This is the third of three articles expanding on comment we gave to South East Business Insider for its October feature on digital restructuring. The others covered the warning signs and where digital investment pays back.
Apply the same test you would to any technology
AI has enormous potential, but businesses need to apply exactly the same principles to it as any other technology: start with the problem.
There is a danger at the moment of organisations feeling they need an AI strategy without necessarily knowing what they want AI to achieve. That produces pilots that demo well and change nothing, because nobody defined what was supposed to improve.
The question is not “where could we use AI?”. It is “which parts of this process are slow, error-prone or dependent on one person’s knowledge?” — and then asking whether AI happens to be the right tool for any of them. Often it is. Sometimes a rule and an integration would do the job more reliably and more cheaply.
Where it is genuinely useful
The useful applications are generally very practical:
- Summarising information — long threads, documents, call notes
- Classifying enquiries so they reach the right person first time
- Extracting data from documents that would otherwise be rekeyed
- Assisting customer service teams with drafted responses and context
- Analysing large amounts of business information
- Creating first drafts that a person then edits
- Helping employees find answers from internal knowledge
None of those is especially exciting. All of them remove work that a person is currently doing without applying much judgement, which is exactly the point. This is the ground generative AI covers well in a commercial setting.
Where it gets genuinely interesting
Where it becomes particularly interesting is when AI sits within a wider automated process.
Rather than somebody manually reading information, deciding what it means and then carrying out the next five administrative steps, AI can increasingly make sense of that information and allow the surrounding systems to determine what should happen next.
That is the difference between a chatbot and an agent that takes actions inside your systems. The value is not in the model understanding the email. It is in everything that happens automatically afterwards — the record updated, the task created, the customer told, the invoice raised.
It also means the AI is only as good as the systems around it. If the data is scattered across spreadsheets and disconnected platforms, there is nothing for it to act on. In most businesses the foundation work comes first.
The distinction that matters
The important distinction is between using AI because it improves a process and adding AI simply because it is available.
A practical way to hold that line is to ask three questions before any AI project:
- What specifically gets better, and how would we know?
- What happens when it gets something wrong?
- Could a simpler method achieve most of this?
If a project cannot answer the first, it is not ready. If it cannot answer the second, it is not safe to put near customers. If the honest answer to the third is yes, do the simpler thing.
The part most businesses skip
Where the data goes matters, particularly for anyone handling regulated or commercially sensitive information. A general-purpose AI tool may be processing your customer data in a jurisdiction you have not assessed, under terms you have not read.
That is a solvable problem — but it is a decision to make deliberately at the start, not a discovery to make later. We cover it in more detail in where your data goes when you use AI.
For a business just starting
Start smaller than you think. Pick one process where a person is reading something and typing a decision, measure how long it currently takes and how often it goes wrong, and change that one thing.
A single working automation that removes four hours a week will teach a business more about where AI fits than a strategy document ever will — and it builds the internal confidence that makes the next change easier.
What Digital Scientists® does
We start with the process, not the model. Where AI genuinely improves it, we build it into the systems a business already runs so the output actually does something. Where it doesn’t, we say so.
If you want an honest assessment of where AI would and would not help your business, book a free discovery call, or read more about our approach to AI strategy and governance.