Your CEO Just Said ‘Use AI or Else.’ Here’s What to Do Next.

Was this newsletter forwarded to you? Sign up to get it in your inbox.

A company memo went viral this week, but not because it leaked. It didn’t have to.

Shopify CEO Tobi Lütke posted an internal message—which laid out new expectations for AI use among Shopify employees—publicly on X. The memo, which urged employees to integrate AI into their daily work, use it during early prototyping, and treat it as a core job skill, quickly spread through tech and AI circles. The headline takeaway? “Reflexive AI usage is now a baseline expectation at Shopify.” 

Translation: If you’re not using AI as part of your everyday work, you’re already behind.

As the lead at Every Consulting, I talk to prospective clients every day about how they can make the best use of AI. Lutke’s memo reflects many of the challenges and opportunities I see teams grappling with. Executives want AI adoption but struggle to differentiate value from noise, managers are unsure how to evaluate AI-powered work, and individual contributors feel paralyzed by the gap between AI's potential and their current comfort level.

Shopify is in the middle of a company-wide training montage. Everyone is. Some are further along. Some are just starting. Even the people most excited about this shift—your CEO included—are figuring out what good usage looks like in real time. Lutke admits as much in the memo: He uses AI constantly but still feels like he’s “only scratching the surface.”

This five-step guide is for anyone trying to understand what Shopify’s new AI expectations mean in practice. What does “reflexive AI usage” look like? How do you go from feeling behind to feeling fluent? Most importantly, how do you make AI work for you—without it feeling like more work?

Let’s dive in.

ChatGPT/Every illustration.

Step 1: Start using AI now

Here’s how most people approach AI: They read articles. Bookmark prompt guides. Open ChatGPT, type “help me be productive,” get a meh response, and close the tab. Repeat until morale improves.

But the only real way to get better at using AI is to use it regularly, imperfectly, without overthinking it. 

The first step is to invite AI to the table—literally. Keep ChatGPT open in a browser tab. Download the desktop app. If you have an iPhone, program your Action button to launch ChatGPT's voice assistant with one press. The goal is to make AI feel like a teammate sitting next to you, ready to jump in. Like a new teammate, they may be awkward at first. They may misread your tone or do things slightly wrong. But over time, you learn how to work together.

Pick something you already do—writing a weekly update, sorting tasks, summarizing calls—and try doing it with AI:

• Use Claude to clean up meeting notes.
• Ask ChatGPT to rewrite a bug or support ticket for clarity.
• Try Cursor to refactor repetitive code.

One trick: Have AI ask you questions first. Instead of just saying “rewrite this,” give it a prompt, like, “Interview me about my goals for this project,” and let it interview you. That back-and-forth helps it understand what you actually need.

There’s also voice mode. ChatGPT’s mobile app makes it easy to talk things out from wherever you are. On your Mac, tools like Wispr Flow let you build voice-driven workflows that feel natural.

Don’t worry about having “the right” process. The goal is to build a habit—a reflex—to include AI in your process.

Step 2: Know how you provide value

Before AI can multiply your work, you need to know what’s worth multiplying.

This is where a lot of people get tripped up. They open a new AI tool and ask, “What can you do?” Instead, start with: “What do I do that matters—to me and to my team?”

If you’re not clear on how you create value—what you’re good at, what your team counts on you for—you risk using AI to generate filler: more words, noise, or decks no one asked for. But if you know your edge, AI becomes a point of leverage.

Try writing on a blank piece of paper or in an Obsidian note:

• What would your team miss if you disappeared for a week?
• What problems do people ask you to help solve?
• What feels easy for you but hard for others?
• What do you, yourself, want to accomplish?

Now pick one of those strengths and ask: How could AI help me do this faster, better, or in a different way than I could have before? Generic AI usage leads to generic results. Push yourself to put your subject matter expertise into words.

Step 3: Develop a documentation habit

If there’s one underrated unlock for getting more out of AI, it’s this: Write things down.

Not for posterity, but for performance. When you combine good documentation with AI, you’re not just saving time—you’re setting up automation.

Think of it like training a new teammate. If you bark vague orders at them—“Do marketing!” “Fix it!”—they’ll flail. But if you give them structure, examples, and context, they’ll understand how to do the job you’re asking of them. The same goes for AI. The people who get the most out of it aren't technical wizards. It’s the clear communicators, managers, and leaders who know how to give good instructions and learn from results.

AI thrives on clarity. It needs structure to do its best work. That structure comes from the context you give it: the way you outline your workflows, capture your steps, and define what “done” looks like.

The next time you are doing something you do often—writing a changelog, triaging bugs, launching a campaign— write down:

• What kicks off your process?
• What steps do you take?
• What does a “good” output look like?

This last point—defining what "good" means—is one of the most critical roles humans play in an AI-powered world. "Good" is deeply contextual and often subjective. What makes a good customer response? What makes a good product feature? What makes a good strategy document? These aren't universal constants. They're shaped by you, your company's values, your team's needs, and your customers' expectations. AI can help you get there faster, but you need to know where "there" is. 

You can capture that context and those expectations in whatever system you normally work: Notion, Google Docs, Voice Notes, Slack, a notebook. The most important thing is creating reusable context, because that document becomes:

• A ready-made AI prompt
• A prototype brief
• An onboarding document for a new teammate
• A starting point for automation

Good documentation turns one-off experiments into repeatable processes, and repeatable processes into opportunities for automation.

Step 4: What you repeat, you can automate 

Maybe you realize you’re using the same prompt to brainstorm new ideas and turn them into actionable plans. Perhaps you keep asking Claude to summarize meetings in the same format, or you always look through your customer feedback using ChatGPT.

You’ve just defined a workflow. Once you have a workflow, you can iterate on it to make it great, and a great prompt is just one step away from becoming a great internal tool or product. Something repeatable, sharable, and scalable. (How do you make a great prompt? Trial and error—and paying attention to what works.) 

For example, I’ve seen clients take their prompts—meticulously crafted to turn news or research papers into key information—and empower a bot that dutifully combs through defined sources before delivering a report to their email each day. 

This is where things start to shift from helpful to high-leverage. Once you notice you’re doing something more than once—and it’s working—you're not just experimenting anymore. You’re building, which is exactly the shift Lütke is pushing for at Shopify. When he says he wants employees to “prototype” with it, what he’s saying is that he wants you to find the leverage points in your day-to-day work and systematize them.

Block one hour this week to review how you’ve used AI in the past month and ask yourself:

• What have I used AI for more than once?
• What worked especially well?
• What took less time than it used to?

Then ask: Can I turn this into a repeatable thing?

• A Claude or ChatGPT prompt template?
• A Slackbot that handles a recurring task?
• A Notion button that auto-generates your weekly update?
• A product that makes you a custom podcast each day?

If it worked once, it can work 10 times, and if it works 10 times, it’s probably worth sharing—or automating. Just make sure it's worth building according to the value you defined in step two.

A common failure mode is experimenting without documenting—you close your chat window before capturing the prompt that worked or the output that was just right. It just takes one more message to ChatGPT to ask it to turn your conversation into a template. Here’s an example:

“Can you turn this conversation into a reusable template? Include the key instructions I gave you, the structure of the output, and suggestions for how someone else could adapt it to their own use case.” 

Try to answer the question from Lutke’s memo: “What would this… look like if autonomous AI agents were already part of the team?”

This question is a forcing function for creative problem-solving. The next wave of groundbreaking AI products will come from people who master their domain first and AI second: Think a designer who builds an AI platform that creates an opinionated style guide from a moodboard; an architect whose AI tools follow them from discussions about materials to 3D renders, enabling seamless exploration; a teacher who uses AI as a way to grade spoken Socratic seminars during which the tool could visualize the conversation or nudge the teacher to encourage a shy student to chime in.  

Your expertise in your actual job is your advantage. You know the problems worth solving because you live them every day. AI is just the new tool in your kit.

Step 5: Share what you learn

By now, you’re using AI regularly. You have a few automated workflows. You’re documenting, experimenting, maybe even prototyping. Now is the time to share what you’ve learned—not when it’s perfect or after you’ve turned it into a polished tool. 

In a company in which employees are learning something new at scale, early transparency becomes leadership. When you share what you’re trying—even if it’s messy—you give your teammates a blueprint, or at the very least, permission to try something themselves.

Pick one thing—an AI-powered workflow, a lesson learned, or a prompt that surprised you—and post it in a public Slack or Discord channel. Or run a 15-minute “here’s what I tried” session at your next team meeting. It doesn’t have to be formal—your goal is to show your process, not just your outcomes. These show-and-tell sessions create vital bridges: The people closest to the problems can explain their needs and solutions, while technical teams can identify opportunities to automate and scale what's working.

Sharing your experiments builds trust and visibility. When performance reviews include questions about AI usage (as Lutke says they will), you won’t be guessing at answers. You’ll have receipts.

Learn AI like it’s your job 

If you take nothing else from this guide, take this: Your next performance review isn’t about whether you used AI. It’s about what AI helped you accomplish.

Getting started with AI is all about overcoming inertia. That first step feels hard because it's new, but humans are remarkably good at making tools part of our nature. We learned to create symphonies from curved wood and string. We turned crushed rocks into computer chips that became windows into all human knowledge. This is just the latest tool—arguably the most powerful one we've ever made, and one of the easiest to adopt. 

You don’t need to build an agent or master prompt engineering. You just need to start the loop: Try, reflect, document, share, repeat.

The people who do that—the ones who start small, learn fast, and make their work visible—are going to define what this AI-powered version of work becomes. Every question you pose to AI, every benchmark you set, every problem you solve shapes how these systems evolve. When you document what "good" looks like, you're training the next generation of AI on what matters. The questions we ask today become the capabilities built tomorrow.

Alex Duffy is the consulting lead and a staff writer at Every, where he writes about empowering people with AI tools and technology in Context Window. You can follow him on X at @theheroshep and on LinkedIn, and Every on X at @every and on LinkedIn.

We do AI training, adoption, and innovation for companies. Work with us to bring AI into your organization.

We also build AI tools for readers like you. Automate repeat writing with Spiral. Organize files automatically with Sparkle. Write something great with Lex. Deliver yourself from email with Cora.

Get paid for sharing Every with your friends. Join our referral program.