training-on-your-data terms apply unless you actively turn them off.
2. ChatGPT Team ($25/user/month, $30/user/month annual): Team plan. Data is excluded from training by default. Includes shared workspace, admin controls. Good for an SMB without specific compliance constraints.
3. OpenAI's no-code GPTs (in ChatGPT Plus/Team): Build a 'GPT' in their builder by uploading documents and writing instructions. Free to create; usable by your ChatGPT subscribers. Limited integration capability.
4. Custom GPT (what we build): A custom-developed AI assistant deployed on your own infrastructure (Azure OpenAI, Anthropic Bedrock, or open-source models). Tight integration with your business systems. Full control over data, prompts, models. Significantly more expensive than off-the-shelf.
How to decide
Pick ChatGPT Plus when
Individual users want general AI productivity. <10 staff. No regulatory constraints. No tight integration needed.
Pick ChatGPT Team when
Whole team needs AI access. SMB with general business data. No specific compliance requirement (no APP, APRA, or healthcare data flows in).
Pick a No-code GPT when
Specific repeated task with documented inputs. The 'task' fits in OpenAI's GPT format. You're already on Plus or Team.
Pick Custom GPT when
Compliance-grade requirement (APP, APRA, ASIC, AHPRA). Deep integration with internal systems. Custom workflow / multi-step process. Need privileged retrieval over your own data. Need to deploy on Australian infrastructure.
Cost comparison (real numbers)
Honest pricing for the four options, for an Australian business with 50 active users:
- ChatGPT Plus: 50 × $20 = $1,000/month. Self-managed.
- ChatGPT Team: 50 × $25 = $1,250/month. With admin controls.
- No-code GPT: Same as ChatGPT Plus/Team — included in subscription. Build cost is your team's time.
- Custom GPT (typical): $15,000–$50,000 build + $1,500–$5,000/month operational + $1,000–$3,000/month inference cost.
Custom GPT is roughly 4–10x more expensive than ChatGPT Team. The right question isn't 'which is cheaper?' — it's 'is the additional capability worth the additional cost?'. For most knowledge-work productivity, the answer is no; ChatGPT Team is fine. For compliance-grade or deep-integration use cases, the answer is yes; Team's limitations are real.
Where the 'no-code GPT' approach falls short
OpenAI's GPTs (the ones built in ChatGPT's builder) are great for some scenarios. They're also limited in ways that aren't always obvious until you've tried them at scale: (1) no real integration to internal systems beyond document upload; (2) limited to ChatGPT-subscribed users; (3) document size limits (typically 20MB per file, 20 files); (4) no fine-tuning on your tone of voice; (5) no audit log or compliance reporting; (6) data residency in US-region OpenAI infrastructure. For internal SOP search or simple task templates they're great — at $0 marginal cost. For anything with real business stakes, you'll outgrow them quickly.
The honest recommendation pattern. Most businesses should start with ChatGPT Team and a few no-code GPTs. Use it for 6 months. Identify the 2–3 places where the off-the-shelf approach genuinely doesn't fit. Then commission custom development for those specific use cases. Custom-from-day-one usually means over-investment.
Frequently asked questions
Will ChatGPT Team comply with our APRA requirements?
Probably not, on its own. ChatGPT Team's data residency is US-based, even though training is excluded. APRA's CPS 234 typically pushes regulated entities toward Australian-region infrastructure for their core risk-bearing data. We've helped APRA-regulated firms deploy custom GPTs on Azure Australia East to meet this requirement.
Can we move from ChatGPT Team to custom later?
Absolutely — that's the right path. We've migrated several clients from ChatGPT Team-built workflows to custom deployments. The data and prompts you've developed in Team transfer over; the team is already trained on AI usage; the custom build focuses on the specific friction points the off-the-shelf product can't solve.
What about Microsoft Copilot? You haven't mentioned it.
Microsoft 365 Copilot is a fifth option — and a very strong one for Microsoft 365-centric businesses. We often recommend it as the starting point if your business runs on Office 365, SharePoint, and Teams. It's $30/user/month for E3+/E5 customers and integrates deeply across the M365 stack. Custom GPT augments Copilot for the use cases Copilot doesn't cover; we don't compete with Copilot for what it does well.
How long until we know which one we need?
30–60 days of intensive ChatGPT Team usage usually surfaces the use cases where it's not enough. Watch for: data sensitivity issues, integration friction, repeated similar tasks that aren't being automated, compliance team raising concerns. Those signals push you toward custom.
Ready to build your custom GPT?
Get a free 30-minute scoping call. We'll map your use case, data sources, and ROI before you commit.
Start the Conversation