Bolt Insight’s Human-in-the-Loop AI Research

In this episode, Lenny Murphy sits down with Hakan Yurdakul, CEO of Bolt Insight, to unpack how AI-native research is reshaping qual and quant. Hakan shares his Unilever-to-founder origin story and the personal “bring summer forward” purpose that drives Bolt’s human-centric philosophy. They dig into BoltChatAI and why AI-moderated qual only works when it’s trained on real human interviews and supported by rigorous human-in-the-loop checkpoints.

The conversation explores Bolt’s Dynamic Personas—living profiles that refresh with ongoing human input—plus the limits of synthetic data for innovation. Finally, Hakan lays out a “Jarvis-style” vision for an AI assistant insights officer, and the duo reflects on what this shift means for trust, ethics, and the evolving role of researchers as strategists and storytellers.

Key Discussion Points:

• Hakan’s path from Unilever marketing/insights to founding Bolt Insight—and the “bring summer forward” mission guiding the company.
• What it means to be an AI-native research firm across quant + qual, and why “qual at scale” is now feasible.
• Human-in-the-loop AI moderation: training on real interviews, quality scoring, and guardrails to avoid “chatbot research.”
• Dynamic (living) personas and meta-analysis: how Bolt keeps personas current and useful without over-relying on historic synthetic data.
• The next 2–5 years: AI automates grunt work, pushing researchers toward strategy, storytelling, and doing more with the same.

Resources & Links:

• Bolt Insight
• BoltChatAI features: Dynamic Personas and Meta-Analysis.
• 2025 GRIT Business & Innovation Report

You can reach out to Hakan Yurdakul on LinkedIn.

Many thanks to Hakan Yurdakul for being our guest. Thanks also to our production team and our editor at Big Bad Audio.

Transcript

[00:00:10] Lenny: Hello everybody. It’s Lenny Murphy with another edition of the Greenbook Podcast. Thank you for taking time out of your day to spend it with myself and my guest. And today I’m joined by Hakan Yurdakul, the CEO of Bolt Insight. Hakan, Welcome.

[00:00:26] Hakan: Hi, Lenny. Thank you. I’m glad that I’m participating to this podcast.

[00:00:33] Lenny: Well, let’s see if you feel that way afterwards.

[00:00:36] Hakan: [laugh].

[00:00:38] Lenny: But [laugh] hopefully you will. And glad to have you here as well. Actually, Bolt has been on my list of companies that I wanted to talk to and get to know for a while. I've been following your trajectory and path overall, but have not had a chance to get to know you or dive into the company, so I am definitely looking forward to this. Which is probably a good segue. So, why don’t you tell us your origin story, tell our audience a bit more about yourself and specifically what it means to bring summer forward. You need to put that in as we talk about this, who you are and a little bit about Bolt.

[00:01:16] Hakan: Absolutely. Thank you, Lenny. So yeah, I come from consumer marketing side. I used to work at Unilever for around 13, 14, years, and before that, Bayer HealthCare, but always been on the corporate marketing, consumer marketing side. Together with one of my co-founders, who was in insights, and we used to work together at Unilever before as marketing and insights, and we left Unilever to create something we wish we had. That’s how it all started with Bolt. But coming to the purpose, yeah, it’s a long story, a bit of a personal one, but happy to share. I grew up in a family, I mean, where my parents were divorced and I was raised by my single mom, and winter meant, like, I came to pretty much an empty house after school. And then when summer came, I used to go to my grandparents’ house, which meant, like, you know, meeting with new people, having conversations, whatever you ask for is ready by the end of the day, the classic grandparents pampering the grandkids. And so, that meant, like, a sense of feeling mattered. And I said, look, my purpose is to bring summer forward for myself and everyone around me. That’s where it comes from. So, it’s very personal, but in the same way, how can I make the lives of people around me better by what I do and what I say?

[00:03:00] Lenny: I’d love that. That almost—I almost teared up there a little bit Hakan because I just imagine you’re coming home to the [laugh] single—[laugh] and in summer, you go to grandma’s house. But that sentiment, I love that sentiment, truly, so thank you for baking that in. I suspect it was hard to live by that idea when you’re working for a large company like Unilever or Bayer. Was that part of the imperative here of not just seeing the business opportunity, but saying, well, I want to make a difference. I want to bring summer forward, and that experience for folks. Is that right, that it was kind of tough in Unilever so you decided, nope, I just need to do this.

[00:03:48] Hakan: Absolutely. I mean, sometimes I was called the crazy kid in the team [laugh]. You know that’s, that’s what results into, I mean, by coming up with creative ideas and bringing some ideas into the discussion where people have not dared to try before. I try to bring summer forward in a good way. But that’s what people used to call me in a nice way. Like, how come the crazy kid, the entrepreneur inside the company? So, I’ve managed to stay in corporate, I think, by creating a legacy which is around, you know, coming up with all those creative, crazy ideas that people have not tried before. So, I think at some point, everybody, including myself, knew that I would step out and try something out myself. And that’s how kind of both journeys started.

[00:04:44] Lenny: I have to ask, at Unilever, did you work with Stan—

[00:04:48] Hakan: Yes.

[00:04:49] Lenny: —during his tenure?

[00:04:51] Hakan: Yes, absolutely.

[00:04:52] Lenny: So, that entrepr—I know he loved the entrepreneurial spirit, so I suspect that craziness, that you had at least had a receptive audience with Stan.

[00:04:59] Hakan: Yeah, absolutely. He used to say—actually, I always quote him with his words around, “Better 80% right rather than precisely late.” [laugh].

[00:05:11] Lenny: [laugh].

[00:05:13] Hakan: Something along those lines.

[00:05:16] Lenny: Yeah, yeah, yeah.

[00:05:18] Hakan: He—I don’t know if he knows this. I haven’t, I think, shared this with him, but he has some ideas, which became the foundation of Bolt as well, later on.

[00:05:29] Lenny: That is not a surprise. He actually played a role in starting IIEX and all of those things. I remember vividly being at dinner, when he was still at Coke, and having dinner with him in Atlanta and, “Lenny, you should go do this. The industry needs this.” “Okay, Stan. That’s a good idea.” [laugh]. Anyway, we can go on and on about Stan, but that’s what he does is helping to inspire people. Well, let’s talk about Bolt. So, some of the ideas obviously came from Unilever gaps that you saw, inspiration from other folks. Let’s take us into that, you know. What does Bolt do today? And—yeah, what does Bolt—what does Bolt do?

[00:06:19] Hakan: We are an AI-native research company, and we have solutions in quant and qual, and we offer self-serve as well as full-service solutions. But we believe in the human-in-the-loop approach in all the services we offer. So, Bolt started as a quantitative research company back in 2020s, and then we moved into qualitative. We were actually one of the first, if not the first, company, to launch AI-moderated qual at scale. Now, it’s becoming one of the norms in the industry and I think the lines between quant and qual is getting blurred. But we started working on this even before the consumer-side AI boom that happened with OpenAI and ChatGPT, back in 2023.

[00:07:23] Lenny: So, obviously you’ve benefited from that boom. I—let’s see if you can relate to this—many years ago, I owned a research company, and I thought virtual focus groups would just—oh, this is going to be fantastic. Everybody’s going to do this. Couldn’t give it away [laugh]. And then along came 2020, right? Obviously, everything went virtual. Were you having a similar AI-moderated groups, and people were scratching their heads thinking, well, what is that? And then along came ChatGPT, and I assume you’ve been able to surf that wave further from there.

[00:08:05] Hakan: Absolutely. I mean, our advantage was, we actually launched our qualitative offering as a—it was called Bolt Chat, without the AI. So, it was a tool where human researchers could connect with the human participants. But then, as we started experimenting with AI, we saw that AI can actually scale things in an unprecedented way. And the idea of Bolt and Bolt Chat specifically was coming from the fact that a lot of insights professionals, marketeers in these big corporates cannot actually find enough time to speak to consumers. So, consumer empathy is a big area because a lot of the marketeers, insights people, they come from, I would say, you know, compared to the general population, come from good, privileged backgrounds, living in main, big cities, but looking after huge brands that touch upon the lives of a lot of people. And you cannot become excellent brand marketing or insight people without having enough connections with the consumers and building empathy. And what we saw was, yes, human-to-human touch and that connection is extremely important, but you cannot scale that because of the sheer time limitations. And especially when it comes to global roles, if you need to reach out to other people in different countries, language becomes a barrier, and due to time-cost efficiencies, it doesn’t become possible to talk to hundreds of people, thousands of people. And AI gave that capability. And from a technical perspective, it also helped us to use the tool, the one that was built for human-to-human connections, to create a lot of large data sets that we can use AI on, use to train AI on. And that’s why we are not just, you know, LLM wrapper, ChatGPT wrapper, but rather created an AI moderator that is built on actual, professionally moderated human-to-human conversations. And I would say that’s been our biggest differentiator since.

[00:10:28] Lenny: Yeah. I mean, as the first company started emerging with post-ChatGPT, a lot were not particularly good, right? I did take a look at them, and thought I see the potential, but you need a moderator to teach this model how to actually [laugh] moderate. And you guys wrote it. Which actually coming from Unilever, right, and the focus on qual and qualitative standards, I expect that was part of the inspiration there as well, right?

[00:11:01] Hakan: Absolutely, absolutely. There was, I think it was called consumer connection passports. They used to have these passports for every employee where you had to complete some of the tasks to be able to get that final passport, like enough connections with consumers. But there were a lot of guidelines that you need to learn and go through. Because, you know, I think the problem with LLMs and no-human-touch approach that some companies have is quite, you know, risky, in my opinion. Because AI is great, but without the right guidance, it doesn’t do research. It just does chats; it’s a chat bot. But research is different, and you need to ask the questions in the right way to be able to get to the unbiased, real answers—

[00:11:56] Lenny: Consistently.

[00:11:58] Hakan: Consistently, exactly, with minimum say-do gap. I always give this example, I hope some of the listeners are not hearing again and again, but it’s again, actually a quote from my mom. She used to say, when a guest comes to your house, you don’t ask them, would you like to drink something? You ask them, what would you like to drink? They sound like very similar questions, but can lead to very different answers and behaviors. So, that’s what research is about. Asking the right question can definitely lead to, you know, the right answer or the wrong answer. So, that’s what we do in the background. That’s why we have a researcher team who can support the clients from project setup to reporting. Even if AI does it all, we have checkpoints in place where researchers, before AI kicks off, checks the project submissions. When the AI report is out a human checks the report and adds human commentary on it. But on top, we have a researcher team whose only job actually to review AI interviews and rate the quality of questions and answers and feed that back into the AI so that it starts to understand what is a good question that leads to good answers or bad question that leads to bad answers. So, that training and control checkpoints needs to be in place when it comes to dealing with AI.

[00:13:35] Lenny: Agreed one hundred percent, and I love that model. The human-AI collaboration augmentation, right? And I think, you know, it’s easy to think how AI can augment us, but it is a two-sided relationship, right, in how we augment AI. Recently, I’ve had similar experiences as the deeper I use, my favorite is Perplexity. I just think Perplexity is just a solid workhorse, right, for business purposes. And the more I train it, the better it gets in meeting my needs and expectations overall, and that’s fantastic, that human collaboration. And I have yet to see any full automated SaaS platform succeed in this industry. There is always a human element, some level of service, not just, you know, kind of, the service you see for normal SaaS platforms, but to have somebody to help through the process, through the research process, has been necessary for success for every single tech company in the space. So, I’m glad you leaned into that from the get-go. Because some people start, no, we’re going to be pure tech. We’re just going to, you know, sell licenses. And, like, we can’t give it away. Yeah, because—[laugh] it’s a little more complex than that. So, I’m glad that you leaned into that from the get-go.

[00:15:04] Hakan: Yeah, unfortunately, you know, some businesses, some companies, fall for that. But I mean, that’s particularly the reason why we don’t even call the team ‘client success.’ It isn’t about client success; we are researchers. You need the researchers to be able to support you. Not because our tool is not self-serve capable. It is fully self-serve capable. You can complete the project without talking to anyone, but even then, we have, in the background people, as I said, like, checking your project setup and checking the AI report before it gets visible. So, if you don’t have that, you’re basically stitching together a chatbot and a couple of front interface to generate a report. I mean, I think soon the danger I see for all those companies are probably because the reason why they behave that way is they’re heavily funded by VCs and a lot of the venture capitalists pushing these companies into doing only SaaS models and license sale and all that, but the danger I see is, AI is also getting better and better in coding, and soon there will be enterprise AI coders, which means, if your tool is not bringing any additional value on top of just lines of code, sorry, a lot of these companies will create their own research platforms and put their own people around that. So, what is the additional value that you’re generating? And again, I think this was actually from Nitesh, if I remember correctly, he used to say—the head of insights at Danone—he used to say, “If everybody is using AI to generate marketing materials, then everybody’s advertising will look the same as well.” So, I think similar with insights as well. If you’re just using the same AI without any additional layers of know-how and expertise on top, what you will get will be generic insights, in my opinion.

[00:17:16] Lenny: Well, and I think your background coming from the client side is people need to hear that, right? You have a very informed opinion about what the real need is. Obviously, there’s many business use cases where full automation, full AI will probably be—just back to Stan’s point, right, 80% right for 20% of the cost that was the other part of his quote—sure, okay, totally get that, but the more strategic and important and consequential the decision is, then the human element, whether that’s intuition, or subject-matter expertise, or just whatever that magical thing is that we do in our brains where we suddenly have an insight [laugh] that’s what differentiates success from a client perspective, would be my take, and it sounds like you’ve built the business both from a technology and a business model perspective, to incorporate both of those things.

[00:18:26] Hakan: Absolutely, absolutely.

[00:18:29] Lenny: So, has that been a challenge? Obviously, one of the issues of being a founder of an early-stage tech company is funding. And you’re right, you know, lots of stupid money goes after the pure tech companies—or maybe not stupid, but you know, they’re always the belle of the ball, so to speak. How’s it been in bootstrapping the company, looking at funding, you know, all of those things to help propel growth?

[00:19:02] Hakan: Absolutely. I think we’ve been fortunate enough to grow pretty much bootstrapping until now. Yes, we’ve had some small seed funding in the very early days, but then we’ve been bootstrapping for a long while now because the company had been profitable, growing really fast. I think we’re one of the few early-stage companies, I would say, who reached, kind of, profitability at, you know, early days. Which is—

[00:19:34] Lenny: Yeah, funny, how bootstrapping makes that happen, right?

[00:19:38] Hakan: Yeah exactly [laugh]. Exactly. Exactly, yeah. Maybe sometimes we don’t throw those big yacht parties, but you know, we invest in our business. It doesn’t mean you know, we’re not bringing cost efficiencies, by the way. Like, our clients, they officially confirm that they have made huge cost efficiencies using our solutions and platforms. The reason being, you know, we build our technology in-house, and we do it with, you know, people who are mostly coming from the other side of the table, being there. So, we don’t need to do a lot of try, test, and pivots because we exactly know what our clients need, we say, so that gives us an advantage. And yeah, and on a positive also note, like, we’re also going to announce a Series A round soon. I’m probably—you’re the first one that I’m sharing this with officially outside the company.

[00:20:39] Lenny: Aww [laugh].

[00:20:41] Hakan: So, thanks for the opportunity. It’s just like, I can’t share names and numbers right now due to some of the confidentialities that needs to be approved before, but hopefully next week it will be out.

[00:20:56] Lenny: Good. So, what—come on, give me—billion dollar?

[00:21:00] Hakan: No, [laugh]. We grow—

[00:21:03] Lenny: It’s AI. It’s got to be a billion dollars [laugh].

[00:21:07] Hakan: [laugh]. No, no AI washing here. It’s all about growing sustainably, creating more solutions for our clients, rather than just mindless growth here and there, in expense of quality. But, yeah, it’s going to be more reasonable numbers, and the money will be used to build new products, new solutions, increasing the capabilities further.

[00:21:34] Lenny: Very cool. We’ll enjoy that ride.

[00:21:37] Hakan: Oh, thank you, Lenny.

[00:21:40] Lenny: Whether it’s, you know, a million dollars or a billion dollars, it’s all propelled growth, and it’s early-stage success, so congratulations. I look forward to hearing more about that. So, you mentioned new products. You know, one of the things that you’ve been talking about is dynamic personas. Obviously, the whole industry now is thinking about this big bucket of synthetic, right, and how we can leverage data to, you know, try and predict human behavior now in a new way. So, can you talk about dynamic personas, and maybe give us a glimpse of what some of that, sweet VC cash may go towards, some of the things that you’re thinking about?

[00:22:24] Hakan: [laugh]. Maybe I start from the end. So, the vision for us is to create that AI insights, assistant insights [officer 00:22:33]. So, we’re working towards that. It doesn’t mean we will have to build and develop every single part of that vision ourselves, but it will be more integrations, connections with other solutions as well. We want to be able to create an agentic system where, you know, it is connected to your internal data lake, your outside data sources, listening, sales data, previous research that you have run, and AI that finds these correlation points, gives you some signals on what’s going well and what’s not going so well, and can autonomously also suggest and run new research to be able to come up with some answers. So, that you know it works on the sides, like, you know Jarvis, almost like augmenting, helping you find some correlation points where you would not be able to spot yourself, maybe, so avoiding some of those silos that most of the businesses have right now, but at the same time, can run new research to bring further insight so that you can make the right decisions fast. So, we’re working towards that. And yes, we started with content. We launched our AI moderated call at scale, so we have all the capabilities. And the last—you know, not last, but another step towards that vision was the personas. Personas in our platform works either and using our own interviews that you have run in the platform, but you can also upload your external data into that and create personas in the system which you can chat with. My perspective on synthetic data and personas is a firm one, and I would say I still have my reservations, and I’ll explain what I mean. For me, I hear a lot of, you know, wrong use cases attached to synthetic data and synthetic personas in the industry. I think if you use synthetic personas for future forward-looking innovation stuff, I think you might be hugely misguided because it’s based on historic data. And then my, kind of, analogy to that is it’s like driving a car by just looking at the rear view mirror. If you use just the synthetic data, it’s like you’re just looking at the past and making some predictions, but you need to always constantly feed that with human data because people change, economies change, lifestyles change, much faster than it used to be. So, if you rely on just synthetic data to validate, you know, your ideas, concepts, or do future ideation work, I think that can become very dangerous in terms of making the right decisions. I mean, but if you’re using it for exploration or understanding categorical stuff, you know, you don’t need to run a new research to find out how many times people wash their hair in a week, you know, things like that. You can, of course, rely on to personas. And what we mean by dynamic personas is—or live personas—is, like, now in our system, you can set up personas, and then it can automatically, on a recurring basis, do interviews with people, collect that information to keep those personas up to date. So, it’s kind of a living system where your personas get updated constantly. We have even, like, applied interesting… services and principles into this. Like, for a UK-based personal care company, what we have done is we’re doing weekly interviews in UK, but in other markets as well, and also, our team is finding out about new products that are getting launched in more influential markets, like, for example, K-beauty, Korea or Japan. Some of these Personal Care Innovations usually start off from these markets. What we do is we find out about these products and then ask about those products and their opinions to consumers that in UK, and then feed that into the persona so that that UK-based AI persona already gets exposed to futuristic stuff. So, when you ask a question to that persona or when you show a new concept idea to that persona, the persona already responds by some of those information coming from the interviews with, you know, futuristic products that we have shown in the past. So, there are some interesting approaches where you can make personas more future-proof rather than just relying on historic data. So, that’s our way of bringing it to life.

[00:28:01] Lenny: I love that, and in violent agreement on—for whatever that’s worth—

[00:28:07] Hakan: Ah, thank you.

[00:28:09] Lenny: The model. I have had very, very similar conversations almost every day now for about two years [laugh] with various folks. This is where we have to go, right? Here’s the use cases for existing data, but it’s not going to work for innovation, right? It’s not going to work when we need to understand more [unintelligible 00:28:32]—I’ve kind of characterized it as the existing data is a mile wide, but an inch deep, so you know, if you’re trying to predict response to—like, visual intensity modeling. You know, sure we can absolutely predict the lady in the red dress is going to pop out [laugh].

[00:28:5] Hakan: [laugh].

[00:28:52] Lenny: So, you know, one hundred percent, got it. And we may be able to say why the lady in the red dress popped out, but we’re not going to be able to predict, but what if it’s a blue dress, necessarily, right? Probably stretching that analogy, but you know what I mean.

[00:29:10] Hakan: Absolutely.

[00:29:12] Lenny: You made a Jarvis reference, so I figured you would get the lady in the red dress reference as well.

[00:29:19] Hakan: Yes, absolutely [laugh].

[00:29:21] Lenny: Which, by the way, well played. Slide in an Ironman reference in there.

[00:29:25] Hakan: For all those Marvel fans.

[00:29:28] Lenny: That’s right, you’re my best friend, now, Hakan [laugh].

[00:29:32] Hakan: [laugh].

[00:29:33] Lenny: Yes, and how long until we are able to just move things around and we’re—anyway, that’s a whole other got that far. So, you are obviously right on the edge, now. You were early in the AI era of research. From what you’re describing to me, I would say that you are, not only were you early, but now you’re on the edge. You’re on the leading edge of what the future looks like and what the business model and products, so that’s fantastic. All of that’s also wrapped up in big picture issues like ethics and trust and transparency and data privacy. So, how are you navigating this rapidly evolving system that also has, kind of, just broad, general, systemic challenges that the whole world is trying to figure out as well?

[00:30:34] Hakan: It’s a very difficult question. I think everybody’s, as you said, like, trying to figure out. What we’re trying to do is… I mean, I’ll go back to the first point I made. Keeping human in the loop is the first and foremost important action you can take when you use AI. I think when we rely on one AI and one software by itself, I think things can go wrong, so definitely having humans in the loop throughout the development process and back office, you know, project management process, but also when it comes to client management, having humans in the loop is absolutely important. Second, I think there is a lot of new certifications and trainings coming in this space. We’re trying to follow them very closely. We have had our, for example—like, in security space—we’ve got our SOC Type 2 certification, but there are others that we’re in the process of. So, having ethical use of AI is extremely important, so we’ve set all those trainings with our team members as well. So, I think it’s all about, how do you use the expertise and human side of things together with AI to serve the clients in the best way possible. Like, we work with, for example, pharma companies as well. They have very different requirements when it comes to the use of AI, like, adverse event trainings, all these are in place, for example, for our research team and employees. So, even when AI has some of those controls in place, we use human controls, as well, on top of that. Perhaps the other thing I can mention is, like, when it comes to our moderation—we call it dual-agent moderation—we have an actually an AI referee monitoring the conversation between AI moderator and the human respondent, and that AI referee’s task is to look into the context of the conversation, looking into the historical data being shared in there, what is the objective of the project, and also looks into other things like if there are any adverse events or other checks that we need to do. And if that happens, AI referee, for example, it even stops AI moderator from hallucinating because before the question is asked, AI referee checks whether that is still relevant and stays true to the discussion guide. So, there are these interesting models that we are using to make sure we have a transparent, trustworthy, and ethical AI solution.

[00:33:35] Lenny: Very cool, very cool. Your answer is unsurprising based on bringing summer forward, right, the that human-centricity, so I see the through-line there from a values perspective. So.

[00:33:53] Hakan: Absolutely. Our mission is, like, we called it actually internally, to make research more conversational. Even that is rooted to my purpose, to be honest. For me, having conversations was… the thing for me. When it comes to when you go to the summer house, you make new conversations. It’s all about conversations. And then I think I haven’t shared that initially, but when I looked at research for 10, 15 years in these companies, what I saw was it was like a blind box-ticking exercise, you know? Like, these surveys were behind the screen, you don’t know the people, you don’t know who they are, you just, like, force them some questions and answers that they have to pick from. And what is the context? What do they think? Why did they have answered in that way, none of that is captured. So, I think the industry will change, for sure. We will see way more conversational models coming into place. The reason why surveys existed is because, like, if you look at the history of research, we had, like, door-to-door, phone interviews, it were still interviews. Yes, you were probably still answering some closed-end questions, and I think I’m not against closed-end questions. I’m against engagements that rely on completely closed-end questions. But then with the online transformation, we brought this concept of surveys for scaling it, but I think we lost a lot in the quality and depth of research as well. And true researchers know that, in my opinion. It will come back. I know it sounds a bit ironic, in a way, that AI makes research more human and conversational, but it is the case.

[00:35:41] Lenny: I again, violent agreement. You know, as we record this today—I’m not quite sure when it’s going to publish, but the new GRIT Report just came out, and the… it’s a tough one in a variety of ways, right? I mean, it’s capturing the shift that we are talking about, and it is unequivocal that that—you know, this transformation that you’re a part of and that we’re discussing is very much showing in the data overall. And it’s a survey, so maybe we need to talk about next year we do something different than a traditional survey. And you’re right. I mean, we came up with the survey with quantitative research as a scaling—it was an engineering issue that we were solving, right? And now we have new tools. And I agree, I think we’re moving into the golden age of research in terms of the ability to engage with humans to get information that helps drive a mutual value exchange and success.

[00:36:49] Hakan: Absolutely.

[00:36:51] Lenny: All the process, how that affects business models, et cetera, et cetera, yeah, that’s all, you know [laugh]—I think we get glimpses of what that’s going to look like. We see that the direction of travel, but it’s going to change and going to change a lot. And some people, some companies, may not navigate that change as well as others. But you know, I live in Amish country in Kentucky; there are still buggy makers, right? But not many [laugh]. There’s still blacksmiths, but not many. So, there is a niche for those, but other technologies transformed aspects of the business. And there’s still going to be—I mean, no shade. There’s going to be a need for buggy makers in the research industry, but not as many.

[00:37:44] Hakan: I’ll share with you an interesting stat. Did you know that all humans in the world speak around 50 quadrillion tokens? And tokens is, like, three-quarters of a word. So, roughly 50 quadrillion tokens. AI in the last 12 months, as of last month, spoke, like, around 30 quadrillion tokens. So, it is estimated that in the next two, three months, AI will actually overtake the human words spoken. So, it’s crazy. Maybe we should call it a world token day, or something, you know? Like, it’s something that humans created will use more words than the humans itself. This is the change that we’re going through, just to, you know, capture the scale of the magnitude of change we’re in. And I think doubting if AI can do good speech or not is just loss of time. I think we need to just learn and guide it in the best way possible for the best use cases. So, that’s my take on AI. I think the sooner we embrace, the better it is.

[00:39:02] Lenny: I get irritated, that’s AI slop. It’s like… okay, I understand what you mean. There’s some things that certainly are poorly written and you know whatever. I understand that, but if it is true, accurate information conveyed in a relevant and impactful way, I think it’s like, you dismiss this at your peril, [unintelligible 00:39:27] AI slop. Some of it is certainly online, right? There’s a lot that’s, like, this really isn’t that great, but when we are trying to communicate information at scale, broad-based, then you cannot debate the efficiency and effectiveness of it, and it gets better and better every single day, literally. I like… I use Perplexity a lot. I also like Grok and they just released Grok 4.1. Just my limited amount of the last 24 hours of using Grok 4.1, it’s like, damn, that’s so much better in so many ways than Grok 4, you know? They just keep leapfrogging in capabilities, which reduces the slop, right? I mean, it’s just getting better and better at a communication standpoint.

[00:40:19] Hakan: There’s nothing to be scared of as well. There are still things that humans do better. I think we have to acknowledge that. For all those people who are scared of AI, I think as long as you get used to it and upscale yourself, I think there’s nothing to be scared of. It just increases your efficiency. You know, the there’s another interesting data that I’ve seen in the last couple of months, like, which was showing AI finding out about the correlation between the chocolate consumption and the number of Nobel Prizes won in countries. So, it shows that the more chocolate you consume, the more Nobel Prices you get [laugh].

[00:41:10] Lenny: But is that a spurious correlation? I don’t know [laugh].

[00:41:14] Hakan: But what I’m trying to come to is, like, AI is great at spotting these correlation, but good researchers again know, correlation is not equal to causation, right?

[00:41:25] Lenny: That’s right. That’s right.

[00:41:28] Hakan: So, when you look at that data, just by AI, you might easily conclude, oh, I probably need to eat more chocolate to get more Nobel Prizes, which is not the case.

[00:41:38] Lenny: Well, I don’t know. I mean, I’ll buy that. I will buy that. If I eat more chocolate, maybe I’ll get a Nobel Prize. It’s a win-win, no matter what, right? So [laugh].

[00:41:46] Hakan: So, to be fair, we may not have spotted that correlation if AI didn’t exist. But then how do you add, you know, your human context, human feeling, human intuition on top of that to go to the real root cause is the art here, you know? And I think that will remain a very important skill reserved for humans as we move forward, especially for researchers, I mean.

[00:42:18] Lenny: Yes. Agreed a hundred percent. You know, my own experience, I was resistant to utilizing AI because, I mean, fundamentally, you know, there’s this public thing that I do with this, but fundamentally, I’m a consultant, and that’s how I really earn a living. And I thought it would cheapen me or devalue me. And I talked about this quite a bit. And then I was forced to utilize it, actually because of demand of, you know, I had, like, three projects all hit at once out of the blue; they all had tight deadlines and, like, there’s no way I can do this unless I use these tools, right, to do what they do. And then I’m like, no, wait, this is a superpower, you know? This is the efficiency gains. But no way could they have replaced me as a researcher, as a consultant, as an advisor because they required not only my management and process, you know, all through, but also my interpretation and contribution on making it all make sense. So, what it saved was just the grunt work of writing, right? Of you know, putting things together, process issues. And so, I agree with you wholeheartedly. So, I want to be conscious of your time as well as our listeners because I think you and I could go on talking for—

[00:43:42] Hakan: I, I—

[00:43:43] Lenny: —a long time [laugh].

[00:43:45] Hakan: —think so as well [laugh].

[00:43:48] Lenny: Yes. And I—

[00:43:49] Hakan: Let’s turn this off and continue, Lenny [laugh].

[00:43:53] Lenny: Okay. And I look forward to having more conversations about this. So, we’re in this period of transition. What does the future of the industry look like for you as you think about the next two to five years, right? What does that horizon look like, not just for Bolt Insights, but as you think about the industry as a whole. Where do you… what do you see happening?

[00:44:22] Hakan: I mean, from an industry perspective, I think it’s pretty much obvious that researchers will become more strategists and storytellers, in my opinion because AI will be able to make most of the operational non-value-adding work automated. So, it means, you know, how do you take all those insights and add the business and human context and create you know impact in the business and create the right storytelling for everyone to understand in the organization? Because one of the researcher’s role is to also create a more consumer, empathetic, consumer-centric organization, and part of that is the storytelling. You know, you can’t expect sending people hundreds of slides, PowerPoints, and expect them to become more consumer-centric out of nowhere. If they read, and even if they read, you know, we have seen many times that people can’t easily become more consumer-centric. So, creating that impact, creating that influence in the organization, is still a researcher’s role because, like, sending answers to questions is just reactive. So, how do you—also maybe by freeing up some time from the repetitive tasks, how do you do some proactive work on top of the reactive work that we’ve been chasing for many years now, is the next step. But on top of that, you know, this is probably a bit of the, you know, controversial side I’m thinking, a lot of the scientists and technologists in this space, they are actually predicting that human cognitive contribution will actually be negative in the next two, three years. So, what do we do around that? How do we remain relevant and value-adding into the equation, is a big question that we should all ask ourselves. I don’t think anybody has the answers, to be honest. And that’s going to be a big change. Imagine being the one in the room with the least cognitive contribution. It’s a bit scary. But I think we need to find our strengths and play on our strengths as well, you know? Some of the points that we have made earlier on is becoming more and more important.

[00:47:10] Lenny: Agreed. I think that last point, as a society right, the—so I often play, which future path are we on, right? Is it Star Trek or Black Mirror? But there’s also Idiocracy, right? It’s like any of those are possible [laugh]. I think we’re heading towards Star Trek, but we may have to pass through Idiocracy to get to Star Trek, is my concern. But—assuming that you’ve seen those movies—

[00:47:40] Hakan: Yes, yes, yes, absolutely.

[00:47:43] Lenny: —and get the reference.

[00:47:46] Hakan: Absolutely.

[00:47:48] Lenny: But, yeah, I hope that we all come to the point where we learn how to think more. That’s certainly, I think what you’re conveying, and my own experience utilizing these tools, is it helps again, process. It’s not a shortcut for thinking, and the thinking is where the real value is, whether we’re researchers or consultants or anything else. So, is there anything that you wish that I had asked or you won this touch on that we did not.

[00:48:18] Hakan: Maybe just one small point, for more the short term, I would say, or mid term. I think what I’m seeing from more than a hundred enterprise clients we work with is, there are—and all the others that we are, you know, in conversations with who are maybe working yet—but what I’m seeing is there are two different approaches in terms of adopting AI and why adopting the AI. One is, can I do the same with less, and the other is, can I do more with the same? Because I think yes, the budgets are shrinking, but as well as the teams, it’s so easy to fall into the trap of, how can I do the same with less money, less people? You know, that’s the easiest approach. But those who are more visionary and look at the holistic picture and more strategically, easily see, you know, how much we were missing due to some of these constraints we’ve had, either human or budget. And AI enables actually, to do more with the same people, same money you have. And I think that’s the direction that we should go after, if not do more, you know, with same or, you know, less budget even. So, I wanted to reiterate that as well, you know? Given we have the perspectives of very different companies, very different industries, and hopefully that’s the vision that a lot of, you know, insights leaders share.

[00:50:03] Lenny: Mm-hm. I love that. That’s a great way to frame it up. So, maybe efficiency versus augmentation, maybe is a way to, kind of, you know, think about that. So Hakan, love this conversation. Congratulations on all the success. I hope that you continue to bring summer closer as you—bring summer forward, sorry—as things progress. Where can people connect with you or learn more about Bolt Insight?

[00:50:33] Hakan: They can check our website, boltinsight.com or can they reach to me as well, [email protected].

[00:50:42] Lenny: All right, thank you. Thank you so much. This great conversation, really appreciate it. I want to thank our editor, Big Bad Audio, thank our sponsors, and thank our listeners because guys, as I said, I’d wanted to connect with Bolt for a long time. Had not. If not for you, listeners, dear listeners, being here, then this wouldn’t have happened, and I wouldn’t have had a chance to meet somebody who I hope that becomes a friend. So, that’s it. Hakan, last words? Anything else?

[00:51:17] Hakan:
I would like to thank everyone as well, and also thank Greenbook and the GRIT Report as well, since you mentioned. We’ve been in the top five most innovative qual companies. Being there in five years of our journey is a big, you know, privilege and proud moment for us, and thanks for that. And thanks for all the people in research professionals who have voted for the GRIT Report, and I was super happy to see that we’re in the top five for the most innovative companies.

[00:51:48] Lenny: Very cool, and congrats. I think your competition is going to get denser—

[00:51:52] Hakan: Yeah [laugh].

[00:51:53] Lenny: —as things progress, but it’s always good to be—

[00:51:56] Hakan: The more the merrier.

[00:51:58] Lenny: —top of mind. That’s right. All right, that’s it for this edition of the Greenbook Podcast. Thanks everybody, and we’ll be back with another one real soon.