What is Shadow AI? (and should you be worried?)

Hi Non-Techies,

Shadow AI might sound like a cool, mysterious new AI tool, but the reality is a little less ‘James Bond’. It’s a term for when people use unapproved AI tools at work.

Today’s newsletter isn’t a lesson in how to stop your team from using these tools (spoiler: you can’t!) but rather a super quick guide to prevent the misuse of these tools (spoiler: you definitely can!).

I’ll share a free custom GPT I built to help you out, too.

Before we get cracking, a quick heads up from me: tickets for my next AI For Non-Techies Virtual Bootcamp are live! Learn how to use AI to create killer content - including powerful copy (that actually sounds like you), dazzling videos and stunning images you’ll want to share everywhere.

It’ll happen between April 15-17 2025, 5-6:30pm GMT. I’ve crammed in £2,659 ($3,372) of value into this three-day course, but entry is just £72/$97 if you book now by clicking the button below:

What We Do in the Shadows

A lot of organisations have a woefully limited (or sometimes non-existent) list of 'approved tools'.

Far too often, this features Copilot - which, as any reader will know, is not my favourite - and perhaps their own locked-down, company-branded version of ChatGPT. They're panicking, and focusing on brands they know and already use. They're not focusing on functionality and innovation.

The problem is that these tools, although fine for beginners, are NOT showing the best of what GenAI is capable of.

For example, last week, I ran a public sector prompting workshop. Some people in the room had ChatGPT, others could only use Copilot on their work laptops, and everything else was banned. We all did the same exercise, but the results from the Copiloters were crap compared to those using ChatGPT. They could see the difference.

"I use ChatGPT with the kids at home", one of them said. "And it's so much better than Copilot".

When we have free access to thousands of tools outside of the workplace, we're going to start to feel frustrated if the tools we have at work are limited. So we'll start using the good stuff regardless of the rules.

One AI Policy to Rule Them All

The best way to build a culture of safe and responsible AI use is through education, not restrictions.

It’s the sort of thing I help businesses with every day, but if you’re looking for an easy place to start, you should create an AI policy if you don’t have one yet.

I built a custom GPT to help. It’s trained to interview you and, at the end of that chat, it’ll generate an AI policy tailored to your organisation. I like to name my GPTs, so if you have any name ideas, please reply to this email and let me know. Policy Polly? Blueprint Barry?

AI Education for the Nation

The organisations with the healthiest and safest AI cultures don’t limit the tools their team can use, they carefully present them with a balanced smorgasbord of innovation and security in their AI tools. It makes for a happier, more productive team and, most critically, it means AI becomes your organisation’s best mate, rather than an adversary that needs to be curtailed. Big thumbs up.

Non-Techies, if you take nothing else from this newsletter, please remember this: Restricting access to great AI tools is a much bigger risk than simply educating your organisation about the best way to use them.

Measuring the Impact of AI: We’re Struggling

I've been digging into HubSpot's fascinating 2025 State of Marketing Report, and one statistic jumped out at me: while 92% of marketers say AI has impacted their roles, only 16% know how to measure its impact effectively.

This disconnect shows a challenge we're all facing.

The most successful marketers are focusing on metrics that connect AI use directly to business results:

1) Time saved on routine tasks

For example: I’ve reduced my proposal creation time down from 3 hours to 1 hour using an AI note taker and a Claude Project (and they’re stronger, too)

2) Increased content production 

How much more can your existing team create if they have less admin to do? They can get from idea to execution far more quickly, research deeply in minutes and repurpose existing stuff.

3) Improved conversion rates from more personalised content 

My own best case study was using a $50/mo tool to create $75m revenue for my clients in an AI-driven workflow. We couldn’t have got those results without AI personalisation tools.

The HubSpot report reveals that organisations with connected systems are 35% more confident in tackling their biggest marketing challenges - showing that integration, not just adoption, is the key to success.

It’s well worth a read:

I need to skidaddle. I’m off to rehearse for 2 major speaking gigs this week at the British Library and the Excel Centre. Let the jittery nerves commence.

See you next week? Great. I’ll bring the chips, you bring the dips.