Having an AI strategy for the sake of it is pointless. You’ll just end up with expensive tools gathering digital dust. I see this pattern everywhere. Companies announce AI initiatives with great fanfare, create working groups, and hire consultants. Six months later, they’re wondering why their £50,000 investment in enterprise AI tools hasn’t moved the needle. The problem isn’t the technology – it’s that they started with the wrong question.
Think of AI as a really focused intern
Here’s how I explain it to people: AI is like a really, really focused intern. You wouldn’t walk into your office and announce an “intern strategy”, would you? You’d identify specific tasks that need doing, then find and train an intern who’s brilliant at those particular things.
The same logic applies to AI. It’s not a strategic initiative in and of itself – it’s a tool that can be trained to be exceptionally good at certain jobs. The magic happens when you match the right AI capability to the right workflow problem.
At Finimize, we serve over a million retail investors and work with 350+ financial institutions. We didn’t start by asking “How can we use AI?” – we started by asking “What’s taking our team too long to do well?”
Start with what’s actually broken
Before implementing any AI solution, we needed to understand where our processes were genuinely inefficient. Our content team was spending hours on quality assurance – fact-checking articles, making sure the tone was consistent, verifying that we hadn’t accidentally used British spelling where we meant to use American… Smart people doing repetitive work that a computer could handle.
We mapped out exactly where time was being wasted. Editors were reading the same articles multiple times – once for facts, once for structure, once for compliance. Analysts were translating insights manually for different markets. Simple tasks that required attention, but not creativity.
Only then did we look at AI solutions. We implemented tools for fact-checks, structure verification, language consistency, and compliance monitoring. The result? 95% efficiency improvements in our content workflows. Our analysts and editors now spend their precious time on deep research instead of proofreading.
The personalisation problem nobody talks about
Everyone knows personalisation matters, but most companies approach it backwards – they start with demographics and try to guess what people want. Age, location, income: standard stuff that feels scientific, but misses the point entirely.
We discovered something counterintuitive. What someone reads tells you more about their investment intentions than their age or marital status. If someone’s consuming content about property investment, mortgages, and savings accounts, they’re probably considering buying a house. That’s more useful than knowing they’re 32 and married.
We built AI systems around this insight: one for users with established reading patterns, another for newcomers. A system that works with reading history can predict what someone wants to consume next with remarkable accuracy. And users getting personalised recommendations show 50% higher engagement than those seeing our standard feed.
The key was understanding the problem first. Generic personalisation wasn’t working. Once we knew why, AI became the obvious solution.
Scale without losing your voice
Here’s another example that demonstrates how problem identification leads to better AI implementation. Financial content needs to work globally, but most translation kills the nuance that makes content valuable. Technical terms get mangled. Cultural context disappears. You end up with content that’s technically correct, but useless.
We identified the specific problem: how do you maintain quality and tone across 30 languages without hiring native speakers for every market? AI translation has improved dramatically, but it needed human oversight for financial terminology and cultural relevance.
Now our content reaches 40 million investors through partner networks worldwide. The AI handles initial translation, humans verify financial accuracy, and local teams review cultural context.
Why most companies get this wrong
The biggest mistake is treating AI as a magic solution rather than a focused tool. I’ve seen companies create AI working groups while banning their employees from using ChatGPT. They’ll spend months developing an AI policy but won’t let anyone experiment with the tools.
The organisations that succeed – like J.P. Morgan – take a different approach. They identify specific use cases, test AI tools on non-critical workflows, and scale what works. They focus on enhancing existing capabilities, rather than revolutionary transformation for the sake of it.
The failing companies believe vendor promises about AI solving everything. They implement AI-first strategies without understanding what problems they’re actually trying to solve. If your base process is flawed, AI just amplifies those flaws faster.
A framework that actually works
The most effective approach starts with identifying your most expensive manual processes. What’s taking smart people too long to do? Where are you bottlenecked by routine tasks that require accuracy, but not creativity?
Map these workflows in detail. Understand exactly where time gets wasted and why. Don’t assume – measure it. You might discover that your biggest problem isn’t what you thought.
Then look at AI tools that address these specific issues. Test them on non-critical work first. Measure efficiency gains, not just “AI adoption”. Build internal capability before committing to expensive enterprise solutions.
Most importantly, maintain human oversight on anything that matters to your business. AI excels at routine tasks, but judgment calls, for now at least, still need people.
The real limitation
AI is the worst it will ever be today. Any current limitations are temporary – the technology improves at an extraordinary pace. Still, many companies get distracted by debates about what AI can’t achieve, instead of focusing on what it can do right now.
The real limitation is organisational. Companies that treat AI as a magic solution waste money and time. Those that treat it as a powerful tool for specific jobs get remarkable results.
The most successful AI implementations don’t start with the technology at all. They start with a simple question: what’s the most expensive thing your team does that doesn’t require human judgment? Answer that first, and the AI strategy writes itself.
