Transcript: Galit Galperin | AI MINDSET | What Does It Really Take to Succeed with AI?

Eyal David: [00:04] Hi, it’s Eyal, you’re on Product Builder. We’re about to start the episode, but before that, go ahead and hit follow on Spotify or subscribe on YouTube and help us reach more ears. Let’s get started. Galit, how’s it going? Great. How are you? Great, we did it, yes, okay.

Meeting Galit and Hi Mindset

Galit Galperin: [00:24] So I’m Galit Gelperin, and I’m the founder of Hi Mindset, which is basically a strategic consulting company for aligning artificial intelligence in organizations. What is it generally for in terms of a business goal? It can cover areas of how I put it into my product and what’s right to put into my product. But it can also be very, very biased, I’ll say, toward the business of what I want to achieve with AI. Sometimes it’s also things that are internal. Not necessarily cross-functional; sometimes it’s marketing teams, sometimes it will be. I’m talking about the people element or even. A new floor in the building called Hi, so what we need at the strategic level is what I come in for. And I basically sit with management teams and carry out all these plans, implementation, and so on. Of course, there’s also academia on the side, which I love. Faculty member at Reichman, and also at Reichman TechSchool, but really, to sum it up, the background is product, of course.

Eyal David: [01:28] What organization do you manage overall, and it also accompanies fashion, am I understanding correctly? And then there’s the academia, so I just— Brings us back to the pre-conversation we started right now, so wait, let’s open with that: you actually started in Orange, your career. Walk us through the journey over time, just a few points, just so we know how to talk about the transition, wow.

How Galit came to product back in the mobile revolution

Galit Galperin: [01:50] That’s it, honestly, people always ask me how you become a product manager. It’s always in some academic course: how do you become one? And I always say that it found me. It came to me when it wasn’t even—well, it wasn’t a profession yet. Or at least not under its official name; it’s a pretty young profession, around 20–30 years old, and basically phones came along and you had to figure out what to do with them. And my first role was SMS product manager, which usually makes people raise their eyebrows completely; WhatsApp, I tell them, no, SMS, I meant SMS, yes. So it’s like you were in the revolution, yes, the mobile revolution. I was there like, this revolution—you came prepared. That also connects to the pre-conversation, right? I came prepared because I was always very, very into gadgets and things; I thought it was a wonderful thing. That you can just walk around with your phone, it’s truly wonderful, right? Today your phone is something that maybe, you know, is your whole life, but back then it was just something you talked through. And my job was to figure out what else you could do with it.

Eyal David: [03:03] Sounds like, sounds a bit repetitive, right? So I’ll take you back for a second to a very interesting post you had that I saw as a basis for our pre-conversation around the index.

An AI-readiness index: who to hire and who to train

Galit Galperin: [03:16] Maybe expand on it. You’re probably talking about an AI hiring index or really a catalog there of roles along some kind of continuum. Yes, so let me explain the background for a moment, because basically. This index is basically meant for organizations and managers who are looking at business reality today and need to decide whom they hire or whom they train or what this team will look like tomorrow morning. There are changes in work, changes in teams, changes—I’ll say—in our abilities as people, maybe in what we need to know. And as a manager, a lot of the time I’m standing there and like, okay, what does this mean, because in the end I know my people and I know their abilities, but I have no idea what it means. AI, or what they need to do with AI, and who’s good at it, by the way, there isn’t necessarily a correlation. It is important to say, yes, not every high performer means they’ll know how to work with AI or be a fit for it, and therefore when I hire, and many times. There’s really a need to beef up the team now, or even if I’m setting up some kind of AI, an AI squad, then who am I looking for? So this index was actually born out of some need I saw among my clients. By the way, my full index, I told you, hasn’t come out yet. It— Really, over the last three years I’ve been doing data, live data of actual market breadth, but. But there was also an index by Zapier’s CEO, and what happened is that it provided the right marker, and basically the idea is to understand how ready the person in front of you is for what’s called the AI era, in terms of depth, usage, agility, the ability.

Eyal David: [05:04] To absorb something new. Now let’s think for a second about why this is interesting. Yes, I think it’s interesting in two directions. First of all, if I’m an organization, if I’m a manager. Obviously, because I want to build, and from that perspective too it can be interesting if I’m building an AI team. On the other hand, it can also be. The world is changing, the team is changing, nothing is certain anymore. Maybe I’m mixing up the discussion words, not product. Maybe in general I want to beef up certain kinds of product people who have specialized now in development and retention, or in completely different domains and users. Capabilities, all kinds of aspects like that, as I assemble my puzzle, the more interesting aspect for our listeners, I think.

Galit Galperin: [05:38] It’s really where I am on this continuum and what exactly I can do. On the other hand, it also gives people the ability to understand where they are on the scale. People really, really love comparing themselves. Sometimes, for sure, when you’re at work and you know. And then you’re next to someone who really knows, and I say I’m for being there; I always say I don’t know enough, even with all the hours I spend with the tools. And then basically you get a bit of a drive to learn, but when you see something like that that—I’ll say—it categorizes, even though of course there are always exceptions, and that’s also very, you know. He is a right-wing figure; it’s not some doctrine you have to stick to, but it gives me the feeling and also helps me understand, even if I’m looking for a job. Or even if I want to level up; by the way, there are quite a few people it made a total career change for. Okay, like really something one thing before and then something else, because suddenly they had a crazy epiphany and they discovered themselves as marketing people who turned into crazy video people because they suddenly understood that they prefer to do their storytelling in a different way. So also to understand for myself: am I good enough, am I worth it, do I still have more mileage to learn, what do they even expect from me in something like this? I mean, today I’m entering a new place — what’s the expectation of what I’m supposed to be?

What to do tomorrow morning to start adopting AI

Eyal David: [07:06] Yes, yes, and I also ran some kind of survey right before we went up, the day before, okay, also kind of to see, of course it won’t surprise you, so the thing that stood out most was, okay, so what should we do tomorrow morning? Let’s talk. Also basically about adoption, you can talk about it at the organizational level, and also at the level of the individual person. The individual employee and what they do tomorrow morning. רגע before we continue, click the link to the Product Builders WhatsApp group in the episode description and join the community there, where you can continue the discussion, ask questions, and be the first to receive recipes for new Tachles episodes. We’re waiting for you there, and now back to the episode. So you’re saying, let’s start from the fact that you work with a lot of tools. Yes, what does it look like for you?

Galit Galperin: [08:00] So first of all, I’ll say that my job, by virtue of the fact that I really have many types of tools and organizations and situations and solutions. Maybe I’m not representative of what’s happening for someone who at the end of the day simply works in an organization, on a team, with, I’ll say, a company. X tools that they can work with in practice, okay? It’s not something you can do at home, but in your work it’s very critical what you’re given. And just this morning they asked me that question: which tool? It’s like there’s this feeling where you say, but how many tools can there even be? And by the way, looking ahead, I say: how many assistants will be put in front of you, there’ll be one. פעיל others that, by the way, Samar already said this a decade ago. Fine, I’m just saying I found my slide where I lecture with this thing; I’m completely on board with it, it really is arriving and happening. And in this question of what to do tomorrow, I say first of all to look inward and understand for a second. Really, what is my situation with this at the personal, intimate level?

Eyal David: [09:08] Wait, let’s start this for a second.

Galit Galperin: [09:12] Look, over the years, conceptually, I always check the perceptions of the person I’m coming to sit with, because we have a lot of beliefs and perceptions that sometimes are limiting, even if we’re very much going back to the index. They want to believe that we’re progressing; there are a lot of people who are afraid of this. There are a lot of people who don’t believe in it; there are a lot of people for whom this story — that every time it’s a different challenge — feels like they’re pulling the rug out from under me, and I don’t have to do this, this isn’t the product manager sitting next to me whom I trust. So first of all, trust and safety at the most basic level, sometimes that also means... Wait, this will replace me; wait, the key person now will do my local product stuff; wait, whose job is this? They don’t like that; there are those who really don’t like it when the company decides for them that now they’ll do their work in a certain way, because if I know how to do discoveries really well, why do I need, let’s do it, I can succeed, let’s do some kind of framing, let’s make it a conversation.

Eyal David: [10:10] Let’s put ourselves in a box for a second, okay, one. Let’s say that someone who doesn’t have a tool they’re using right now — let’s even stretch it to GPT, Gemini, whatever — that’s the baseline if we look at this index; if you’re not using it yet, then... Find another podcast to listen to. Right now, the second thing I think is that whoever thinks this revolution won’t come to replace them... And they work in high-tech right now, so I’m also taking them out for a second. Maybe on your podcast you’ll talk about it and pick it up from there, I don’t know, but for this podcast, let’s stretch that line and go north from there. And now we understand that companies are becoming more efficient, we understand that these tools are there. And now, like, let’s actually think about it: of course it’s scary, of course nobody likes change, but it’s happening, all the roles are changing, okay?

Why you must make time to experiment

Galit Galperin: [11:04] The first thing is to make time, and I know there isn’t any because nobody has any, but what you need to understand here is that the technology is taking it; it’s kind of pulling ahead of us. In many ways it’s exponential; if you didn’t get on the train now, by the time the next train arrives you’ll already be late, and I’m not saying that because of FOMO, why? Because the more you... And the story here is to experiment; the more you experiment, the more you’ll fall, and then you’ll learn more. Okay, because if you think that just talking about it normally and asking it and saying it — write me the user story, that’s the story — then you’re not in the event; you’re not in the event. In my sessions I constantly show דווקא the place... To think with it דווקא the place of the high, and דווקא there to take it, even though it’s hard, because it’s generative, because you need to choke it from five hundred places so that maybe it’ll be an ail I can trust when I’m talking about it. But דווקא there you find the gems that maybe as a person you didn’t know, didn’t think of, and you’re limited by them, okay, all my blind spots. Beyond that, it allows me to do things I couldn’t do before — and really, really. One of the things I love most is the scale of something, okay, of ideas, of responses, of articles, the whole story of scale. You can’t argue that it saves you time; you can argue about how accurate it will be, and there are mitigations for that too, but. That’s the power of managers and female managers; it didn’t exist before. By the way, not just them, yes, but they’re just incredibly with a lot of my irritation, I’ll say in the product lifecycle of product people there is that, but even if I say entrepreneurs, okay, in the end you have a lot of a chain of actions that is not one action. I’m not writing the same code now, and that’s what I do. I have a lot of things that I consciously, they’re different even in the brain, okay: doing user ideas, doing analysis, doing strategy, doing prioritization—it’s not the same area. And I think that if you know how to dedicate yourself first of all at the level of commitment of the mask, it’s a bit like a gym, only for the brain. Put in the hour of the day when you know you’re at peak clarity, not leaving it for before bed, and really do experiments, small experiments. Saying, “I’ll try to do this,” and then finding out it doesn’t work—why doesn’t it work? Okay, that’s fine too. I’m not saying there are only successes, but try it—try it now in three tools, try it in a tool that, say, is just a Jewish tool. That’s suited to the task—understand that at all. And by the way, that’s the difficulty, I think, for a lot of people I meet: understanding what is good for what. What is good for what? Why is Claude better than GPT? Now again, I’m getting into the weeds, but it can equally well be not getting into the weeds, okay? If today I want—today I was asked for a podcast to be read aloud. But she tried it in NotebookLM and had to try with 11, Levsky wanted it word for word—this is what it’s called, it’s not the tool, come learn. Your work, what you’re trying to create, the storytelling doesn’t matter—the output does. Which tool is suitable, and if it’s not a tool, which chain of actions is suitable for it—that’s also something that people somehow avoid, which I really, like, say there’s no point, because in the end that’s the story, it’s stacking.

Eyal David: [14:37] Okay, so I’m curious to really see what you see people doing. I know my own struggle. What’s your struggle? My struggle is first of all really defining now the time around some goal. I actually don’t go with an hour a day because, a bit like at the gym, I’m looking for a minute. Yes, to reach a real goal.

Galit Galperin: [14:55] And you still don’t reach a real goal within the hour?

Eyal David: [14:58] If I devote an hour, I won’t get there. Almost?

Galit Galperin: [15:00] You need more?

Eyal David: [15:01] Yes. Today it’s less so because today I’m already working in it; basically every time I add another layer, so I already really know the basics, let’s say, because I got through the first month with training in the gym. And I work according to a program, let’s say.

Galit Galperin: [15:16] Look, you’re at the top, okay? It’s not that I’m saying if a person is capable of making a plan that allows them now a day to experiment.

Eyal David: [15:26] It’s coached, but in most of the reality of the people I know and meet, it פשוט isn’t, so let’s go, let’s see the index because we promised—let’s break it down here באמת. We said those who don’t use the tools, okay, we took you off the index, what does it look like.

Basic use vs. really understanding how an LLM works

Galit Galperin: [15:42] So first of all there’s some issue of there’s basic use, okay, that I think today, really, like, ninety-five percent are there, and they’re really in day-to-day use and in questions, but I always argue there’s a layer that’s above, first of all, the layer where you understand—you understand how it works. In almost every room I enter, I’ll ask how it works. 80 percent. Maybe more. I’m talking about tech companies—they didn’t know how to explain to me. Now why is that important? Because once we say don’t know, we miss a lot of risks. Inaccuracies and things that could derail us in our work—what you can explain so we can stretch it, I can pause and stretch the line a bit to talk about what high slope is, because it’s something that kind of the word hasn’t reached us yet, or it’s not very common, but in the United States it’s already a thing, and basically let’s say you and I work on a team and I now need to send you some kind of spec. And since I don’t really know the tool’s limitations, or I don’t fully notice that it’s hallucinating. And I gave it a task and it did the task and I’m happy and satisfied—I made my B and sent it to you. Now you have to deal with what happens with high slope, which means not only do you have to do your job, now you also have to do my job. Because the machine didn’t do it well. Now we see this in a lot of things—I was just talking about product, but you can also take it into the creative worlds. Also in brand worlds, in almost every field there is high slope. It usually happens when not enough context is given to the machine about who we are, what we want to present, and so on. He says that because people really love it when it does it for them. Which, by the way, is exactly me, like exactly this event, but it’s really tempting because it writes it for me in a second, so what if it’s not the word I meant to write, but it wrote it, so okay, like. That’s the first thing in my eyes, because then it really takes it off you; it takes it off you twice. First, because wait, I have to do your work too, and teams don’t like that either. The second time, it reflects on you that, so to speak, you don’t really know how to use it, even at a very basic level, and you need to bluff your way through; the part after that can even point out another thing. You know that, like, unrelated, you asked about criticism, right? Say for a second: this isn’t fair play; this isn’t the right metric to put here.

What AI slop is and why it's dangerous for teams

Eyal David: [18:19] When you asked me what I do with it and where I struggle, I’m now in the studio all day going over the same feedback.

Critical thinking as a skill for working with AI

Galit Galperin: [18:26] That’s mostly what I do. By the way, that’s that’s a very interesting thing you said, because I think it’s one of the things necessarily tied to the index; that is, this index overall models the skills. That I have, or or or the level of skills I have that are required for AI, and one of the skills required for AI is necessarily critical thinking, not in the sense that now I’ll go and look for where it lied to me, because we’re not getting out of the event, although there are mitigations to do מראש but. Those. That’s what I told you in the pre-conversation. I’ll tell you what I think; you tell me what you think, and I’ll okay or not. I’ll say it and guide you, and you’ll write, and it’s fine if you also tell me maybe the mistake, because you’ll bring me a source, but I’ll lead the event. And if I’m not leading the event, say I’m doing something in a field I don’t understand and I want to learn. Then all the more so I’ll try to verify it really, like, I’ve just met Yial and I’m not sure if he’s great or not, and I’ll interrogate him until I feel confident, so. It’s the same thing, only in a kind of machine-multiplied way, and it’s a skill that can really reflect on you that either you don’t care at work or you don’t understand. No no no, if you don’t understand, apply critical thinking, like, enough so that either you’ll care or you’ll know.

Eyal David: [19:56] Just, I’m opening a parenthesis for a second, we’ll come back to the smell among kids—this generation growing up now. Like, it’s a thing to strengthen: the ability to make mistakes completely until we get to, like, that window of faith.

Galit Galperin: [20:10] I’ll tell you why this is something that both bothers me and intrigues me. Because children—everything human, even dogs, okay—when you get feedback from me about something, you avoid it. Meaning this whole story of the no-lits and in the end doesn’t claim properly; everything is all about everyone succeeding and everyone being beautiful and everyone being good, and of course if she succeeded in making the TikTok recipe, how come I don’t, how do I burn this thing. Okay, it’s just that all the time they show me a TikTok recipe that I’m supposed to meet at that level, and factually that doesn’t happen. So, like, also give kids this feeling that it’s a place where you can be wrong and it’s safe for you to be critical, and you need to ask, you need to think. I think we’ve landed here on a really big Duelingo failure, let’s keep going for a second. Too bad you said that; it’s sitting with me in my colon exactly like that, and it’s super frustrating.

Eyal David: [21:10] I think that’s it; this is actually a muscle we can work on. We talked about several places where one can improve, so this is a place that really. Worth deepening into. It’s a skill that is really being critical and also basically putting your ass on the line for it; it’s hard for people, really hard for people.

Galit Galperin: [21:30] They don’t have—I’ll say this is a crude generalization, yes, I’m also making things up—but I think that when there’s something that performs for me. I’m not interested in putting in effort. I’ll give you an example, דווקא from the field of images. I was a documentary photographer in the past, among other things. We didn’t start, okay, fine, and when you shoot documentary. My job was to capture as naturally as possible; basically, not to lean into the wall and not be noticed, to the point that I’d catch the moment, but even within capturing a moment. I need to aim the camera and understand for a second what the aperture is and how much light is coming in and whether I’m at the right angle and all the rules that apply to an image. I think people don’t think about this thing inside their prompt. Okay, they say, “A person sitting with headphones.” A bit like the first prompts were for, what’s it called, for images—I’m blanking, I’m without coffee, yes, the first prompts were really you would just inject this event, but still you can inject the event; now it has a few different things, but what I’m trying to say is that even today people won’t tell you “at aperture 2.” These are technical details that anyone making video or image takes into account in the hardware they’re working with, okay. And the one who says “an image,” you’ll see the simplicity in it, but the more you deepen, that is, if you will craft it like a photographer. You’ll get the photograph’s image because you defined it. By the way, you asked how products are defined and how requirements are defined—it’s the same thing. I always say a prompt is like a פרדי, it’s like.

Why a good prompt is like a photographer's work

Eyal David: [23:23] We’re going into a world everyone talks about of builders, where the developer will do product, product will do UX, everyone—so what you’re בעצם saying is I don’t know where this is going, yes, what you’re בעצם saying is that it’s not that simple.

What it really takes to be a builder in the AI era

Galit Galperin: [23:38] I’m saying that this system needs to be understood; whoever understands it manages to be a builder. There are nuances, because you need to understand that there are nuances, and this morning I literally said that all the personal competitions, this twit one prompt on love, it’s fine, but it’s exactly like a casino. You’re not really, you don’t know what’s going on with you, it’s not you, you’re not really in control, you’re just pulling something now and it’s a slot machine that will now give you an output, but it’s not really an output that will work for you again, or in another tool, and so on and so on. When you’re a builder, a builder, then you need to know exactly. Which tool, which prompt, which chaining, and what the outcome is. And you need to craft it; that won’t help. I’m now doing another one of builders for over two years now, and I see that you need to understand the need, you need to, and that, by the way, is exactly a skill-wise thing. And that’s also why this talk about generalist and specialist, and in the end you need to, like, understand where you want to get to. Because when you just want a person with an earpiece it’s cool, but if now you want it to be in your brand, for it to be like your podcast, maybe even for it to resemble you. You already need to work a bit harder, and it’s no longer just the tool doing what I want; I’m already a bit hungry with it, you say, I’m giving feedback. It’s not that magic of “I built 15 agents overnight.”

Eyal David: [25:06] Two things come up for me; we’ll actually come back again to the index, I’m going to pull us there so I can also move forward from it, and I think the example of the photographer is the clearest. Maybe you want to add to it, so two things came up for me that I was thinking about, one. An audience of product people who usually came from a previous role and have a perspective, or maybe a broad one, maybe a business role, maybe a technical role too. So דווקא this closeness of this profession to a builder sounds to me much easier and more logical. That’s one, because if I’ve just been a developer my whole life, or I’ve been in development my whole life, I don’t have a lot of skills that can help me with prompt tuning. The second thing I’m thinking about. How much more grace does this thing have, like until they’ll be good enough to basically skip even that and give me the middle, or until the 80, 85 percent, with those at a very, very good level that won’t require this fine-tuning.

Galit Galperin: [26:00] That’s a great question, because even today you already have within certain products what I call mitigations, which are called prompt enhancement; that’s an invention of product people who said. I don’t want to do this; I want to ask, and then what actually happens in the back is that we translate the intent into prompt enhancement, and what’s called do the work. The problem with that, by the way. Is that most of the time, because it’s generative, it adds all kinds of embellishments that I didn’t necessarily intend. I would be so happy if the moment they fine-tuned the prompt it would give you the whole thing, and then again we get into this event of disappointment, so. So I think it doesn’t matter how much we move closer, there will always be products that try to make it accessible, okay, or to wrap up how much, technologically speaking, we move closer. I think it still has a bit of time. Like, it’s not. Yesterday there was the CEO of Anthropic who put out a very much, like, letter saying we have two years if we don’t get our act together, so to speak, it’s on us, and he specifically, whom I very much appreciate and like to read. It’s very possible that it will be good enough, but I am not at all convinced. That they’ll succeed in getting it, again, at the natural level of it, as a language model, to stop making hallucinations.

Eyal David: [27:30] So like you’re saying the dependence will remain, basically, yes.

Galit Galperin: [27:34] Yes, now again, unless everything is wrapped in a very accessible way, but that’s kind of a band-aid, you understand, because if you arrive at a product that isn’t wrapped, then you need to, like, know now. Learning, and I’m into it, I’m swimming in it. Product managers are people who, in my opinion, are first of all builders. By nature, people who are interested in everything, people who are curious how things work, people who take integration from one world and put it into another world, people who understand people’s psychology and what drives people, people who like building solutions. That’s all, in my opinion, by definition, okay, a builder. Now there may be some who aren’t builders, but product managers, in my opinion, are. Are the first to be that, what I said, like they’re first in line to be builders. It’s not that I think developers can’t be builders. Because a developer too, in the end, a developer who now works in the method of spec-driven development, okay. That he understood he can’t just write everything according to what he feels like, but that there needs to be a method and pre-mortem and background, and so on, he can also succeed; it’s not that he won’t understand the need. He’ll understand, okay, his strength in the skill, his strength will be fixing the code. I’ll say the shitty thing that came out of AI, but also that, for example.

Eyal David: [28:57] It’s a matter of technological progress, you understand, I even think developers have a better chance, really, because in the end in companies they sit on the valve. If you call them mechanics, then דווקא they can have more influence; they’ll acquire the product skills, let’s call it that.

Galit Galperin: [29:15] Not understanding the problem—you kind of say that, but it’s not certain they want it; like there are code people I know, I see, who love code, not to say hate the AI that writes code for them because it made them become the direction. Okay, within five years, okay, but a person learned their whole life to write code because they love it, just like you learned product management and you love it.

Eyal David: [29:36] I also learned it already, I understand where it’s going, and in the end that’s the force that drives us, I think, the world, and yes.

Galit Galperin: [29:45] So let’s look now, do you think—I’m not sure that a developer and that is really, like, a question. The whole thing is terrible, horribly crude in a human way, but in the end it’s a skill. Still, I’m not sure every developer wants or is able to be a product person, and I’ll tell you one more thing: let’s bring in the third side we haven’t talked about. Which is UX slash design. How do I know? A lot of times, what’s called “there’s no such thing in the event” because you see, you see that no matter how amazing the functionality can be, if it’s not coded and the idea and everything, but it hurts your eye and you can’t use it, you can’t use it — it’s unusable. It’s unusable, or it looks vanilla, kind of like all the other high-end products, and you say, I’m not putting my Google into this thing. You see it. It’s not that now you’re asking me whether in another year or two it won’t happen anymore — probably it will, but I don’t know.

Eyal David: [30:43] It’ll be good enough, like, it’ll be good enough.

Galit Galperin: [30:46] I, in general, would want. The power of “there” to give me a recommendation. Meaning, it will know me and it will give me the recommendation, and the skill of who I am, less good in UX, bye-bye, I’ll recommend to you how it works, here. Accessibility, wow, design, wow, UX, wow — what do you want, choose. But it’s also, you know, evolution, and I think it’s going to happen, so we’re really going to be in a world where I’ll say the competition will be over the skill set, which is… By the way, the index fits best right next to it, which is the generalists.

Eyal David: [31:21] Okay, okay, so we’ll get back to that. So really we said there are companies that adopt tools, and there are companies that adopt them at the most basic level. There are the more advanced companies, actually, where this gets stretched into: I already work on a regular basis, most of my work is actually in choosing, not so much most of the work, I’ll work with tools, or I’ll say it this way: I…

Building internal tools, no-code and mini tools

Galit Galperin: [31:43] I’ll build tools for myself. I call them little tools, okay? Or no-code tools. And actually, I know there’s also a matter of distinguishing the essential from the secondary in my work, and that— That’s a thing, a thing that’s taking off on its own. I’m starting to hear these chants everywhere. Because sometimes people get so deep into this whole thing of “I use an AI for everything now” that they use AI even for things that don’t really work or fit, or that I need or want. Leave yourself something you actually like, too. I will never give up talking to a client for feedback, okay? I don’t want to do synthetic clients. It can model it for me, but it’s not a real person. But I do, I mean, I do need to distinguish, and again, everyone has something different they don’t like, you understand. I don’t like “non-function” in a requirements document, okay? That’s it. The machine will do it for me every time, and I can’t, I can’t — my brain shuts down, I can’t think about it. You hate user stories. Everyone should take this wherever they want, make the tools, and know how to bring the best result. In the end, I want it to be a tool that helps me.

Eyal David: [32:52] I don’t want it to replace me. Okay, so let’s also put some… what’s happening there, basically. The next level is people who build things, yes, in no-code, and they also know how to identify those places where they want to insert bottlenecks — that’s usually what they call them.

Galit Galperin: [33:08] Or bottlenecks, or really things where you know exactly, you say, why am I still doing this, like, what—

Eyal David: [33:18] By the way, that’s one of the things I like most: I open Cursor, all kinds of desktop applications that do great things. It’s amazing. How do you manage them — that’s the question. What do you mean?

Galit Galperin: [33:27] You said desktop applications. Okay, now there’s a lot because of the ease — I called it the unbearable ease of creation. There’s some talk about, like, disposable software. Like, I didn’t invest in it, it doesn’t cost me anything, I’m not in an SaaS model, and I wouldn’t take a trip with an app specifically built around my trip. Now, who does that? You make an app for a trip, you do it for all trips, not specifically for a podcast trip I made all kinds of things. It really overwhelms me. I need to say that I have a lot, and how do I manage it? That’s the question: how do I manage it, because I discovered — by the way, Claude, a friend at work— Cursor — that I found, it’s a bit embarrassing to say how many, but let’s just say I found a double-digit number of tools that help me with the podcast, to do what I need: the preparation and the research and all that. Some are even duplicates. Because of this whole disposable thing, and then you ask, how do you manage all this? It’s like, it’s not maintainable, because you can create so easily, so here you made your mother some page and here you made… It balloons. It becomes something that itself starts turning into trash a bit.

Eyal David: [34:45] No wonder I’m also talking about this freed up stuff that’s in your Excel of ideas — I think that’s really the point, yes, it really is a point. Also maintenance-wise, and really if I want someone from my team to work with it, then I suddenly need to take it out, whatever, and make it— And also, like, there are a lot of things I build that in the end…

Galit Galperin: [35:08] Solved something for me, I was super excited, and then afterwards I tried it. The question is whether you need to solve it for a one-time thing, whether it’s worth the hassle, and on the other hand whether it’s worth the learning. Meaning, there’s some kind of tension here between what I build and what feels cool to me, and what I understand that if you now have a team member who’s now working on the podcast — and this happened to me— I found seven tools that were the same thing, not exactly the same, with improvements, but suddenly I said, wait, wait. I need to manage this for a second, I need to keep it, I need to mean, suddenly you have so many, many things and it becomes more work, and that’s also something people say a lot — that sometimes it created more work for them. Like, where did this come from now? I have more work, come on, let’s make it shorter, and that discourages people. Because in the end, the road to success is paved with a lot of work. A lot of your work is with yourself, as you say, sitting with it for an hour, even if now you sit for nine hours and in the end came away with four tools. Tomorrow morning you’ll remember the tools; you’ll work with the tools, they’re meaningful.

Eyal David: [36:12] So good, I asked how do I manage, so I actually have in Notion a project called Toolbox. And basically there I manage the tools, there I manage—yes, but also I kind of, I understand very much what you’re talking about. Okay, but wait, I do want to stick to our format, so okay, so probably I’m placed in this index. After that there is an index of what, of a company that we’re also building Cursor, that’s it, from what’s in the diagram.

Cursor, code, responsibility and what happens before production

Galit Galperin: [36:45] Look.

Eyal David: [36:47] Do you have a question? Sorry, I’m writing at the end, even though I know code, I work with Cursor a bit like I would work now with no-code. I don’t really look at the code, maybe I’m wrong, maybe I’ll write a few tests, but I’m not looking at the code—how much difference is there באמת between these levels.

Galit Galperin: [37:07] Okay, now you stole my previous question—look, Cursor. Cursor, okay, it doesn’t matter, I’d describe it as something that by definition is UI-less and stresses people because it’s not pleasant at the experience level, lots of windows, lots of code, everything looks—I’d say that’s unpleasant. Look, today there are places that are already talking about 80% of the code being written and not looked at, and so on and so forth. I don’t think that today there is anyone working in Cursor who isn’t coding, who is literally sitting and looking at the code. Okay, and I think that will also continue to evolve; in the end the question is where you’re headed. Because we both understand that if now you wrote something in Cursor and tomorrow you want to take it to production, you’d pass it through a human. Who is a developer, and it would go through; in your case maybe you know the code, so it’s easy for you to insert yourself into the event, but you’d want some kind of validation. And not code on the fly. Okay, at no level of responsibility does that mean you’d do it, like for people who do “build, build, build” and put it live and don’t understand why all their databases are out there. It’s a bit like that CloudBot thing that just came out, that everyone is very eager to install, but the thing exposes that a lot of money can be stolen through it. So like with everything, it’s a matter of responsibility and so on; I do think that for simple businesses building tools for themselves. No, it’s not very important for you to look at the code right now; what is important is to do—I’ll say mitigations—maybe yes, maybe that’s very important to me. I think it’s important for you at the learning level with your husband, but if you already know everything, what do you say, I’m letting go—not for me, for the organizations you mentioned. No, I’m not talking about organizations, I’m talking about the working person who does it in organizations; in organizations there’s no such thing. What internal tool? There’s no such thing. You don’t—you don’t want that either; it’s enough that it has data. No, I don’t see that today in any situation you’re working in either; there’s no such thing. How do they work? Sorry, we’re opening parentheses again. But you’re going very much in the direction of development, because in Cursor. In the business side, there aren’t today organizations that give Cursor to anyone who wants it—no, but today everyone is building code. Everyone is building internal tools; internal tools can be tools that are in an enterprise package environment, and I’m allowed to build them, and there’s someone sitting in-house whose job is to decide what I’m exposed to and what not—to take it all, as a bunch, and say Eyal earns xyz—that’s not the kind of internal tools. These are internal tools where you and I are on the same team, and now we feel like it. We have something that you know is a problem, and we’ll solve it with a tool that we’ll both use, or chain between us—that’s great, cool, and I’m for it, and the more there is, the better the work will be. Let’s not forget the goal; the goal is not to work with My, I don’t do that either, and it’s also not always right—you know why? Because again, what is efficiency, how do you measure efficiency? It’s not. Look, you can measure how long it took when he did it, and even now that’s not right because I already had internal, and not for efficiency—what other reasons are there? Because it’s a different way to work, because it gives you, I’ll explain, because it gives you an entirely different dimension. If I have a Racer. That is now sitting and bringing me all the things that used to take me months and I had to have a person do Racer for me. It’s not just a matter of efficiency and speed, because sometimes it may not be faster for me to read everything it brings me, but suddenly it pours out a huge amount of knowledge and data that very much affects my decisions. That’s innovation, that’s looking at things differently, that’s knowing how to stay up to date in a world that never stops updating itself. Listen, sometimes I say things in the morning that by evening I get—very evening—they suddenly blow up and change the company’s entire business direction; I agree, I’m also in such a competitive world as an organization, so I really agree, it’s also because your business model changes. Here’s what just happened with OpenAI, okay? Until OpenAI said there would be ads, half the market shook, but wait, what’s happening, what will happen now? There are entire companies whose business was built on that. Okay, now sometimes that’s great, sometimes it’s like that, sometimes it isn’t—but again, that’s only Racer, that’s only an example; but take, למשל, teams that need to work together. Okay, like the whole SDLC story, everything about how we develop a product — if I know better how to connect with my developer. If I can ask the codebase questions and not end up with a mess when I arrive, that improves me. It’s not necessarily more efficient. It could be that it took me more time now because I spent an hour asking the code something I wouldn’t have invested in before. On one hand, on the other hand, I saved him time that maybe is worth money; it depends on time. I’m not more efficient. Efficient at the organizational level, I meant. Okay, but the organization doesn’t look at it that way; in performance, I’m measured.

Eyal David: [42:23] Okay, but in the end that points to more developers, and that costs money. Let’s put that on the table, what you’re saying, because I agreed with you — agree with me, come on, come on, and let’s move on, so. Okay, I interrupted you, and I interrupted your whole train of thought and we got into this, and that’s fine, so let’s continue with this index. I think north, we talked about code, that was basically the issue. And let me say for a second, I’ll give one more line — I think farther north than that I already heard Hiten Levoff on some podcast, and he talked about how he sits on his phone, and yeah, I also, I’m also, dude, it’s not, bro, farther north, it’s.

Galit Galperin: [42:57] It’s an addiction. I also, I also found myself. In a more serious message, honestly, I was in the middle of some code and didn’t want to close the laptop. I put the laptop next to me and at every station I did one more thing, like I really wanted to see the product finished and I couldn’t, really couldn’t, like, bear the fact that it was from the phone, it’s. It’s possible, but it’s kind of, really kind of, I’ll tell you, world possibilities, you can’t really create like that. It’s kind of maybe an extension to continue. I think the next stage, I call it more mindset order. Mindset, which is also why I really named the company that way, because I understood at a very early stage. That when you’re something, there’s some stage you reach where you realize something, you have some sort of eureka, usually tied to some very big value you suddenly received. And by the way, it doesn’t matter whether it’s personal or work that you. That you suddenly start thinking that with everything you say, wait. Maybe it could help me with this, or how it could help me with this, or you really start examining it in this lens. And that’s already a place where I’d say it’s more maybe code — I don’t really like limiting it to code, because you could also be just a project manager or an HR person; you don’t have to be a product person, okay, it’s really a mindset that says, wait. Maybe I can solve this with AI, and then I start looking for the tools, and sometimes it’s also people I find, it doesn’t matter in the end. And I know, by the way, someone who is HR, who פשוט calls the “mindset orders” — that’s what she calls it. Because she took some process and understood she wanted to change it and started looking, and she took herself to a workshop and she took herself to experiments and she took herself down a path, and she brought someone in — her mentor in the organization — and she built. And she suddenly understands that this is something she couldn’t solve until today, and in her head something changed toward this thing. I mean, I’ll say, I’m all about this mindset. Wait, not sure I need — I don’t know — to take it outside and bring someone in, or maybe I can’t do it alone. It kind of depends where the limitation sits for you; sometimes it’s.

What AI Mindset is and why it matters

Eyal David: [45:22] If I look at it as a CEO at the organizational level, then I’m looking through mindset glasses, and I do want a person in the end.

Galit Galperin: [45:31] By the way, that’s not something I know how to say like you do, because not everyone wants a person in the end; some want fewer people in the end, no, I mean.

Eyal David: [45:42] I now want to execute some goal, like in the example of that HR woman; I don’t know in her role, and in the end I don’t want to move the cheese in my organization, I don’t know what that is, who said because. I want to keep advancing on the business path I’m on, I don’t know if I want to, maybe I want to bring in some—you want to, but today you don’t have something very healthy, so how do I deal with it, how do I actually look at it as a CEO.

Galit Galperin: [46:12] I think you’re the CEO who doesn’t understand that he has to — first of all, that’s a backup. And this year is stronger than year seven, because you understand that you can’t, that you understand what it means — you can’t. You can, but the question is how much you stretch it; כלומר, how much is still left for me.

Eyal David: [46:32] No, I’m coming into the event wanting to try it. I’m saying I want it, but a person to start it — I want, right, exactly. Exactly how do you bring the person to start it, because how do I go into such an effort — what I would say for recruitment, not recruitment, but for just entering this state, that it was a mindset, a backup, for example the HR’s gift, it’s clear to me. In the organizational example, I’m trying to understand.

Galit Galperin: [46:55] I’m starting to break down the different modules; you have to understand what it means for your organization. Sometimes that means, by the way. It can be as simple as what devices they need — I’ll buy what’s missing; tell me what you need for your work and I’ll bring it to you. Which sounds like an easy question — not an easy question — but let’s say that’s the case here, and it can be at the level. I’m kind of comparing it now, I’m kind of applying it, but I’ll compare it to the COVID days, okay? When COVID arrived. It put a lot, a lot of companies in a pants-down situation, like, what do I do now? People aren’t out there, I don’t know how to sell. Suddenly there were — I don’t know if you remember this, I suddenly realized there were tons and tons. Of these small commerce initiatives that suddenly just sell you things because nobody goes out, and who would have thought it would be, like, a situation like that. So sometimes reality puts you in some place that, I’d say, introduces a new floor. That suddenly, wait, it gives me different things; maybe I’ll try to do something differently. דווקא, not stay in this place, but wait, I’ll really open some sort of floor of innovation, of doing things differently; maybe I’ll change the business model, maybe I’ll examine it. Once, twice, I need to understand what is stopping me. Because sometimes speed stops me. I mean, my organization is very cumbersome. Sometimes we’ve worn out, we’ve exhausted people; my organization doesn’t know how to learn what to do, what to do. Learn—I’ll bring people, I’ll bring machines, I’ll teach the people to work with the machines, like what does that mean, how will that look? It touches every profession, like a motor manager. In the end, I need to manage in quirks. I just, just today I said in the morning that five were born to me; at night, by the way, they moved a workshop file for me. From place A to place B, which I didn’t leave, but they had a party at night. How do I manage that? Who tells me how to manage that? Who tells me how to manage that? Who teaches me this thing, and if I’m now a senior manager in an organization, how do I manage this? I’m in sales right now; what do I do now when all kinds of things are being wanted from me on the internet, and they do scraping on me, and someone needs to verify the information, and how do my people even know now how to verify this information בכלל? Suddenly they’ll have hundreds of leads, when before it took them a month to get one. What does that mean for this business at the scale it has to deal with? The whole reality is not a reality where, like, you understand everything; it’s new. The question is just where?

How an organization starts adopting AI seriously

Eyal David: [49:20] Like, what I understand from what you’re saying is it’s very hard to plan if I’m in this hat. I need to understand that I’m probably going into some kind of grace period where I’m learning for a bit, right? Because I want to see some ground move. You’ll say, with that example you gave about leads, maybe I can, maybe I can’t, and then suddenly after some time, when I see what’s happening, I can fine tune it, I can manage risks, I can plan suddenly—that’s basically what you’re saying.

Why you can no longer plan a year ahead like before

Galit Galperin: [49:44] No, I’m saying—and I’m saying this based on how I interpret the world, of course. That doesn’t mean it’s right or wrong—but I’m saying that you’re not entering a grace period; you’re entering a period with a lot of uncertainty, and you can no longer plan for a year. Not that I once thought you could, because it was nice to say where I want to be next year. It’s fine to set the goal, but what you need to understand is that the iterations between you now and you in a year are much shorter and will be continuous learning. That means it’s not a grace period of learning; it’s learning all the time. I do things fast, I fail fast, I do things fast, I fail fast, I got leads, I want to check them now—I’m not going to wait a quarter to check the leads.

Eyal David: [50:32] And now if you need to compare that to people.

Galit Galperin: [50:36] The same thing. Look at people, I mean, רגע I need to say, I’m wired for learning, okay? You can put it on the table: I’m a person who learns every minute of the day that I can learn something, I learn something. And I understand that that’s not a trait everyone has. But I think it’s a duty today for everyone to understand what in the gap between what I do in life and how it can help me in life, I need to learn. And even if that means learning that this tool is, sorry for the French, crap and doesn’t help me, and I can’t rely on it and I abandon it—that’s also learning to me. Even to say it’s not good, or even to say it’s not good, what am I not understanding? Or to say, wow, this is good, I want you to bring me another one like this.

Eyal David: [51:22] So what are the things that right now, tomorrow morning, you can teach us about learning that work for you?

How to learn well when the pace of change doesn't stop

Galit Galperin: [51:28] First of all, I think it’s a combination of format and time. I mean, people learn differently. I know quite a few people whose best way of learning is really in a podcast, in traffic, on the way, because it occupies their head, and then their thoughts race, and they don’t even realize they’ve learned, but the next day suddenly, how do you say? Wait, I do it from the phone, and wait, I’ll do it from the phone, because suddenly it came up. There are people who put themselves in a learning group, a class, a course, whatever. There are people who get it at work, okay? And they have something like a Pareto principle of, come choose how you want to do the learning, but know yourself, because it’s not right to put in general, okay? In learning, learning needs internal motivation. External motivation will almost never lead to meaningful learning, and therefore I think the duty is קודם כל to understand what works for me. I like learning from offices, from writing, what’s good for me? Maybe I actually like watching LinkedIn, doing the swipe, getting my dopamine, and then going into Chat and trying to do what the person did, or whatever. Whatever format works for you, great. Because what happens a lot of the time is people say to me, tell me after one word, and then I say, great, but I can give you five agents who want from me 200 to 500 sources—is that what will help you learn? No, because that’s how Galit likes to learn. Galit needs bullets. I understand quickly; I don’t need to read articles now. That’s how it is for me. Either there are academic articles, pilea, that I read, or bullets. Who is this suitable for?

Eyal David: [53:02] So you start with characterization, that’s what you’re saying.

Galit Galperin: [53:04] A full characterization of where I am, what I need. I might also say, “Wow, I don’t know what I need.” And then you see, wait a second — and again I’m saying time and format, because that time too, where is it? You’re telling me I need much more than an hour; you’re, so to speak, a king, but people don’t have more than an hour. Inshallah, but yes. No, not inshallah, really, I’m telling you, because that’s what I tell people: set aside an hour, and they tell me they don’t have that hour. There is no minimum hour; less than an hour won’t help. Stick it to your run, what difference does it make? Like, do it wherever you know it fits, okay? Or there are many people, you know, it’s a bit like with dieting, when people say, how will I succeed at dieting? Some like doing it alone, some need a fitness coach, and some need a dietitian sitting right here, or a group that shames you, where you come and step on the scale in front of everyone, and only then does it work for them.

Eyal David: [53:58] What’s the additional tip for someone who’s already at a more advanced stage? I’m very anxious about your time, that’s why I’m generating.

Galit Galperin: [54:03] Don’t be anxious. I think the tip, I’ll say, for someone who’s more advanced, is simply to try to stretch it toward something they haven’t tried yet. Because when you say advanced, I’m thinking, okay, that’s someone who has already built the tool, who maybe already did the no-code, who can already jump ahead. You know, it’s very broad; I’m trying for a second, I’m trying, like, I think that someone who got on the train and is curious already knows what to do. It’s framing your brain. Because you’re already in the experiment, you’re already in the cursor. Okay? You’re already, so to speak, you’re already in exploration. So now it’s basically just a matter of curiosity. What else can I do? Just, I made a presentation in HTML, I can make it so that it also, that I’ll click on the phone and it’ll work, I can, meaning, you really start experimenting with a lot of things, which again, for everyone it’s something different, because let’s not forget, there are also video people, there are also voice people, there are also, I mean, you can do a lot of things, you can make an agent, and then you can decide you want to make a fully agentic one, and then you can decide you want it in Hebrew; that’s harder, okay? Like, each person takes, I’ll say, their own area, but if you’re already in code, then you already, you already know.

Eyal David: [55:27] Okay, and I’ll ask you one more question, to wrap up. If you had to choose now a few fields that are worth building businesses and products in? Choose even one or two. We already understood this won’t be an evergreen episode, so like, give it to me straight.

Where Galit sees opportunities to build products and businesses

Galit Galperin: [55:46] A field to build a business in. I’ll say there are two fields in my eyes.

Eyal David: [55:55] We had a few things here, by the way.

Galit Galperin: [55:56] Yes, but I think at a very, very broad level, I’ll say it at the level of fields: it’s really this story of cyber and defense, because it will be, and it’s already a jungle, very scary. And the second thing is at the level of management. Because in the end, really, we need to manage all of this. There are so many, so many things today that are, like, stitched together and not managed. Someone, something, a system — you’d need that.

Eyal David: [56:29] We’ll get that, and also basically manage the learning process itself—that’s something to wonder about if the models get into that.

Galit Galperin: [56:36] The models are there, the models are there; unfortunately, by the way, there was one model that Google made, it was in beta and it was really for teachers. I taught it to lecturers in academia, okay? Because it really helps you prepare the lesson.