The common belief that AI is only for large organisations is a misconception. AI is for every business, but it only becomes effective when there is structure behind it.
The question becomes: do you have repeatable, defined processes and reasonably clean data for AI to work with?
AI readiness comes down to data maturity and process structure, not company size.
At a practical level, that means:
No duplicate records (customers, item numbers, etc.)
Data captured and updated in a consistent way
Repeatable processes, not dependent on individual interpretation
A few duplicates are not a dealbreaker, but systemic chaos signals deeper organisational problems.
Once you have a repeatable process, you have a foundation you can actually improve with AI or automate with agentic AI.
However, businesses still on spreadsheets, multiple bespoke systems or at the “everything just happens in someone’s head” level are at a significant disadvantage: fragmented data, manual duplication, and results you can't fully trust.
If SMEs haven’t moved to an ERP like SAP Business One, they often have siloed data that doesn’t provide a complete picture. In fact, according to Decidr’s 2025 National AI Readiness Index report, 21% of SMEs identified data quality or availability as the main barrier to adopting AI.
Another disadvantage of manual processes is that multiple people are updating records in different ways. That means the data is not robust, it’s not auditable and it’s not reliable enough to support automation.
Bringing AI into that mix is challenging because there are no clear start and stop points, and measuring success is hard because the process is always changing.
One of the biggest misconceptions about AI is that AI is a whole-of-business strategy.
It isn’t. It is a tool to improve or solve a specific, existing problem or process.
It sits alongside an ERP. It doesn't replace an ERP.
But without a repeatable process, AI has no structure to operate in.
Using a structured ERP like SAP Business One forces process maturity by default. Transactions such as sales orders, invoices, debtor records, and product costs are captured in a structured and repeatable way.
That structured data can then be enriched with AI to surface insights (buying patterns, geographic sales trends) and to ask questions.
How often are we selling this product? What is our largest margin item? Which regions are driving the most growth?
Most ERP users are sitting on valuable insights they never see because they lack the time or tools to run the analysis. But AI can help surface those insights quickly.
Most businesses are using AI as a chatbot (Claude, Gemini, ChatGPT, etc.) for one-off questions, not integrated into the ERP.
For example, businesses might use AI to speed up processes like searching for alternative suppliers, checking delivery-related issues, or gathering information they would previously have searched for manually online.
The next step is closing the loop – feeding AI outputs back into ERP processes rather than switching between disconnected tools.
That’s where practical use cases start to emerge:
Weekly credit worthiness checks across customer databases to identify emerging risk
Updates on market segments or industries customers operate in
Real-time insights delivered to sales or service teams during customer interactions
The value comes from using ERP data more intelligently and continuously enriching it with external context.
Ultimately, success comes down to measurable outcomes. Are customers more responsive? Are they buying more? Is decision-making faster or more accurate?
If the answer is yes, you expand the use case. If not, you adjust or move on to another approach.
The best place to start is small.
Choose one to three discrete, repeatable processes and focus on them for one to three months. The goal is not to implement AI everywhere at once, but to refine specific use cases and properly evaluate whether they’re delivering value.
Before introducing AI into any workflow, businesses need to ensure their processes and data are consistent – the same task being done the same way across the organisation.
If there are multiple ways of doing the same thing, the underlying data can end up being different depending on the method used. That creates a problem when insights are automated, as they may be incomplete or incorrect.
From there, businesses can begin measuring outcomes.
Is the process saving time?
Improving response rates?
Reducing operational costs?
Is it creating a measurable improvement for staff or customers?
It’s also important to recognise that AI has a real cost. Large language models (LLMs) consume tokens every time they process a request, and those costs can escalate quickly if use cases are poorly defined or not properly controlled against ROI.
There are also governance considerations. AI agents may reference information, regulations, or documentation from overseas markets and apply logic that doesn’t align with Australian business requirements or compliance expectations.
The key takeaway is to keep a human in the loop, particularly in the early stages. Businesses should approach ERP-connected AI carefully, with clear governance and guidance from experienced partners before moving into more advanced automation.
AI works best when applied to a discrete, definable, and measurable process. Without that foundation, results are inconsistent and difficult to trust.
The role Key Business Solutions plays is helping businesses identify the right processes, ensure the underlying data is consistent and structured, and implement the right tools for the right job.
Ready to explore? Start a conversation with our team today.