Perspectives

If your team has access to AI tools, do you know what’s actually changed?

Data & AI

Spoiler alert: it’s training that will determine whether businesses integrate AI successfully into their operations.

Our head of AI & data Kai Henderson reveals the impact of a planned AI adoption at Marex

We’ve proven that bringing AI and data into our own business at Marex works.

Over the past six months, we’ve made a real focus of introducing AI into our internal team. Not as an experiment on the side, but properly, in how we actually work day to day.

And we did it carefully.

No big launch. No giving everyone a login and hoping for the best. We planned properly and used AI agents for specific, well-defined tasks. Then, using my background in data science and coding, I built workflows and tools around how our team actually works.

Some of the key outcomes:

Several processes reduced by 70% time or more.
One process that took 64 hours last month now takes 18.
The team are happier as they don't have to do the manual, repetitive processes.
Projects are being completed quicker.
Allocated more time for technical reviews by our experts.

Here’s what I learned along the way, and it’s the bit I think most businesses get wrong.

Rolling out AI is not giving everyone access to Copilot or Claude and calling it done.
That’s the version most companies try. Licences go out, a few people get excited, most people use it like a search engine, and six months later nobody can say what changed.

What actually worked for us was three things:

Training people properly. Not a one hour demo. How to use it well for their actual job.
Being clear on what AI shouldn’t be used for. That conversation matters as much as the one about what it can do.
Building the right tools for the processes that are manual, repetitive and eating hours every week. I built these for the team, and it’s where the 64 to 18 came from. Not a chatbot. Proper workflows built on clean, structured data, which takes data and coding expertise to get right.

Now that it’s proven in our own business, we’re helping other companies do the same. Making the transition properly, with their data and their people, rather than buying licences and hoping.
If your team has access to AI tools but you couldn’t tell me what’s actually changed since, that’s usually the gap.

👋 I’m Kai. I help businesses maximise AI and gain insights on their data. I write about what works and not the hype.

Perspectives