This week, Shopify CEO Tobias Lütke told everyone who works there to basically get with the AI programme.
- Learn how to use it (and you’ll learn best by using it).
- Make it part of every prototype.
- Don’t expect to get money or new staff to do a task unless you’ve proved AI can’t do it.
- Share what works with your colleagues.
- We’re going to make AI knowledge part of your performance reviews.
Pretty clear. But the actual email was hundreds of words long, including lines like:
Having AI alongside the journey and increasingly doing not just the consultation, but also doing the work for our merchants is a mindblowing step function change here.
Stagnation is almost certain, and stagnation is slow-motion failure.
What we need to succeed is our collective sum total skill and ambition at applying our craft, multiplied by AI, for the benefit of our merchants.
In fact, there’s a seven-paragraph build up to the numbered section titled ‘What This Means’, with plenty of big ambitions, personal anecdote and sincerity – which would make quite an engaging speech.
Unfortunately, I suspect most people scrolled straight past it. Which is exactly what makes me think he wrote most of it himself.
AI often gets to the point faster than people do
We’re much more inclined to fall prey to the idea that we need to give lots and lots of background first, even though we happily skim past that in every email we get. After all, we feed AI a diet of buzzwords and corporate speak that humans wrote.
But it’s quickly learning that we don’t like to read what we write. So I put it to the test, using GPT-4o and this prompt:
Write a memo from the CEO of a big ecommerce platform saying that AI use is now a mandatory, expected part of everyone’s work process. Tell them they won’t get any new resource for tasks until they’ve proved AI can’t do the job. And that it’ll be part of their performance reviews. Set the expectation that everyone will learn to be good at AI by using it, and will share what they’ve learned for the benefit of the business. You can use a little bit of personal anecdote, but prioritise being clear.
The 300-odd words it came back with were still a tad lofty for my liking, and I would have liked to see some bullet points – but that’s on me. Some AI output can be perilously close to malicious compliance: you get specifically what you asked for, and almost nothing you didn’t ask for.
Still, it included lines like:
When you find something that works, tell others. Write it up. Make a Loom video. Get it in front of your team. We grow faster when we share.
Compare that to:
We’ll be sharing Ws (and Ls!) with each other as we experiment with new AI capabilities, and we’ll dedicate time to AI integration in our monthly business reviews and product development cycles. Slack and Vault have lots of places where people share prompts that they developed, like #review-ai-use-cases and #ai-centaurs.
If you want people to get on board your bandwagon, less is more
To be clear and persuasive:
- get to the point as quickly as you can
- use as many words as you need to get the job done, and then stop
- use short bullets or numbered lists.
You can put all that information in an AI prompt to make it better, too. Because AI’s doing is only as good as your thinking. When I asked it to try again with that extra guidance, it decided the subject line should be:
AI is now a core part of your job
You tell me how that feels compared to:
AI usage is now a baseline expectation
We do lots of thinking and writing for businesses and teach them how to write good copy and prompts too.
Let us know if you’d like a chat about it.Written by Alex Goldstein, Creative Director (the writing kind), at Definition
