Last Tuesday, a product manager in our community told me she’d been “meaning to start using AI” for eight months. She has a Claude Pro subscription. She’s opened it maybe five times. Every time, she types something in, gets a result that feels… fine, closes the tab, and goes back to the way she’s always worked.
She’s not lazy. She’s not behind. She just doesn’t know where to start when “use AI at work” could mean a thousand different things.
If that sounds familiar, this guide is for you.
Why “use AI at work” is harder than it sounds
Most advice about using AI at work boils down to “just start using it.” That’s like telling someone who wants to get in shape to “just go to the gym.” Technically true. Practically useless.
The problem is that AI tools like Claude and ChatGPT are general-purpose. They can do almost anything, which means they give you no obvious place to begin. You open a blank chat window and think: “Now what?”
This is actually a new kind of challenge. Every other tool you’ve adopted at work had a specific function. Excel does spreadsheets. Slack does messaging. Zoom does meetings. You knew what the tool was for, so you knew how to practice.
AI doesn’t work that way. It’s more like hiring a new team member who’s brilliant but has no idea what your job involves. You have to tell it what to do, give it context, evaluate its work, and push back when it misses the mark. That’s what we mean when we say using AI is a managerial relationship: you’re managing inputs, goals, and outputs. Not pushing buttons.
Once you see it that way, the question shifts from “how do I use AI?” to “what parts of my work would benefit from a capable assistant who needs clear direction?”
That second question is much more answerable.
The four types of work AI can help with
Before you open Claude or ChatGPT, it helps to have a mental model for where AI actually fits into professional work. Not every task is a good fit. And the tasks that are good fits might not be the ones you’d expect.
Here’s a framework we use in our coaching practice. Think of your work in four categories:
1. Drafting work (the blank page problem)
This is any task where you’re starting from nothing and creating something: emails, reports, proposals, presentations, project briefs, status updates. The blank page is where most people waste the most time, not because the work is hard, but because getting started is hard.
AI is excellent at first drafts. Not because it writes better than you (it often doesn’t), but because it gives you something to react to instead of starting from zero. Editing is always faster than creating.
Where to start: Open Claude or ChatGPT and paste in your last meeting notes. Ask it to draft a status update for your manager. See what comes out. Then edit it until it sounds like you. That editing process, where you push back on what the AI gave you and shape it into something real, that’s the core skill.
2. Synthesizing work (the information overload problem)
You have a 40-page report to read before tomorrow’s meeting. You have 200 emails to process. You have six competing proposals to compare. You have meeting transcripts piling up.
Synthesis is one of AI’s genuine strengths. It can read a long document and pull out the three things that matter to you specifically, if you tell it what matters to you. The key word is “specifically.” Asking Claude to “summarize this document” gives you a generic summary. Asking Claude to “read this proposal and tell me the three biggest risks to our Q3 timeline” gives you something you can actually use.
Where to start: Take a document you need to read this week. Paste it into Claude. Instead of asking for a summary, ask a specific question about it. “What are the cost implications?” or “What assumptions is this proposal making about our team’s capacity?” You’ll be surprised how much faster you can prepare for meetings this way.
3. Thinking-through work (the reasoning partner problem)
This one surprises people. AI is genuinely useful as a thinking partner, not because it has better ideas than you, but because it can help you build scaffolds for complex thinking.
You’re preparing for a difficult conversation with a team member. You’re trying to figure out why a project keeps stalling. You’re weighing two strategic options and can’t see past your own assumptions. These are all places where talking it through with Claude or ChatGPT can surface angles you hadn’t considered.
Where to start: Next time you’re stuck on a decision, open Claude and describe the situation. Not “make this decision for me” but “here’s what I’m weighing, here’s what I’m worried about, help me think through the angles I might be missing.” Treat it like talking to a thoughtful colleague who has no political agenda.
4. Process work (the repetitive task problem)
This is the category most people think of first, but it’s actually the last place I’d recommend starting. Process work means tasks you do repeatedly in the same way: formatting data, generating weekly reports, filling out templates, converting information from one format to another.
AI can absolutely help with repetitive work, and the time savings can be significant. But it’s the last place to start because it often requires more setup, more specific prompting, and more iteration to get right. Save this for after you’ve built some comfort with the first three categories.
Where to start: Identify one task you do every week that follows the same basic pattern. Write out the steps you take. Then ask Claude to help you build a reusable prompt that handles the routine parts. This is where 20 minutes of setup can save you hours over time.
Your first week: 20 minutes a day
The biggest mistake people make is trying to “transform their workflow” all at once. That doesn’t work. What works is small, consistent practice that builds on itself. Here’s what I’d suggest for your first week.
Monday: Pick one email (10 minutes)
Find an email you need to write today. Something real, not a test. Open Claude or ChatGPT and give it context: who you’re writing to, what the situation is, what you want the email to accomplish, and your general tone.
Read what it gives you. It probably won’t sound like you. That’s fine. The point isn’t to send what the AI wrote. The point is to have a starting point that you can edit. Most people find they can write a polished email in half the time when they start from a draft instead of a blank screen.
Tuesday: Prep for a meeting (15 minutes)
Take the agenda or pre-read for an upcoming meeting. Paste it into Claude. Ask three specific questions about the material: “What are the three most important decisions we need to make in this meeting?” or “What’s the strongest argument against this proposal?”
Use the answers to prepare your own talking points. You’re not outsourcing your thinking. You’re giving yourself a head start.
Wednesday: Summarize your day (10 minutes)
At the end of the day, open Claude and describe what you worked on. Ask it to help you write a quick summary of your priorities for tomorrow, or draft a brief update for your team. This is a low-stakes way to practice giving AI context about your work.
Thursday: Think through a challenge (20 minutes)
Pick something you’re stuck on. A project issue, a team dynamic, a decision you’re putting off. Describe the situation to Claude in detail and ask it to help you think through your options. Push back on its suggestions. Ask follow-up questions. This is where the “managerial relationship” starts to click: you’re directing a conversation, not just asking for output.
Friday: Reflect on the week (10 minutes)
Open Claude and tell it what you tried this week. What worked? What felt awkward? What would you want to try next week? This isn’t busywork. The reflection helps you identify which use cases are actually saving you time versus which ones feel like novelty.
[NEEDS COMMUNITY QUOTE: A member sharing what their first week of daily AI practice felt like, the specific moment it clicked]
Common mistakes (and what to do instead)
After coaching over a hundred professionals through their first weeks with AI, here are the patterns I see over and over.
Mistake 1: Starting with the hardest task
People hear “AI can help with anything” and immediately try to automate their most complex, high-stakes work. That’s like learning to drive on a highway. Start with low-stakes tasks where a mediocre result doesn’t matter. Drafting an internal email is a much better starting point than writing a client proposal.
Mistake 2: Giving no context
“Write me an email” is the most common first prompt I see. It’s also the least effective. AI doesn’t know who you are, who you’re writing to, what your relationship is, what you’re trying to accomplish, or what tone is appropriate. The more context you give, the better the output. Think of it this way: if you handed this task to a new hire on their first day, what would you need to tell them?
Mistake 3: Accepting the first output
Most people type a prompt, read the response, and decide AI either “works” or “doesn’t work” based on that single exchange. That’s not how the tool is designed to be used. The first output is a starting point. The real value comes from the conversation: “That’s close, but the tone is too formal. Make it more direct and cut the second paragraph.”
AI is a do-first, understand-later tool. The understanding comes through iteration, not through getting it right on the first try.
Mistake 4: Trying to learn AI “properly” before using it
I see this one constantly, especially with experienced professionals. They want to take a course, read a book, watch a tutorial series, and feel prepared before they start. But AI is a practice skill, not a knowledge skill. You will learn more in one week of daily use than in a month of reading about it. The understanding comes through doing, not preparation.
Mistake 5: Going it alone
Using AI can feel isolating in a weird way. You’re having these conversations with a tool, getting results nobody else sees, and you’re not sure if what you’re experiencing is normal. Am I using this right? Is this output good enough? Should this be taking me this long?
Having a community of people figuring it out alongside you changes everything. You see approaches you wouldn’t have thought of. You realize your struggles are normal. You pick up techniques from watching how other people work.
From the practice: what we’re seeing in coaching
Every week, we run community sessions where professionals share what they’re building and where they’re stuck. A few patterns keep showing up.
The “aha” moment isn’t about prompting. People expect the breakthrough to be a clever prompt trick. It’s almost never that. The breakthrough is usually the moment someone realizes they can paste their actual work into Claude (their real meeting notes, their real project brief, their real email thread) and get genuinely useful output. The tool becomes real when the context becomes real.
The second week matters more than the first. Week one is exciting and novel. Week two is where people either build a habit or drift back to their old patterns. The ones who stick with it are usually the ones who found one specific use case in week one that saved them real time, even if it was just 10 minutes.
People underestimate how much context AI needs. The single most common issue we see is people providing too little context and then being disappointed with the output. When we sit with someone and help them build a proper prompt (with background, audience, purpose, constraints, and tone), they’re often shocked at how different the result is. It’s not about AI getting better. It’s about you getting better at directing it.
[NEEDS COMMUNITY QUOTE: A specific member describing a use case that saved them meaningful time, with concrete details about the task]
[NEEDS REAL EXAMPLE: A specific before/after from a coaching session showing the difference between a low-context prompt and a high-context prompt with the resulting outputs]
What happens after the first week
If you follow the daily practice above, you’ll start noticing something by the end of week two or three. You’ll start seeing opportunities without looking for them. A report lands in your inbox and you think, “I bet Claude could pull out the key risks in this.” A colleague asks for feedback on a proposal and you think, “Let me run this through Claude first to sharpen my thinking.”
That shift, from “I should try to use AI today” to “AI is just part of how I work,” is what we’re after. It doesn’t happen through a workshop or a training session. It happens through consistent, daily practice with your actual work.
Here’s what the progression typically looks like:
Weeks 1-2: You’re consciously deciding to use AI. It feels like an extra step. Some tasks go faster, some feel clunky. This is normal. You’re building a new habit, and new habits feel awkward at first.
Weeks 3-4: You start reaching for Claude or ChatGPT without thinking about it. You develop a few go-to use cases that reliably save you time. You get better at giving context, and the outputs get noticeably better as a result.
Month 2 and beyond: AI becomes part of your workflow the way email or your calendar is. You stop thinking about “using AI” as a separate activity. You also start noticing tasks where AI doesn’t help, which is just as valuable as knowing where it does. You develop judgment about when to use the tool and when your own expertise is what the situation needs.
The professionals who are getting the most from these tools right now aren’t the most technical people. They’re the ones who show up every day, try something, learn from what works and what doesn’t, and keep going. It’s the same principle behind everything we do: the skill develops through practice, not preparation.
The $20 question
One practical note: the free versions of Claude and ChatGPT are fine for experimenting. But if you’re serious about building AI into your work, the paid tier ($20/month for Claude Pro or ChatGPT Plus) is a different experience entirely. Faster responses, better reasoning, longer conversations, access to the latest models.
Think of it this way: $20/month is less than most people spend on coffee in a week. If it saves you even 30 minutes a week (and it will save you more than that), the math is obvious. This is the most affordable professional development investment you’ll find right now.
And you don’t have to figure it out on your own. This is going to be the most disruptive moment of our professional careers, and it is better to go through this moment together than alone.
Start here
You don’t need to overhaul your workflow. You don’t need to become a prompt expert. You don’t need to understand how large language models work under the hood.
You need 20 minutes, a real task, and a willingness to try.
If you want to practice alongside other professionals who are figuring this out in real time, that’s what we built MVP Club for. We’re not a course. We’re a community of people who believe that AI belongs to all of us, and that getting good at it is more of a human challenge than a technical one.
Come see what we’re building. Take the AI readiness assessment to see where you stand, or just show up to a session and see how it feels.
We’d love to have you.