I’ve spent enough time around AI adoption now to notice a pattern. Ask a business how AI is going for them and the honest answer is, lately, “we’ve got Copilot, and it’s not great.”

That gap between the hype and the reality is exactly where Samantha North operates. Sam runs an AI transformation agency, helping companies bring AI into their actual workflows rather than bolting on the latest feature because it made headlines.

In our chat about smart AI usage, she gave me a principle worth stealing, no matter what you sell.

AI as a sounding board, not an author

Sam’s approach to AI started with her PhD. She spent six years at the University of Bath studying social data science and analysed Twitter data from both sides of the Brexit debate to understand how online tribalism works. ChatGPT and Claude entered the stage just before she finished, and she used them to organise her research and stress-test her thinking. What she never did is hand over the writing. 

It was never just, “hey, Claude, write me an article about X”. It was like “I’m going to talk at length about all my ideas on this particular subject. I’m going to use this AI tool to help me organize and structure them, maybe come up with some patterns and intersections that I hadn’t thought of”. 

I’m always going to be driving this thing, at the wheel of it. 

Three years on, that’s still how she works. Focusing on the human element is the principle she now brings into every client engagement.

The £4m case for AI 

The clearest proof of that approach is a project she completed for a marine safety company. The business bids for government and MOD contracts. One of their team members had to check multiple tender portals every day, getting overwhelmed and missing opportunities as a result.

Sam got them a system that pulled data in from those sources and used AI to score each opportunity against the company’s own capabilities and goals. So, the human on the other end only had to read an email update each morning rather than trawl through portals. 

“I think we found four million pounds worth of opportunities for them within the first month of operating that system.” Sam told me.

The bidding itself still needed a human. The AI’s job was to make sure nothing worth bidding on ever went unseen.

What I also liked about this story was that this client didn’t come from a sales pitch. The managing director found Sam through a YouTube video where she’d talked through building an early version of the same system. He messaged her directly on LinkedIn.

A minefield worth clearing

That project pointed Sam toward a specific gap in the market. UK SMEs bidding for government contracts are, in her words, navigating “such a minefield.”

It’s an opaque process where a handful of large incumbents scoop up most of the opportunities, even though the government has signaled it wants smaller, more innovative firms to win more of that work.

Most of the SMEs she now works with are stuck running that process through spreadsheets and manual admin. To help them, Sam takes the messy, high-friction workflow and streamlines it with AI and automation. 

he finds a process nobody has time to do well. Then she lets AI carry the parts that don’t need a person’s judgement. Finally, Sam makes sure there’s a human firmly in charge of the parts that do.

It’s the only chance for a small team to manage such complex processes properly.

Stay the author, not just the editor

My next question to Sam was whether she’s optimistic or pessimistic about human judgement in an AI-heavy world. She landed somewhere in the middle.

“Judgment and expertise should not be replaced, and don’t need to be replaced” she told me.

But in some ways, that replacement has already started. When the current wave of AI tools landed in late 2022 and early 2023, nobody really taught people how to use them properly.

There’s always going to be this chunk of the population that are just using AI as like outsourcing their thinking and their judgment, right? And I think we’re already starting to see this cause issues. 

The result is AI slop flooding the internet. LinkedIn’s response is a new reporting feature, the “Seems like AI slop” button.

That hasn’t put Sam off using AI, and it shouldn’t put you off either. She still uses Claude to structure her LinkedIn posts, keeping herself in the driver’s seat.

Every post starts from an idea, an experience or an opinion that’s genuinely hers, voice-dumped into the tool before AI does any organising. That’s what makes the difference between a text you’ll want to read and one you’ll just scroll over. 

“That knowledge is not common knowledge at the moment. I don’t know how to make it common knowledge, but it should be” she added.

Why this matters for MSPs

Most of your clients this year will fall into one of two camps. Either they’re barely using AI at all, or they’re using it in a way that quietly hands over judgement they should be keeping for themselves. 

What I took from talking to Sam is that the real opportunity is not selling a smarter tool. It’s helping clients stay in the driver’s seat and apply AI to the process that’s actually broken. That could mean ticket backlogs, alert triage, or getting a quote out before a competitor does. 

Whatever the process, make sure a human is still the one deciding what matters. Get that distinction right, and you’ll be solving a problem most of the market hasn’t even noticed yet.

If you liked this article, follow us on LinkedIn, Reddit, X, Facebook, and Youtube.