AI created hybrid worker schedule

This remote/in-office schedule is one of three case tasks I requested from GPT and will show you how to do to automate normally tiresome tasks for hybrid managers and workers.


With all the big-picture discussions about how AI is going to change the future of work, it’s easy to lose sight of how AI could change our working lives right now. But the more time I spend using ChatGPT, and honing the skill of coaxing really useful responses from the machines, the more I see the ways AI could help address the biggest pain points we experience as hybrid team managers, or as hybrid and remote workers.

I’m going to share three of these today, with some pretty detailed examples—because the “aha!” moments with AI happen when people see the nitty-gritty of how to formulate prompts to get effective results. Throughout this newsletter, I’ll highlight tips on how to get the most from working with AI — even if these aren’t the three problems you personally need to solve.

Much of this comes down to what’s termed “prompt engineering”: Coming up with the right framing and cues to produce useful results from the AI. This post will offer you lots of ideas about how to write effective prompts for a range of workplace situations.

NOTE: This week’s post focuses on working with ChatGPT, which is the interface I use most often when working with GPT-4, a product of OpenAI. If you’re doing work that requires real-time or recent information from the Internet, the new Bing Chat interface is also powered by GPT-4. 


Roadblock #1: Designing a hybrid schedule.

So much of the hybrid transition comes down to figuring out exactly who is going to be in the office on which days. The idea of “pick your own schedule” seems great in theory—until you schlep to the office, only to discover that everyone you want to meet with is at home. And if you avoid that by just having everyone come into the office on the same two or three days, you lose out on the potential cost savings and environmental benefits of reducing the size of your office space.

Ideally, you’ll come up with a coordinated team schedule that provides plenty of opportunities for in-person collaboration among the people who work together most—so that home days can be mostly meeting-free, which is better for focused work and personal flexibility. But coming up with a schedule that reflects a range of professional relationships and personal priorities isn’t easy—unless you are an AI!


GPT to the rescue!

I took ChatGPT’s planning powers for a spin by whipping up a spreadsheet that represented the collaboration needs of an imaginary 8-person team, as well as the requirements and preferences of each team member. Then I used a series of “concatenate” formulas in Google Sheets to turn those into a description of each person’s needs.

The near-instant result? A 5-day-a-week game plan for how to optimize the team schedule in a way that would support in-person contact, minimize the need for video calls on at-home days and provide for a couple of meeting-free windows each week.

I was so impressed by GPT’s ability to solve this brainteaser that I turned my spreadsheet and workflow into this Hybrid Team Planner, which you can copy and use on Coda.

Snapshot of landing page for Hybrid team planner. Visit for a machine-readable version of the live page.

This interactive Hybrid Team Planner was built in Coda, and is powered by GPT—and I created it in a day. It’s a handy tool, and an example of what’s now possible to build with zero programming, just by integrating a little AI.


How to use AI to solve other work problems.

You can adapt the strategies I used to build this hybrid team planner whenever you have a problem you want to tackle with the help of AI. Keep in mind:

  • When you have a work challenge that reads like a logic puzzle, consider giving it to ChatGPT. This is where AIs have a real advantage!
  • Spring for a ChatGPT Plus account to get access to GPT-4, the latest model. It’s a dramatic improvement over GPT-3.5, and the impact it can have on your work makes it the best $20 you’ll spend each month. (Try it for a month, and let me know what you think!)
  • Use spreadsheets to compile multiple factors or data points into a single natural-language query. The “concatenate” function is your friend: It can take ingredients from different columns or rows and compile them into the kind of natural-language query ChatGPT expects. If you don’t know how to use it, ask GPT to write your formula.
  • Begin your prompt by giving ChatGPT a specific job title or role, so it knows how you want it to think about its assignment.
  • Tell GPT the format you want for your result; if you want a table, tell it the specific columns and rows you want, and give it an example row: Please format my results as a table, with one row for each company, and columns for COMPANY NAME | COMPANY HQ LOCATION | NUMBER OF EMPLOYEES


Roadblock #2: I miss riffing with my colleagues.

People often tell me that when they’re working remotely, they miss the spontaneity of the office: they miss the opportunity to pop down the hall when you have a quick question, or to bounce an idea off somebody else for a quick second opinion, or to ask for help developing an idea. And when it comes to brainstorming a new project, product, or strategy, there’s really no replacement for a group of creative people putting their heads together for an hour, ideally in a room with a giant whiteboard.

That’s what I thought until recently—when I started using Bing Chat and ChatGPT to take on the role of in-house sage or creative muse. Instead of asking a co-worker to re-read my drafts or double-check my Excel formulas, I ask an AI. (They never complain about being interrupted.)


What AI brainstorming looks like.

ChatGPT has become my go-to brainstorming partner, especially for the parts of writing I find trickiest: picking titles and subtitles for my work, and drafting conclusions (the literal worst!). Here are snapshots of the process I used to get a title for this post:

SNAPSHOT OF CHATGPT INTERACTION: Prompt: You are an email marketing consultant who is paid based on the open rates and forwards of eachnewsletter you manage. You are managing a newsletter about the future of work that focuses on hybrid work, generative Al and the future of work. The latest issue has a working title, "Three hybrid work problems Al can solve", and shares examples of using Al to craft a hybrid team schedule, facilitate brainstorming and look for a new job. Can you suggest 10 titles and sub-titles that will optimize open rates for this newsletter? GPT REPLY: 1. Title: "Hybrid Work Revolution: Al Unlocks the Future" Sub-title: "Discover How Al is Transforming Team Scheduling, Brainstorming, and Job Hunting in Hybrid Work Environments" 2. Title: "Unlocking Hybrid Work's Full Potential with Al" Subtitle: "Revolutionize Your Schedule Management, Collaborative Thinking, and Job Search Strategies" PROMPT: Wow, those are some really generic and un-sexy titles, and your subtitles are SO LONG. Let me give you some examples of other recent newsletter subject lines, and you come up with 5 new options for "3 hybrid work problems Al can solve". Subject Variety is the spice of work The Robots are Coming: How to prepare for the future of work with ChatGPT TV that works harder: 8 ways TV can boost productivity The Joy of Tedious Work The Hybrid Worker's Guide to Information Overload Reinventing organizational culture What's your hybrid plan for 2023? [Year Compass/Hybrid planner]
Just like when you’re brainstorming with a human, getting great results from AI brainstorming takes time. You can see how much back-and-forth it took to get to this week’s subject line by reading the entire chat here.  Now that I’ve taught GPT what I want in a subject line, it’s much quicker to suggest great options for subsequent newsletters.


The secrets of brainstorming success.

As I’ve tapped ChatGPT for an ever-wider range of brainstorming challenges, here’s what I have found effective:

  • Tell GPT what its top consideration should be when it’s giving you ideas (I told it to optimize for email opens and forwards).
  • Give it a descriptive starting point and then tell it how you want to do better than your starting point. Let it know that you want options that are funnier, more consumer-friendly, more powerful, etc (you can ask for that even if you’re looking for ideas rather than wordsmithing).
  • Give it examples and documentation that it can use to guide its brainstorming.
  • Try and try again. It took twenty rounds of back-and-forth for me to get to the right subject line. That didn’t take that long, however, because GPT iterates quickly.
  • Make it do tricks. If you were brainstorming with a human, you would probably stump your colleague if you asked for product name ideas that were puns on the names of famous physicists, or if you said you only wanted 5-word report titles where at least 4 of those words began with the same first letter. But GPT excels at these kinds of stunts.
  • Provide feedback throughout the process. Tell GPT what you do and don’t like about its various suggestions so it can do better with each round.
  • Keep your own voice human. If you give commands like you’re talking to a robot, your own creativity will stay limited. Riff as if you’re talking to a person and you’ll keep your own creative juices flowing.
  • Train for re-use. ChatGPT treats each brainstorming session as one extended conversation, and forgets what it’s learned when you start a new session. But you can resume a previous session by going to your history in the sidebar, so it’s worth investing extra time in back-and-forth if you’re creating a session you can re-open and re-use for similar purposes in the future.


Roadblock #3: I need a new job.

Sometimes, the only way to fix an unsatisfying hybrid work arrangement is to find a role, organization, or boss who suits you better. 

If it’s time to start job-hunting, ChatGPT can help! A friend recently asked me for advice on how to use GPT to draft cover letters, and I tested a few strategies that can help generate effective letters customized for each position—which makes it easy to apply for lots of different jobs.

Here’s the kind of starter prompt I recommend:

You are a talented and articulate human resources consultant and ghostwriter who is working for a new client: [Description of applicant] who is applying for [type of job] jobs that will advance them to the next level in their career. You need to write application cover letters in their voice, reflecting their particular strengths and underlining their qualifications for each specific position. We will provide you with each job posting and ask you to write a customized letter for that opening, but first, you need to get comfortable writing in the voice of the applicant, and you need to understand their qualifications. We will provide you with their résumé; use this to review their qualifications. Then stand by for an example of a job posting this person has already applied to, and the cover letter they wrote for it.

Then paste in the résumé, sample posting, and sample letter, probably over the course of a couple of messages, telling GPT to stand by in between. When all the background info has been sent, finish with something like:

Now, review this job posting and draft a cover letter in the same voice as the example above, but tailored to this particular posting. Please highlight past experiences or skills that you noticed in the applicant’s résumé that you think are relevant to this posting. Be sure to note: [add 3 or 4 bullets about the specific things you want GPT to highlight in the letter]

Here are the bullets I gave ChatGPT when I challenged it to draft a cover letter for a job writing consumer advice for pet owners:

Portion of GPT prompt reading: "// IN THE COVER LETTER, HIGHLIGHT: - Alexandra's experience writing review-driven tech and travel stories for the Wall Street Journal - Alexandra's longstanding admiration for the Wirecutter (she even wrapped all her Christmas presents in homemade wirecutter wrapping paper and ribbons) - Alexandra's expertise on pets as a longtime dog owner, trainer of an autism support dog and daughter of a dog breeder"

And here’s the letter GPT was able to write based on my résumé, sample letter and the job posting I provided:

CHATGPT interface shows draft letter: "Dear Hiring Manager: lam excited to apply for the staff writer position at Wirecutter, specifically to cover products and advice for pets. As a long-time fan of the publication, I was thrilled to see an opening in the area that is both a personal and professional passion of mine. My experience writing review-driven tech and travel stories for the Wall Street Journal has instilled in me a deep appreciation for research and the ability to explain complex topics in an accessible manner. My passion for writing extends beyond the field of tech and travel, and into the world of pets. As the daughter of a dog breeder, a longtime dog owner, and a trainer of an autism support dog, I have an intimate understanding of the unique needs of pets and their caretakers. I believe that my expertise in these areas makes me the ideal candidate for this role. At the Wall Street Journal, I was responsible for writing a range of pieces that aimed to simplify complex technology issues for readers. In doing so, I became an expert in product evaluation and the creation of comprehensive how-to guides. I also possess the necessary skills to collaborate with editors and other writers, create test plans, and bring ideas from pitch to publish. My experience writing for a large publication has also taught me the importance of responding to reader feedback and questions, and I am excited to continue that tradition at Wirecutter. I am excited to bring my experience to the world of pet products and advice. As someone who has long admired Wirecutter's commitment to providing in-depth, accessible product reviews, I am excited at the prospect of working with a passionate team to help pet owners make informed decisions about their pet's needs. Thank you for considering my application. I look forward to the opportunity to further discuss my qualifications. Sincerely, [Your Name]"


Hey, I’d hire her!

Getting a letter this good did require a little thinking: In my first try, I trusted GPT to write a decent letter based on just the posting. It wasn’t until I added the bullets for ChatGPT to cover that I got good results—so good I am half-tempted to apply to the Wirecutter’s posting for a pet writer.


Writing great drafts with GPT.

The strategies I used to get solid cover letters from ChatGPT can be applied to many situations where you want GPT to give you a complete draft:

  • If you need to give GPT a long briefing before asking for help, tell it to “stand by” as you add additional pieces in subsequent messages. A sample résumé, job posting, and cover letter, followed by the job posting you want a letter drafted for, will together amount to a longer prompt than GPT can take in one input. But you can split them over two or three inputs, and then ask GPT to give you its results.
  • If you run out of messages in the course of this process, take a break. Even ChatGPT Plus subscribers can only access 25 GPT-4 messages every 3 hours; when you reach the cap you’re bumped down to GPT-3.5. At that point, it’s better to pause and then come back to the session later (it will be in your sidebar history) rather than continuing. Otherwise, you’ll feel like your HR consultant was just lobotomized mid-session.
  • Stop generating if GPT is off-track. Sometimes GPT just gets it wrong, so if you see that it’s going in the wrong direction, hit “stop generating” and then provide some feedback on what was wrong with its initial effort and what you want it to do instead.
  • 10% goes a long way. If you’re prepared to put in 10% of the effort to create your draft—5% in giving GPT some context and bullets, 5% in cleaning up the final draft to sound more like what you want—GPT will get you the rest of the way.


Problems, be gone!

Yes, ChatGPT can be an amazing resource in addressing the problems that arise around hybrid work—but the strategies I describe above can be used in a near-infinite range of contexts. Whatever your professional frustrations or opportunities, it’s time to start exploring the ways generative AI rewrites the possibilities. You’ll only find out if you give it a try, and then keep on trying.

This post was originally featured in the Thrive at Work newsletter. Subscribe here to be the first to receive updates and insights on the new workplace.