I Want to Get Started with AI...
So...how do I actually do that?
I love it, I absolutely love it, when people tell me they want to get started with data or with analytics or with AI, when they break through the idea that it’s scary mathy tech or just tech that will never actually be useful to them and realize that it’s not magic, it’s not hype…it’s just a bag of tools that they can use to do better work.
But then that’s where things go off the rails.
People worry that they won’t be able to figure out how to code fast enough, or that they were never good in their computer science classes, or think that all they have to do is start plugging everything into AI and it will make better decisions for them, and that’s when we have to just STOP. Breathe. Take a minute.
Using AI ten years ago meant you probably needed to be versed in Python, understand probabilistic design and a lot more stats than any of us wanted to understand, and have access to a hell of a lot of compute power. Just like working on a car ten years ago (or twenty) meant you popped the hood open and could access and replace just about everything with a wrench and some elbow grease.
These days, you need to work on your car and you need computers and tools and god knows what else, and you open the hood and there’s just a sheet of shiny plastic. And that’s about how it feels to work with AI as opposed to the old days—you don’t need to get under the hood, you just need to understand how to work with the interface.
So, no, you don’t need to learn how to code (unless you want to, because there are tons of AI tools that will both train you and help you write it). And no, you don’t need to become a machine learning engineer.
You need to have a functional understanding of what AI is and what it does, and figure out what you need it to do to help you work better.
Let’s decode it. 🚀
So you want to use AI…
But you’re not sure what that entails. Let’s break it down.
Step One: Shift your mindset.
Around this time last year, as one of my first articles on The Talent Code, I wrote about finding your role in operationalizing AI in your organization using an analogy I have used ever since (and probably beat to death, sorry not sorry because it works):
AI is a car.
It’s a tool/set of tools. A vehicle. You use it to get where you’re going more efficiently than you would by horse, buggy, or foot.
However, you have to understand how to operate it effectively, and that depends on your role.
Many people believe they need to have a full mathematical and computational understanding of AI, but that’s like needing to understand combustion before driving a car. The functional knowledge you need to use AI is very different.
The functional knowledge you need is not about how the car works in detail. You just need to know enough about its functionality to understand how to drive it. Not the mechanics or physics of how cars convert gas into motion but that pushing the gas pedal makes it go and that it uses fuel to do so.
For AI, at a minimum, these are some of the functions you should know:
AI predicts patterns. It does not “know” truth.
Generative AI produces probabilistic outputs.
AI can hallucinate—it’s a pleaser and will “make up” things to get approval.
You are accountable for anything you submit or publish using it.
That’s enough to begin.
Step Two: Start small, personally.
Before you start creating enterprise AI strategies and plotting how you’re going to integrate it into your workflows, let’s try a few things first. Small things. Low risk things. High frequency things.
Here are just a few things you can try with some example prompts (a lot of these are ones I use—I save them in a Google doc so I can grab them easily and keep consistent summaries and tone):
Draft and rewrite emails.
✳ Example Prompt (Draft)
Act as a senior policy advisor. Draft a concise email to a division chief explaining that our AI pilot is delayed by two weeks due to data integration issues. Keep the tone professional, transparent, and solutions-oriented. Include next steps and an offer to brief live if needed. Limit to 200 words.
✳ Example Prompt (Rewrite)
Rewrite the following email to make it clearer and more executive-ready. Reduce unnecessary detail, remove passive voice, and tighten the language. Keep it professional but direct.
[Paste email]
Summarize long documents.
✳ Example Prompt
Act as an executive briefer. Summarize this 18-page report into a 1-page executive brief for a senior leader. This executive brief should highlight 1) key findings, 2) risks, 3) decision points, 4) recommended actions. Keep it under 500 words.
[Paste document]
Convert meeting notes into structured outlines.
✳ Example Prompt
Convert the following raw meeting notes into a structured outline. Organize into 1) decisions made, 2) open questions, 3) assigned actions with owner and deadline if stated, 4) risks or concerns. Make the format clean and easy to share.
[Paste notes]
Brainstorm outlines or frameworks.
✳ Example Prompt
Act as a workforce transformation strategist. Help me develop a simple 4-part framework explaining how AI impacts organizational design. Include a short label for each pillar and a 2–3 sentence explanation. The audience is federal senior leaders who are new to AI.
Compare policy language.
✳ Example Prompt
Compare the following two policy excerpts. Identify key differences, overlapping language, potential conflicts, and areas requiring clarification. Present the output in a two-column comparison table with a section for each of these areas.
[Paste Policy A]
[Paste Policy B]
Create checklists from narrative documents.
✳ Example Prompt
Convert the following narrative policy guidance into a practical compliance checklist for supervisors. Organize into actionable steps. Use clear, directive language.
[Paste narrative document]
Translate technical writing into plan English.
✳ Example Prompt
Rewrite the following technical explanation in plain English for a non-technical audience. Keep it accurate but remove jargon and spell out all acronyms. Offer definitions when possible, and break those out for use in a sidebar or callout box. Use simple language and short sentences.
[Paste technical text]
My prompt formula.
To get my AI tools to give me the results I want, I try to use this formula:
Role + Task + Constraints + Output Format + Audience + Tone
You’ll see that in most of the prompts above. To figure this out, when you’re tasking your AI tools to create something for you, ask these questions:
What perspective do you want the tool to take for what audience?
What is the task and what are the outputs?
What limits do you want it to have on the output or when accomplishing the task?
What format do you want the output in?
Do you want it to take a certain tone (optimistic, scientific, etc.)
I’ll often tell my AI tools to come up with three different options or recommendations and then work through a couple more iterations to get closer to what I want. Chances are pretty good I’m still going to go through and edit the thing myself—for content and clarity if we’re talking meeting notes, for voice if it’s a memo or paper—but still, having a tool create a rough outline for you and help you brainstorm what content you need to include in a document or email saves a lot of time.
If those feel a little too…”workish”…try these.
Help me figure out what to eat for dinner.
Upload a photo of your fridge or pantry contents and ask for 3-5 dinner ideas using only what is visible, or plan out meals for the week to use up what’s in the fridge and recommend a consolidated grocery list to complete the meals.
Try a prompt like this:
✳ Example Prompt
Based on this photo of my fridge, suggest 3 healthy dinners I can make this week using mostly what I already have. Highlight what ingredients need to be used soon and create a short grocery list to fill any gaps.
I do this a lot. When you’ve spent the entire day making decisions on things (the average person makes 36,000 decisions a day!), sometimes you just don’t want to decide what to fix for dinner.
Help me figure out what to do on a trip.
Give it your flight arrival time and where you’re staying, what kind of time you’re working with, how much walking you want to do, mention any dietary restrictions you might have, and give it your general vibe. Are you a museum or a theater nut? Foodie? Want to check out some top wine bars or tasting venues?
Try something like this:
✳ Example Prompt
I’m landing in Denver at 3pm on Friday and staying downtown. I leave Sunday at noon. I have activities planned for Saturday afternoon and night but the rest of my time is unstructured. I am looking for good coffee, interesting bookstores, and one memorable dinner that isn’t touristy. I don’t want to rent a car. Build me a flexible, walkable weekend plan with backup indoor options in case of weather.
You can also ask it to put together a packing list based on weather. A whole lot more fun than sitting in your hotel and using the “around me” function on your map app.
Build your life systems.
Workouts, family chore rotations, study plans for certifications (hello, SHRM-SCP and CAP, two I am working hard on right now), date night idea banks, home decluttering plans…you can build all of these things using AI.
I use it a lot to fit workouts into my available time, or to build workout plans for the hotel I’m staying on while on a work trip. Here’s an example for that:
✳ Example Prompt
I am at a hotel on a trip for four days where I have access to dumbbells up to 40 lbs. Taking up to 45 minutes per day, build me a 4-day strength-focused workout plan that incorporates the same progression elements of my daily home gym workout plan but with dumbbells only.
[paste current daily home gym workout plan]
Here’s another one that I use to reduce my Sunday scaries.
Help me design a Sunday reset routine that takes 90 minutes and prepares my home and schedule for the week. Break it into timed blocks. Tell me where I can introduce automation to scrub my calendar for conflicts. Build in time to address priority emails and administrative actions.
Have fun.
AI isn’t just for fun image generation. You can turn it into something fun and creative. Turn your vacation into a “Top 10 Highlights” newsletter for family, feed it a half-baked idea and have it help generate a plot for a novel or a series pilot, generate your work Spotify playlist theme for the week.
Here’s one I did for my son about my dog to help him understand my work:
Write a short children’s bedtime story starring a golden retriever named Darby who learns how to be brave on his first day at the office, helping people solve their HR problems.
Why start personally? Because when you’re just thinking about AI for work, you overthink it. We’re all worried about breaking something or getting in trouble or getting judged at work in a way we’re not when we’re just looking at the stuff we do around the house.
Think of it as giving yourself the freedom to let out your inner kid, the one who wasn’t afraid to experiment and try new things.
Step Three: Find the “machine work” in your role.
Once you’ve played around with AI in small scale, step back and think about the work that you do in your role that feel mechanical, repetitive, and routine.
Ask yourself:
What repetitive tasks do I perform weekly?
What structured outputs do I generate repeatedly?
What formatting or compliance work drains my time?
What summaries do I rewrite over and over?
These are AI-ready tasks. AI is good at pattern recognition, reformatting, summarizing, generating rough drafts, and creating structured content. If your office is fielding AI tools that do this for you, consider feeding these tasks to those tools.
In our case, the Department of War fielded GenAI.mil, a suite of commercial large language models geared to help folks with this kind of work. Memo summaries, policy comparison, formatting, and more.
Now I’m just waiting for them to get Copilot working or partner with one of the other companies that converts info papers to slide decks so I don’t have to make slides while working on the gov side. I hate making slides.
Step Four: Learn to prompt strategically.
No, I’m not saying you need to figure out what it takes to be certified as a prompt engineer (because I’m pretty sure someone has come up with a certification and a bunch of courses for it), but you do need to be clear.
Think of AI as Drax the Destroyer from Guardians of the Galaxy.
Bro is extremely literal. He is going to take what you say the way you say it and that’s what he’s going to do.
AI? Similar. It’s getting better at recognizing tone, but basically AI is going to do what you tell it to do, so you need to be clear about what you want and what you don’t want, how you want the result delivered, and what you’re trying to do.
As I mentioned earlier, AI performs best when your prompts include:
Context
Objective
Constraints
Output format
Remember, here’s my prompt formula:
Role + Task + Constraints + Output Format + Audience + Tone
For example:
“Act as a policy analyst. Summarize this 12-page memo into a 1-page executive brief for a senior leader. Highlight risks and decision points. Use bullet format.”
That’s it.
And don’t be afraid to correct it if it gets it wrong and try to fine tune the result.
Good prompting is just good communication. That’s all coding is—a conversation with the computer in machine language, except with the interfaces we have these days, you don’t need the machine language (as much).
Step Five: NOW you can start designing.
Once you’re comfortable using AI for yourself, you’ll start seeing opportunities everywhere. Now you can start looking at your team’s workflow, finding tasks that are predictable and repetitive, and looking at where judgment still needs to remain human.
There’s another framework in here that you need to know, though—and that’s the one my team uses to redesign business processes along with automation.
ELIMINATE - SIMPLIFY - AUTOMATE - ELEVATE
We’ve got a lot of details on how that works in this article here:
But basically if you don’t restructure your business processes as you introduce automation, eliminate what you no longer need to do, simplify the chain in what you do, then automate, and elevate the human element, you’re going to get some efficiency…but not enough to justify the expense of the tools you’re investing in (AI is a lot cheaper than it used to be, but it ain’t free!).
Step Six: Be critical in your judgment.
AI doesn’t replace you. It elevates you. Which means you need to critically judge its outputs. The more you use AI, ask yourself:
Does this output make sense?
Is it missing nuance?
Is it overly confident?
Could this introduce bias?
Would I stand behind this decision?
Because even if AI assists you, you are the one who owns the consequences.
You also need to be critical in your judgment of what you should and shouldn’t be doing. I mean, even though it’s easy to build code with AI, no one is going to judge you if you haven’t gone out and created your own app or built a chatbot from scratch. Your question should be more along the lines of:
What do I need to be able to do with AI within my role?
If your job is workforce design, leadership, or operations, your role is to understand what AI can do, what it shouldn’t do, where it fits, and how to govern it.
Knowing how the tech works can help with that, but let’s be realistic about what you have time to focus on—until you can claw back some efficiency with AI, that is!
Sound like fun? Let’s create a 30 day AI starter plan!
So the first week, you’re going to…
Ha. Nope. We’re going to do this a little differently.
Here’s your prompt. Go to the AI of your choice, copy/paste/edit, and let’s start building your “learn by doing” plan for AI!
Act as a coach helping non-technical professionals build confidence using AI tools. Create a 30-day “learn by doing” plan to help me get comfortable using AI in my daily work and personal life.
(Customize This Section):
I work in [HR / talent management / workforce planning / recruiting / learning & development / other]. My typical responsibilities include: [list 3–5 tasks you do regularly]. I have basic familiarity with AI but no technical background. My objective is to become comfortable with AI and understand how it can fit into various tasks I do around my house so I can think better about how to use it at work.Keep daily tasks small (10-20 minutes). Start simple and gradually introduce complexity. Start with personal use cases but gradually start including work. Include reflection prompts each week to build judgment. Assume I am cautious about data privacy and governance. Please keep the instructions friendly, clear, encouraging, and non-technical.
Output Format (Choose One):
Option A: A simple numbered 30-day list
Option B: A printable checklist with checkboxes
Option C: A calendar-style weekly layout (Week 1–4)
So should I take a class?
You can take classes, or you can learn by doing. Classes are great to give you a primer about what the tools are, but they aren’t so good at telling you what they’re for, or building comfort with using them.
That’s what I’m aiming to do here. To get you trying things and figuring out new uses, if you aren’t all that experienced with AI or you’ve maybe tried using ChatGPT to make a couple images but that’s about it, or even if you are more experienced and are looking for ways to train your teammates.
Start small. Rewrite an email, build a checklist.
Experiment. Notice what works and what doesn’t.
Read widely.
Stay curious.
And most importantly, talk about it. Share what you’re doing and what works and what doesn’t. Ask others what they’re trying. Compare notes.
The future of work won’t be shaped by the people who took an intro class. It will be shaped by the people who experiment and apply. And the best user of AI are the ones who will engage with it thoughtfully, responsibly, and consistently.
And this is how you get started with AI.
