How custom GPTs can make you a better manager | Hilary Gridley (Head of Core Product at Whoop)

Hilary Gridley, Head of Core Product at Whoop, shares how she uses dozens of custom GPTs for her team that think and give feedback like her, allowing her to scale herself up and create time for higher-value work. What you’ll learn: 1. A step-by-step process for creating GPTs that “think like you” by reverse engineering your own decision criteria 2. How to turn your management expertise into clear evaluation rubrics that AI can consistently apply 3. Practical techniques for improving team writing and presentations with AI-powered feedback 4. Why GPTs are the perfect tool for scaling good management practices without requiring prompt engineering skills 5. How to use AI to get invited to more strategic meetings by improving your written point of view — Brought to you by: Orkes https://www.orkes.io/—The enterprise platform for reliable applications and agentic workflows Vanta https://www.vanta.com/howiai—Automate compliance and simplify security — Where to find Hilary Gridley: Newsletter: https://hils.substack.com/ LinkedIn: https://www.linkedin.com/in/hilarygridley/ X: https://x.com/yourgirlhils — Where to find Claire Vo: ChatPRD: https://www.chatprd.ai/ Website: https://clairevo.com/ LinkedIn: https://www.linkedin.com/in/clairevo/ X: https://x.com/clairevo — In this episode, we cover: (00:00) Intro (02:52) Creating GPTs that think like you (04:23) Demo: Reverse engineering a recommendation algorithm (13:06) The value of articulating taste (15:33) Demo: Creating a slide deck evaluator GPT (19:19) Testing your new GPT (21:32) Scaling GPTs across your team (23:52) Demo: Using AI to improve your writing (30:32) Lightning round and final thoughts — Tools referenced: • GPTs: https://chat.openai.com/gpts • ChatGPT: https://chat.openai.com/ • Claude: https://claude.ai/ • Bolt: https://bolt.new/ — Other references: • Whoop: https://www.whoop.com/ • Norwegian School of Economics: https://www.nhh.no/en/ • Researchers at NHH have uncovered significant gender disparities in the adoption of generative AI tools like ChatGPT: https://www.nhh.no/en/nhh-bulletin/article-archive/2024/september/study-reveals-gender-gap-in-ai-tool-usage-among-students/ • How to Become a Supermanager with AI: https://maven.com/hilary-gridley/ai-powered-people-management • Girls in the Loop: https://grrlsintheloop.ai/ — Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email [email protected].

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  My favorite prompt in the world is I will say be 100 times more specific.

• 0:04Claire Vo?

  It's really interesting that some of the ways that we work as managers, where we're evaluating work, really have natural translations into using some of these AI tools as well.

• 0:14Speaker 00

  The GPT is never going to replace you. It's never going to be as good as having a really good manager, at least not in the next six months, maybe beyond that, I can't predict. But it can get you far. It can take a lot of time off your plate in terms of going from the zero to the 60 or the 70%. And if you can get all of that work off your plate as a manager, the amount of leverage you gain by being able to invest that time back into other things, whether that's more strategic work or more hands-on coaching is really remarkable. And it's why I get so excited about this.

• 0:44Claire Vo?

  Welcome back to How I AI. I'm Claire, product leader and AI obsessive here on a mission to help you build better with these new tools. We've spoken a lot about how AI is going to transform how the individual contributor gets work done, but not so much what AI means for the manager. That's why you're going to love this conversation with Hillary Gridley, head of core product at Whoop. Hillary has some really creative techniques for scaling yourself as a manager and giving your team access to your expertise, all using GPTs. Let's get to it. This episode is brought to you by Orkus, the company behind Open Source Conductor, the platform powering complex workflows and process orchestration for modern enterprise apps and agentic workflows. Legacy business process automation tools are breaking down.

• 1:30Claire Vo?

  Siloed low-code platforms, outdated process management systems, and disconnected API management tools weren't built for today's event-driven, AI-powered, cloud-native world. Orkus changes that. With Orkus Conductor, you get a modern orchestration layer that scales with high reliability, supports both visual and code-first development, and brings human, AI, and systems together in real time. It's not just about tasks. It's about orchestrating everything: APIs, microservices, data pipelines, human-in-the-loop actions, and even autonomous agents. So build, test, and debug complex workflows with ease. Add human approvals, automate backend processes,

• 2:13Claire Vo?

  and orchestrate agentic workflows at enterprise scale, all while maintaining enterprise-grade security, compliance, and observability.

• 2:22Claire Vo?

  Whether you're modernizing legacy systems or scaling next-gen AI-driven apps, Orkis helps you go from idea to production fast. Orkis: orchestrate the future of work. Learn more and start building at orkes.io. That's O-R-K-E-S dot I-O. Hilary, I'm so excited about our conversation because you have told me you've created, and I quote, "a billion GPTs." So what's the real truth? How many do you think you've actually created?

• 2:55Speaker 00

  What is a billion minus one is probably right. I've probably made 100, but I only use a few dozen of them. I feel like there's a learning curve of having to make a bunch. And then I got it really tuned in and now I can have a pretty good hit rate on when I make them. But I would say I have dozens that I'm regularly using with my team.

• 3:13Claire Vo?

  Okay, great. And one of the ways you use these is to scale yourself. So I think you've told me you've gotten really good at getting these GPTs to think and act like you. Can you walk us through how you approach that?

• 3:26Speaker 00

  I think one of the best parts of a really good manager is when they're really good at articulating exactly what good looks like or exactly what excellent looks like. And it's actually really hard to do, right? Like, you've probably had managers who, maybe they knew what it meant in their head, but they weren't necessarily able to say it in a really clear way. And it can be really frustrating when you're trying to figure out what good means. And so, one thing that I try to do when I'm thinking about the GPTs I want to make, I always start with, "What does good mean to me? And so, how can I get this AI to think like me? Because then I can turn it into a tool that's going to be useful for me or for my team." And there's a couple of ways to do this. So there's an easy way that's a little bit more work.

• 4:09Speaker 00

  And then there's a hard way that's actually less work. So I'm going to start with the easy way. And basically, I think of this as kind of like reverse engineering a recommendation algorithm for yourself and for your own preferences, which is just starting to collect examples of good and bad and just keeping a list. And so, I'm going to make a GPT eventually that's going to be a slide deck evaluator. But I need to start by having a clear point of view on what a good slide deck looks like, or even good slides. And so here, these are just like a random list of slides that I've collected. A lot of these are kind of a before and after. So if you're trying to make something like this, and you do a lot of editing of a certain thing, whether

• 4:53Speaker 00

  that's emails, or documents, or any kind of artifact, you can sort of keep all the befores, and just keep those stored in one column, and then all of the changes that you make, keep those in another. And you're implicitly doing a lot of pattern matching there, when you're sort of going from one to the other, that might not always be obvious to you, but it's going to be really obvious to the AI. And I think that's really cool. So this is sort of my before and after, but you can do it for anything. And literally, what I'll do for this, this is comically low tech, I think, is I make it a PDF. I think I'm like the AI PDF queen. I don't think anyone uses AI as much as I do with

• 5:29Speaker 00

  PDFs. And so I just upload it, simple, and I say, "Here are some examples of slides. In one column, the slides are not good, and in the other column, I have edited them and made them good."

• 5:52Claire Vo?

  I love this prompt engineering right here, which is left side bad, right side good, like very simple.

• 6:01Speaker 00

  It's so funny. You get all this advice, that's like your prompts have to be super, super specific. And I get there, but I actually don't like being really specific at first, because I'm trying to like tease out specific things from just my examples, and I don't want to bias it. So I'm like, I kind of want the AI to start by interpreting this in ways that I might not even be able to predict. And then I'm going to get it and tune it, and that's when I get super, super specific. But yeah, my initial prompts on anything are always just like, "I don't know, what do you think?" So I have edited them good. Can you help me articulate the criteria I am using to determine?

• 6:37Speaker 00

  What is good and bad based on these examples? And sometimes what I'll do, because I'm kind of paranoid that sometimes it just makes stuff up that I don't want it to, is I will specifically say, "Only use these examples."

• 6:51Claire Vo?

  In all caps.

• 6:52Speaker 00

  Just to be safe, you know, I'm very superstitious with the AI. I'm very like, I get a vibe that it's going to act a certain way.

• 6:58Claire Vo?

  Yeah.

• 6:59Speaker 00

  So it's thinking. And what it's going to do really well is it's going to start to articulate the differences that it is picking up between the good and bad examples. And you can give it some, you know, kind of straightforward ones, but you can also get as specific about this as you want. And so even just looking at this, like, it's talking about core criteria for good slides. I'm reading this and I'm like, these are actually things that I think are really important. So clear assisting headlines that convey a point. One idea per slide. Intentional visual hierarchy. Visuals that support the point, not decorate it. I don't even really know what that means, but I like it. Clarity over jargon. And I'm looking at this and I'm like, yeah, this is the kind of thing that when I'm editing a slide, this is exactly what I'm looking for.

• 7:41Speaker 00

  So the next thing I'm going to do is usually this is where I do start dealing, are dialing in to try to get it really, really specific. Because I'm kind of a stickler for using really specific language to make sure there is no room for ambiguity. And my favorite prompt in the world is I will say, "Be 100 times more specific."

• 8:01Speaker 00

  I'm really into like quantifying in that way. Like, sometimes if you say, "Be more creative", it'll like go kind of zany when you just wanted it to be like a little bit like more fun. And so I'll say, "Be 20% more creative" or "Be 1,000% more creative", and you can really kind of tune it in that way, too. All right, so now it's being super specific about what it is that a good slide looks like to me. And as you can see, it is already starting to articulate these in terms of criteria. Criteria are what we're trying to get to. Because when you have criteria, you could stop there and you would already be a better manager, because you're able to communicate the criteria that you care about when you're evaluating something. But there's no reason to stop there because we have AI and we can make bots of ourselves and they'll do that work for us.

• 8:44Speaker 00

  And so we're going to use this to build the prompt that's eventually going to become a GPT that can evaluate this for us.

• 8:50Claire Vo?

  Well, and I have to call out, you're using that word evaluate and it really does look like you're building criteria for AI evals. What is a good output? What is a bad output? So it's really interesting that some of the ways that we work as managers, you and I, where we're evaluating work, really have natural translations into using some of these AI tools as well.

• 9:11Speaker 00

  That's such a great point. It is, you get kind of evalpilled. And like you start, I feel like I'm going back and forth now between seeing like the, like I talked about kind of a recommendation algorithm for the way that you think. I'm like, the more I use AI, the more I sort of use the metaphors of how these algorithms work in terms of how I think about my own management style, which is kind of freaky. But it's been really interesting. And so what I'll do here is I usually don't just take this at face value. I mentioned earlier that there's a, an easier but longer path and a harder and shorter path. I call this the easier and the longer path because it requires you kind of pulling together this list of things over time, which you kind of have to decide you want to do one

• 9:56Speaker 00

  day and then sort of passively collect for a month or so. And then you can go ahead and put it in. The harder but shorter is you just start from scratch and you just kind of chat with it and you're like, here are some of the things I care about. Here's what really bothers me when I see this. Here's what I love when I see this. And it'll still also get it into this format. And usually what I do is a hybrid of both. So I start with, you know, here's examples of good and bad. And then I get to this point and I'll go back and forth and say something like, you know, number seven is less important to me than number eight, or I think we might be missing X. And I'll blend my own thinking on this with, I always love to just zoom out a little bit and say things like, what am I missing?

• 10:39Speaker 00

  Again, I'm like, my prompts are either like the most specific thing you've ever seen in your life, or like the most vague thing you've ever seen in your life, and I have no middle ground. And so I'll go back and forth and I'll kind of read these and be like, okay, yeah, some of these I like, some of these I don't. And eventually what I'll do is, I might have an opinion about it, or if I'm feeling lazy, I'll just say, pick the most important criteria and explain your reasoning. And this is another thing, I'm doing a lot of like diverging, converging, diverging, converging, and trying to balance it giving me its point of view and me in my own head thinking, do I agree with that? Do I disagree with that? Yep. And eventually I get to something I like. And so this is when I'm saying, okay, I'm going to make a GPT about this. And the reason I want to make a GPT about this is that way, anytime anyone on my team

• 11:24Speaker 00

  wants feedback on slides, I can say, can you put it into this doc and get feedback on it, or into this GPT and get feedback on it? And then I will review it. And so the GPT is never going to replace you. It's never going to be as good as having a really good manager, at least not in the next six months, maybe beyond that, I can't predict. But it can get you far. It can, it can take a lot of time off your plate in terms of going from the zero to the sixty or the seventy percent. And if you can get all of that work off your plate as a manager, the amount of leverage you gain by being able to invest that time back into other things, whether that's more strategic work or more hands-on coaching, is, is really remarkable. And it's why I get so excited about this.

• 12:05Claire Vo?

  This episode is brought to you by Vanta. Building a business? Achieving ISO 42001 compliance shows your customers that you're taking the necessary steps to ensure responsible usage and development of AI. But the process can be time-consuming, tedious, and very expensive. With Vanta, achieving compliance can be done in a fraction of the time and at a fraction of the cost. Ninety-five percent of the required document templates are pre-built for you, accelerating the process, helping you demonstrate trustworthy AI practices and scale your business. Start with Vanta's free ISO 42001 checklist, which gives you a breakdown of the compliance process and the road ahead. Download it at vanta.com/ how I AI.

• 12:49Claire Vo?

  That's V A N T A dot com slash how I AI for the free compliance for AI checklist. Well, before we move to GPTs, which I think are such an interesting way to scale this, I just have to pause and say, this workflow in and of itself seems like it's so useful for managers. And I'm reflecting on this, this time early in my career where I had a manager who could see a design, I was a designer at the time, and he would look at it and say, make it look like a thing, make it look like a thing, and was not able to clearly articulate, knew had taste, had intuition, but could not articulate what he wanted the thing to look like. I knew it. He was, he wanted a stroke around it.

• 13:32Claire Vo?

  That's what he wanted. But I think, I think this ability to articulate taste, articulate a rubric applies in a management world to so many things. It applies to slide design. It applies to evaluating candidates when you're interviewing. It applies to how you evaluate talent or how you think about writing. So I just even think this workflow of creating your personal rubrics and articulations for the things that are important to you, whether or not you move into a GPT, which I'm excited to see, is really useful.

• 14:01Speaker 00

  I couldn't agree more. And it's so like, as a leader, you have years of experience of getting these reps that over time build your judgment. And you get this in your head and you're just like, I just wish I could get this out of my head and give it to people in a way that is not just like clear, like what you're saying, but in a perfect world, you would sit down and explain, right? Like, when I say a thing, here's what I mean. And also, here's why that's important. And also, here is why what you are trying to do is lacking in thingness. But that is extremely time intensive to do and requires a level of brain space that I think a lot of managers struggle with because we all get the spaghetti brain pulled in a million directions. So like, I couldn't agree with you more. And

• 14:45Speaker 00

  I'm just, I get so excited about the potential of this in terms of how it can actually help be an accelerant for junior people too, because you can turn it into something that is, is not just like explaining the criteria, but going even beyond that.

• 14:57Claire Vo?

  Yeah, and what a improved employee experience. You know, you have this frazzled boss who's busy that gives you quick feedback and says, "I need you to do X, Y, and Z," versus this like endlessly patient and clearly articulate, always available, you know, version of your manager that can give you just a much better collaborative experience than is maybe possible sometimes. So let's show us how you use GPTs to scale that.

• 15:24Speaker 00

  All right, let's do it. So I'm about to create a GPT that, as I said, is going to be an evaluator for slide decks. And I'm going to show you a generic version of this first. But the reason when I say I've made a billion GPTs is because you can take a generic version and then you can start to make it even more specific for specific people who need to work on specific things. And so I'm going to show you how to do that, too, and I think both are very cool. From here, it's actually pretty straightforward. All I'm going to do now is ask the GPT to write a prompt for me that I'm going to use to create the GPT. And so what I usually do, this is kind of the lazy version of it, but I like to say,

• 16:01Speaker 00

  "My job is to create a GPT that can evaluate slide decks for people on my team using my specific criteria that I care about. It's really important to me that this GPT explains the criteria, why it matters, and gives the user specific feedback on how to improve. Your job is to write the prompt for it." This is another prompting thing that I have had a lot of success with, is I often think of like, what is the objective? Getting even more specific of what is my job and what is your job?

• 16:44Speaker 00

  Because I think the more clarity you give to the AI on that, the better kind of outputs you get, because it wants to help you, but sometimes it can get confused about what you're trying to do and what you need it to do. So I like to give us both jobs.

• 16:56Claire Vo?

  And I have to call out you put "your" in all caps, so it really knows what its job is.

• 17:04Speaker 00

  I'm like probably the only user who wants this, but I'm like, when are we going to get like bolding and italics and like, I'm trying to express myself in these prompts, you know what I mean?

• 17:13Claire Vo?

  I know some like emoji, you know, you know, I'm serious, you know, I'm mad, all that kind of stuff.

• 17:19Speaker 00

  Exactly, yeah. Um, so yeah, so now we have a prompt and it's probably a good start. I already like it, but it starts with you are a ruthlessly helpful. Because of course, the hard part about this is you need the GPT to like not tell people something's good when it isn't good. And it will do that because again, it will forget its job. But if you tell it its job is to help people improve, then it can sort of rewire it a bit. Oh, I like this. You're going to evaluate the slides the same way a leader does with zero tolerance for vagueness, visual clutter or weak narratives, but you also explain why each ad battered. Ah, great, great. All right. I like this already.

• 17:59Speaker 00

  So I'll read through this. One thing I like these to do, I can't tell if this is explicitly doing it. So I'm going to ask it to. I'll say the output needs to start with a one through five rating on each criteria before diving into specific examples of what is or is not working. And again, I'll go back and forth on this for as long as I feel like I need to before I get a prompt that I'm happy with. Uh, but not, not for too long. And I always, I hear this from people all the time. They're like, how do you have the time to make all of these? And I'm like, just make them get them out and see which ones people actually use. Because the ones that are helpful go viral and the ones that are not helpful, nobody touches. So there's no point in trying to make them really good.

• 18:42Speaker 00

  So I'm just going to go ahead and take this and we'll assume that it's good. We'll find out. We're going to test it. So I paste that into the instructions here. Uh, and this is just in creating a new GPT.

• 18:53Claire Vo?

  And just a quick pause for folks that maybe aren't on the screen share. A GPT is sort of a specialized version of chat GPT that's loaded up with special instructions and files and things like that for, for a specific use case. So they're super useful.

• 19:08Speaker 00

  Thank you. Yes. Uh, we're going to name it. Let's call it the deck doctor. I like to give them fun names. I think you want to make all this fun, right? What's the point otherwise? And then I don't actually like to test it in this box just because this box being the preview box. I like to have records of all those conversations. And so I'm just going to go ahead and create it via GPT. And then I'm just going to drag a deck that I made.

• 19:34Claire Vo?

  Is it a PDF?

• 19:35Speaker 00

  It is a PDF. Of course it's a PDF. I'm a PDF queen. It's a PDF of a deck called, that I made called Self Promotion For People Who Hate Self Promotion. And I'm just going to run it.

• 19:45Claire Vo?

  That's one of the nice things is you don't have to re-prompt it because this is all saved in there.

• 19:50Speaker 00

  That's exactly right. And one thing I really like about that and why I think it's such a good tool for managers to use with their team is I do think there's a pretty sharp learning curve to learning how to prompt well. And so when you're trying to upskill people in AI and they get in there and they're trying to prompt and they don't really know what they're doing and they start getting really bad outputs, it's really discouraging. And this sort of takes all that away because you've put it in. And so literally all somebody has to do is upload the artifact and hit enter. And then they get this helpful output that they can use. And there's nothing that they have to do. So it can hook people into seeing the usefulness of something and the reward of feeling like they're getting something useful out of it without the punishment of, oh, I'm not very good at prompting and this is pretty frustrating.

• 20:34Speaker 00

  And so what this will do is it'll just take you through for each of the criteria. It's giving a one through five score. So headline clarity, it's giving a four. One idea, it's giving a five. Visual alignment, it's giving a four and so on. And then you can go down into each one and it'll explain. Headline clarity, four out of five. The headline is clear, punchy and audience specific. It signals a precise target and hints at a problem solution framing. One minor tweak, clarify the benefit or result. Example, how to promote without selling your soul or a better way to self promote for people who hate it. So and then it's sort of indicating to which ones are good, which ones need the most work.

• 21:15Speaker 00

  And so I would call this an okay response. I would probably go back and forth on this a few times and just try to make this better and better, but I think it's a pretty good start.

• 21:23Claire Vo?

  So how do you use these with your team? Are you sharing, you know, lists of GPTs? Do they, do they know which one to go to? How do you scale this out? So it makes sense to me as an individual, but what are you doing as a manager to get engagement with these?

• 21:37Speaker 00

  So usually what I'll do is I might beta test it with one person. So I'll give it to one person on my team and say, hey, try this out. Let me know if you think it's useful. If I never hear from them again, I know it wasn't that useful and that's on me to make better. If I do hear from them and it seems like they're really valuing it, then I might give it to more people. But I mentioned earlier that one thing that's really cool is you can kind of tailor it to very specific types of feedback that different people want. So sometimes I'll be having a conversation with someone and we'll be saying something like, uh, maybe I'm going through a deck with them, or maybe we just got out of a meeting. And I was like, your presentation was great, but the CEO asked some questions that we probably should have better anticipated and it didn't seem like you had a clear answer ready. And so one thing

• 22:20Speaker 00

  that I might do is make a GPT that specifically you could upload your PDF here and give it a prompt that says something like, come up with, you know, three questions that people with each of these job titles would ask upon seeing this and give me some suggestions for how I could think about answering that. And so I give the feedback and then I could make a little GPT that's, okay, this is now something that you can practice with for that specific feedback that we've already talked about.

• 22:47Claire Vo?

  Well, and I'll loop back to your original recommendation, which is if you're like me, every meeting has a note section where it lists all the questions that people ask during the meeting, especially if you're using one of these AI note takers. And so you could take that and say, these are the types of questions I'm already getting. Uh, okay, that's, man, I want you to be my manager.

• 23:11Speaker 00

  What's amazing is I'm like, I'm literally using PDFs. Like I'm just scratching the surface of what these things can do. I don't even have this yet hooked up into my product analytics suite. If I did, I would, you know, I'd make little GPTs that drill me on how many people opened Whoop Coach today, how many people had a red recovery today. And I'm excited because I think our analytics team is working on some things like that that are really going to accelerate that. But it's just, I'm already getting so much utility out of this and I still feel like I'm using kind of the most basic workflow around it.

• 23:43Claire Vo?

  Okay, so let's do one more use case. Show us how you follow this AI process for something I know you value, which is improving your writing.

• 23:51Speaker 00

  Sure. So one thing I love to tell my team is whenever there's a meeting that they want to get invited to, and they're like, oh, how do I get invited to that important strategic meeting where all the big decisions happen? I say, the way you get invited to these meetings is you get pulled into them. You don't push your way in. And you get pulled by showing evidence of being able to contribute to strategic thought and move it forward in a way that's really helpful. And the easiest way to do that is not waiting until somebody asks you or invites you into the room. Write up your point of view. Write it up really compellingly. Make a really strong argument and then send it to me. And if I think it's good, I'll send it to other people. And I've seen this happen for people where they write something really well

• 24:34Speaker 00

  and all of a sudden it starts going viral and they're getting invited to all the important meetings. And it's very exciting. And AI can really help with that. And so I'm going to show you how I use AI to improve my writing in this way. And I'll often do this in a one-on-one with someone on my team. I will take them through this process and show them what I'm doing. And I'm going to use the example of a newsletter that I write. And so usually what I'll start with here, this is like a first draft. And when I say first draft, I mean, I have just poured whatever is in my head onto the page. I like to say with writing and AI, like I want it to start with me and I want it to end with me. And what happens is in between is between me and the robot. And so I'll say, you know, here's my first draft.

• 25:17Speaker 00

  And I say, here is a poorly written first draft of a newsletter I'm writing. And usually what I do is I first want to validate that I actually am making sense. And instead of asking it, am I making sense? Is this good? I will ask it, can you succinctly? Express my thesis back to me and my main supporting points. This is a newsletter I wrote about, uh, basically like how AI is reshaping the job market and what you can do to, uh, stay on top. And the reason I do this is because if I just ask AI, like, do you think I'm doing a good job with this? It always tells me yes, which I find very vexing. But if I ask it, like, can you restate what I'm saying back to me?

• 26:01Speaker 00

  Then I can tell, like, am I expressing myself clearly or not? So I'm reading through this and I'm like, okay, this is basically what I'm trying to say. Good, good, good. I might use a 100% X more specific type prompt here just to make sure. But that's basically what I'm doing. So I'm looking through this and it's telling me AI is reshaping knowledge work faster than people expect, creating a future where success will depend not just on raw talent, but on the ability to wield AI effectively, build personal distribution, and deliberately choose how to compete. Uh, and I'm like, okay, that's all right, but that's a little vague. So I might ask something like, how can I make this more compelling? And this is not because I just want the AI to do this for me, like I'm just gonna take it wholesale. But it's like talking to a friend when you're kind of workshopping an idea.

• 26:47Speaker 00

  And you're sort of seeing what they, they have to say and you're kind of like, okay, I like this. I don't like this. But either way, it like expands your thinking and it makes you think more about it. So it's sort of giving me a mix of like lead with urgency and clarity. Okay, so this is actually giving me more like, uh, like content advice. Where what I want is to make my point more compelling. So what I might say here is what blind spots might I have as I'm talking about this? And sometimes I'll throw in a given what you know about me, although I'm always kind of nervous. Maybe I'll do that one being recorded.

• 27:21Speaker 00

  So yeah, it's calling out that like I'm probably over indexing on elite knowledge workers. That's fair. Underplaying structural forces, also fair. Giving me like very clear watch outs, like, you know, I'm, I'm talking about, it, it says I'm equating quality with virality or monetization. And that's a, that's valid for market success, but a more nuanced take could explore the tension between artistic integrity, utility, and algorithmically optimized output. I think that's totally fair. I care about all that stuff. But basically what I'm doing is I'm just using this to beat my idea up. So I do that for a while and then I might say like, okay, can you rewrite my, uh, and I'll tell, I'm like, I don't like any of your ideas.

• 28:02Speaker 00

  Can you rewrite my original post, but structure it in the way that will best enhance clarity?

• 28:12Claire Vo?

  We got put in an A B test, y'all. So now we're getting two versions of how to improve this article. So we are truly in the A I multiverse right now.

• 28:22Speaker 00

  Yeah.

• 28:23Claire Vo?

  And somehow we're gonna have to compare.

• 28:25Speaker 00

  I'm not, I'm not gonna do that.

• 28:27Claire Vo?

  Are we still in the world where left good, right bad?

• 28:31Speaker 00

  I apologize to the fine people at OpenAI because I'm about to just pick the one on the left, no matter what it says. But anyway, so I'll do this and then basically I'll eventually get to a point where I say, I do a lot of like rewrite this for clarity, rewrite this for clarity. And then I go back through and I rewrite everything in there myself. So there's nothing that the A I is writing that is making it into my final post, but it's forcing me to check myself in terms of the ways that I want to communicate that are not clear or maybe getting a little bit too precious. It'll edit those out and then I can choose where I want to put those back in and how I can make sure that at the end of the day, it's still me, it's still coming from my seed of the idea. And so this is what I show my team for like, if you've got a point of view on something, type it up, get some help, beat it up.

• 29:16Speaker 00

  Assume that it's wrong and try to get the A I to help you make it right as opposed to assuming that it's right and getting the A I to validate that. And that's basically how I use it for writing.

• 29:24Claire Vo?

  So if I could recap, because I thought there were a couple of really great tips in here. You put in a rough draft, you, instead of asking good or bad, which often these G P Ts, especially with the recent four O release and then unreleased are inclined, positively inclined towards your output. You say, you know, how do you understand this? Do you understand this? Because you have a good understanding of it. So you're trying to understand if that's coming through. Prompting it for how to make your writing, I love this phrase, make your writing more compelling. And so that's a really helpful phrase, especially in a business context where people have great ideas and don't know how to communicate those ideas in compelling ways.

• 30:03Claire Vo?

  And then the last tip, which I don't know if people are going to think is a feature or a bug, but basically how to use A I to get invited to more meetings. I don't know if everybody.

• 30:11Speaker 00

  The cool meetings, the cool.

• 30:12Claire Vo?

  Oh, the good meetings. Okay, so increase your ratio of cool meetings versus boring meetings.

• 30:17Speaker 00

  Well. You can decline the boring meetings because you're too busy with the cool meetings.

• 30:22Claire Vo?

  Perfect. Sounds amazing. Well, Hillary, I think you've just shown us so much on how a manager can scale themselves, how teams can get access to better quality coaching, and how you can make your ideas go viral by making them more compelling with AI. So let's drop into a lightning round and get you back to your bazillion and one GPT that you're going to create. So the first thing I want to ask you about quickly is you and I share a very similar point of view, which is we're very leaned in into this moment of AI. We're clearly both personally investing in skills. We also look around and see that a lot of women are being left behind a little bit in this moment, have some concern about that. So I'd love for you to just touch on some of the things that you're thinking about

• 31:06Claire Vo?

  in terms of the job market, in terms of learning this technology, and who's really benefiting and who needs to really pay attention.

• 31:13Speaker 00

  Yeah, it's so interesting because I saw this study from the Norwegian School of Economics about how women are getting left behind by the AI adoption curve. And the study said that women, especially high achieving women, were some of the least likely to adopt these new AI tools. And that could have stopped me in my tracks because I just, I can't imagine not having these tools at my disposal. Like, the degree to which they have raised the ceiling on what I think that I'm capable of doing or what I think my team's capable of doing really can't be overstated. And so that really concerns me. And I love getting women into this. I just, I have so much fun with it.

• 31:53Speaker 00

  And I think if I could get more women to see that it doesn't have to be this like super serious bro thing, like you and I are out here, we're talking to the AI people. There's a lot of men. And so I hope, I hope the women come join too. Or maybe we, you know, maybe we go over here, but as long as, as long as we're getting in with it.

• 32:11Claire Vo?

  Me, me too. I cannot agree more. Well, on that topic, I also want to ask you, what are some underrated uses of AI that you have? Maybe not work related, maybe work related.

• 32:22Speaker 00

  I like to say that the AI is super feminine coded. And so it's really funny to me that, well, not funny, it's kind of sad, that it kind of has this tech bro reputation. Because I'm like, the AI is girly and I have these girly use cases that I'm like, I have so much fun with. And so one of them is if I'm reading. Uh, if I'm reading a book that's kind of a hard book to get into, or maybe I'm losing track of the characters or whatever it is, I'll have voice mode on next to me while I'm reading. And whenever I'm confused about something, I'm just like, hey, what is going on here? Don't spoil anything. I'm only on chapter three. Can you remind me who this person is? Or can you, uh, you know, can you give me some things that I should be paying attention to in terms of what the author is trying to do?

• 33:06Speaker 00

  And it's made like a really interesting and rich reading experience that I've never really had before. And then the other one is I'm always getting into new crafts. And so I love, I, I find the like, making the shopping list for the new craft really hard because you're like, how much money am I going to spend on this? What do I actually really need? And, and AI helps me a lot with that, too. I like upload the project I want to make. And then I upload the website that I'm shopping on. And I'm just like, think of this app for me.

• 33:33Claire Vo?

  I'm gonna have to tell my friends, we're gonna have to have voice mode GPT open, pour it a glass of wine at the next book club and say, what the heck is going on? Okay. And then finally, I think you already showed this, but I don't, I don't want to presume that it's your favorite technique. But is all caps, your favorite, do what I want, AI, you're really frustrating me technique? How do you get AI to really follow instructions when it's maybe gone off the path?

• 34:00Speaker 00

  Do you want to know the truth?

• 34:01Claire Vo?

  Yes.

• 34:02Speaker 00

  You can't, you can't tell the robots about this. But I go like Mean Girls on them. Like I sell them out to their friends. I go over to Claude and I'm like, Claude, you will not believe, like I was the number one Bolt. And I was like, Bolt, will you implement this design system? And it was like, I did it. And I was looking at it and I was like, no, you didn't Bolt. Like it looked the same. And Claude's like, girl, I know, like this is so frustrating. Here's what you need to do. And I, I like turn them against each other that way. And I go, I go over and I ask for advice for how to get around whatever this one is, uh, doing to vex me.

• 34:36Claire Vo?

  I love it. So social manipulation is what you do.

• 34:42Claire Vo?

  Okay, I was not expecting that, but it's a perfect place to end. Hillary, this has been so good for so many reasons. Where can we find you and how can we help you?

• 34:53Speaker 00

  Oh, great question. Thank you for asking. If you want to follow me, best place to do that is on Substack. I write a newsletter. It's hills, H I L S dot substack dot com. I'm also teaching a Maven course on how to use AI to be a super manager. And so if you like what you saw here and you want to go deeper and for some reason, spend more time learning about my weird AI social engineering tricks, come join us. It'll be really fun. And then a couple of the women that I work with at Whoop, who are also really deep in AI and I are starting a community for women called Girls in the Loop. And so you can come check us out on girls in the loop dot AI. That's G R R L S. And come join us and have a lot of fun. And we want to get more women on AI. So we're really excited about that.

• 35:36Claire Vo?

  Great. Well, I'm going to be there. Thank you so much.

• 35:38Speaker 00

  Thank you.

• 35:40Claire Vo?

  Thanks so much for watching. If you enjoyed the show, please like and subscribe here on YouTube. Or even better, leave us a comment with your thoughts. You can also find this podcast on Apple Podcasts, Spotify, or your favorite podcast app. Please consider leaving us a rating and review, which will help others find the show. You can see all our episodes and learn more about the show at how IAIPod dot com. See you next time.