AI is worth the investment when it removes repetitive work your team is already doing. It's a waste of money when it's added because it sounds impressive. Here's how we tell the difference.
We get asked to 'add AI' fairly often, and the honest first response is usually a question: what task is taking up your team's time right now? AI earns its cost when it removes repetitive work someone is already doing by hand. It becomes an expensive decoration when it's bolted on because a competitor mentioned it.
Where it consistently pays off
The clearest wins we see share a pattern - there's a high volume of similar decisions, and historical data showing how those decisions were made. Customer support is the obvious one: a well-trained assistant can handle the same twenty questions that make up most of your inbox, and hand off the genuinely unusual ones to a person.
Forecasting is another. If you're ordering stock based on intuition and last month's numbers, a model trained on your own seasonal patterns will usually do better - not because it's clever, but because it can hold more history in mind than a person can.
Where it disappoints
AI struggles when there's no data to learn from, when the decisions are genuinely novel each time, or when being wrong is expensive and unrecoverable. It also struggles when the underlying process is broken - automating a bad workflow just produces bad outcomes faster.
“If a process is unclear to your own team, an AI system won't clarify it. It will just make the confusion harder to see.”
The data question comes first
Most AI projects that stall don't stall on the modelling - they stall because the data was scattered across spreadsheets, systems, and people's heads. Getting that data clean and connected is usually the first real phase of work, and it's worth doing even if you never build the model.
A reasonable way to start
If it works, expand it. If it doesn't, you've spent weeks rather than a year finding out - and you still end up with cleaner data than you started with.
