AI is everywhere. Every vendor pitch includes it. Every conference has sessions about it. AI innovation is the latest hype cycle in full swing, complete with obviously AI-generated stock images (like the one on this article).
For mid-sized businesses, the question isn’t “Should we use AI?” but “What AI innovation actually makes sense for our business, and what’s just expensive distraction?”
Having worked in IT for 30 years and recently completed an MBA focusing on Design Thinking, I’ve seen enough technology hype cycles to recognise the pattern. Some AI applications genuinely solve problems. Most are solutions looking for problems to justify their existence.
What AI actually does well
AI excels at pattern recognition in large datasets and automating repetitive tasks. That’s not revolutionary – it’s just computers being good at what computers have always been good at, but faster and with less manual programming required.
- Improving existing processes: Supply chain optimisation, inventory management, logistics routing. AI isn’t inventing new approaches – it’s calculating faster and adjusting to changing variables in real time. Incremental improvement at scale.
- Automating routine work: Data entry, invoice processing, basic customer service queries, scheduling. If you’re paying people to do repetitive tasks that follow predictable patterns, AI can probably handle it cheaper. Not exciting, but useful.
- Analysing data you already have: Most organisations collect more data than they analyse. AI tools can spot patterns humans would miss or take weeks to find manually. Predictive maintenance in manufacturing, demand forecasting in retail, risk assessment in finance – these work because they’re processing existing information differently, not creating magic.
What AI does badly
- Strategic decision-making: AI can analyse data and suggest options. It can’t understand your organisation’s culture, your competitive positioning or whether now is the right time to make a particular move. That still requires human judgment.
- Understanding context: AI generates responses based on patterns in training data. It doesn’t understand what it’s saying. This creates problems when context matters – which is most business situations.
- Replacing human judgment: Chatbots can handle “Where’s my order?” They struggle with “This is complicated and I’m frustrated.” Customer service AI works for routine queries. It fails spectacularly when empathy matters.
What businesses should actually do
- Don’t fire your staff to fund AI: The best AI implementations augment human capability rather than replace it. Your customer service team using AI to handle routine queries so they can focus on complex issues? Useful. Replacing experienced staff with chatbots to save money? You’ll discover why that was a mistake when customer satisfaction tanks.
- Start with the problem, not the technology: “We should use AI” isn’t a strategy. “Our customer service team spends 60% of their time answering the same five questions” is a problem AI might solve.
- Look for repetitive, data-driven tasks: If you can write clear rules for a task, AI can probably automate it. If the task requires judgment, creativity or understanding nuance, AI will struggle.
- Use existing tools first: You don’t need custom AI development. Tools like ChatGPT, Claude and Copilot for content drafts, automated transcription services, basic chatbots for websites – these exist, they’re affordable and they work reasonably well. Start there before spending serious money.
- Measure actual impact: AI vendors promise efficiency gains, cost savings, improved customer satisfaction. Measure whether you actually achieve these. Many AI implementations create more work than they save because nobody tested whether they solved real problems.
The Design Thinking perspective
My MBA research focused on human-centred approaches to technology implementation. AI projects fail most often because organisations implement technology without understanding user needs.
Talk to the people who’ll actually use the AI tools. Watch how they currently work. Identify where AI genuinely helps versus where it creates friction. Pilot small, measure results and adjust based on reality rather than vendor promises.
AI is powerful for specific applications. It’s not magic and it won’t transform your business overnight. Treat it like any other technology investment: identify the problem, evaluate whether AI actually solves it, implement carefully and measure results honestly.
The hype will continue. Your job is to ignore it and focus on what actually works for your specific situation.
