Klarna made headlines in February 2024 claiming their AI chatbot could do the work of 700 customer service representatives. A year later, they’ve quietly started hiring humans again, admitting that “in a world of automation, nothing is more valuable than a truly great human interaction.”

Even Anthropic’s CEO recently claimed that “everything you do is eventually going to be done by AI systems,” predicting a billion-dollar company with just one human employee by 2026. Yet at his own company, where 70% of code pull requests are now AI-written, engineers are still needed to “orchestrate” the AI codebase and attend meetings.

This pattern is everywhere. Everyone’s either terrified AI will replace them or convinced it’ll solve all their problems. Both camps are missing the point entirely. The government is throwing money at businesses to adopt AI through Enterprise Ireland grants. Startups are raising millions with AI in their pitch decks. Almost all companies claim they’re investing in AI, yet just 1% believe they’re actually good at using it.

AI isn’t your replacement or your strategic brain. It’s a very capable assistant that excels at specific types of work. Instead of replacing customer service reps, use AI to handle the routine stuff – password resets, order status checks, basic troubleshooting. Let humans deal with complaints, complex issues, and anything requiring empathy or judgment.

Have AI create initial drafts of routine emails, product descriptions, or social media posts. But everything needs human editing for tone, accuracy, and brand voice. AI doesn’t understand your company’s personality or strategic context. AI can spot patterns in sales data, categorise customer feedback, or summarise long documents. But always verify the conclusions – AI can miss context and occasionally hallucinate facts.

Letting AI send customer emails without review, make pricing decisions, or handle complex negotiations is asking for trouble. These require human judgment, relationship understanding, and strategic thinking.

I keep seeing the same mistake. Companies are looking for the biggest, most impressive use cases when they should start small and build competence. Pick something specific and repetitive that doesn’t require creativity. Customer inquiry categorisation. Expense report processing. Meeting note summarisation. Get good at that before expanding.

Every AI output needs human review initially. Create clear guidelines about what needs checking and who’s responsible. Most AI failures happen because nobody’s actually reviewing the output. Your people need to understand what AI can and can’t do. They should know how to write good prompts, spot when AI is going wrong, and understand its limitations. This isn’t intuitive and requires proper training.

More importantly, create clear policies about what AI can be used for and what’s off-limits. Document these guidelines so you’re covered if regulators come asking questions later. Most AI compliance issues come from people using tools without proper oversight, not from the technology itself.

Track time saved, not tasks automated. If AI helps your sales team spend 30 minutes less per day on admin, that’s 2.5 hours per week for actual selling. Measure what matters.

Companies succeeding with AI aren’t seeing dramatic transformation. They’re seeing lots of small improvements that add up. I helped one client set up AI to categorise and tag incoming leads from different channels, which has increased efficiency massively. Now their marketing team can spend that time on actual campaign strategy and relationship building instead of data entry.

Another client uses AI to categorise customer feedback, which used to take hours of manual work. Now their customer service manager can spend that time on actually improving service instead of spreadsheet organisation. These aren’t revolutionary changes. They’re practical improvements that make work more efficient and more interesting.

The companies struggling with AI usually have one thing in common: they expected too much too fast. They wanted AI to handle complex customer complaints without realising those complaints often involve emotional customers who need human empathy. They wanted AI to write marketing copy without understanding that good marketing requires deep knowledge of customer psychology and brand positioning.

They treated AI like magic instead of like a very good intern who needs clear instructions and careful supervision.

Stop looking for the perfect AI use case that will transform everything. Start with something small and boring that takes your team time but doesn’t require complex thinking. Get good at using AI for routine tasks with proper human oversight. Learn what works and what doesn’t. Build competence gradually.

Most importantly, keep humans in control of anything that involves judgment, creativity, strategy, or relationships. AI can help with the groundwork, but the important decisions still need human brains.

The companies that will benefit most from AI are the ones that stop thinking about it as revolutionary technology and start thinking about it as a really capable assistant that’s available when you need it.

For most businesses right now, AI is a tool that helps you work faster and handle routine tasks more efficiently. Yes, it might eventually transform entire industries, but that’s not what’s happening for the majority of companies today. Right now, it’s mainly making the boring stuff quicker so you can spend more time on the work that requires human judgment.

The sooner businesses stop expecting immediate magic transformation and start using AI for what it can actually do well today, the sooner they’ll see real benefits. Those benefits are genuine, they’re just more practical than revolutionary for most of us.

Because that’s what it is. And honestly, that’s pretty useful.

Jenny Martin is a fractional CMO working with scaling businesses across the UK and Ireland.

Book a Free Consulatation