"If there's something that's terrible, it's this fuzzy value proposition with AI slop." Alex Osterwalder, co-founder of Strategyzer and inventor of the Business Model Canvas, returns to The Innovation Show with a field report on AI value proposition design. A few weeks ago he argued that AI should never replace the human struggle of value proposition work. Drawing on recent work with a large client's product owners and product managers, he now shows Aidan McCullen where AI earns its place, once a team has done the homework that gives a language model something worth working with. In this conversation, Alex reveals: - The four evidence-backed artifacts every team needs before AI writes a word of its value proposition - Why high-level input to an LLM produces high-level output, and how to stop it - How Strategyzer fed its own customer profiles and value stacks into AI and was surprised by the hero copy that came back - The Value Stack, a new tool that sorts benefits into a key value driver, supporting benefits and table stakes - What Gmail's one gigabyte of free storage in 2004 teaches about the single benefit that brings customers to you - Why a large European marketing team demoted its own "differentiators" to table stakes the first time it used the Value Stack - How the marketer's job shifts towards overseeing AI, much as software developers now oversee coding agents - Why agents can run 100 customer interviews in an hour yet still cannot decide what to build - The reason Alex says AI has no taste, and what that means for product judgment Chapters: 00:00 Sponsor Message 00:22 Why AI For Value Props 01:44 Field Insights And Demand 03:26 Homework Before AI 04:43 Four Core Artifacts 06:33 Generating Assets With AI 08:10 Marketers As Overseers 10:01 Value Stack Explained 15:15 Agents And Human Judgment 17:41 Wrap Up And Giveaway 18:55 Sponsor Closing Notes About Alex Osterwalder: CEO and Founder of Strategyzer, inventor of the Business Model Canvas, and lead author of Business Model Generation and Value Proposition Design. Explore his tools at https://www.strategyzer.com. And I am biased but a damn good guy! Win a digital copy of Value Proposition Design, with thanks to Kyndryl. Subscribe to the Thursday Thought on Substack to enter: https://thethursdaythought.substack.com Get 20% off the Strategyzer platform with code INNOVATIONSHOW20 at https://www.strategyzer.com This series is brought to you by Kyndryl, who run and reimagine the technology systems that drive advantage for the world's leading businesses. With a unique blend of AI-powered consulting built on unmatched managed-service capability, Kyndryl helps leaders harness technology for smarter decisions, faster innovation, and lasting competitive edge. Learn more about Kyndryl and the Kyndryl Institute at https://www.kyndryl.com. About the Host: Aidan McCullen is the 2025 Thinkers50 Innovation Award recipient, keynote speaker on AI, disruption, innovation and change, host of The Innovation Show, and author of Undisruptable: A Mindset of Permanent Reinvention (Wiley). Learn more at https://theinnovationshow.io/about-aidan-mccullen About The Innovation Show: the Thinkers50-recognised podcast where square pegs find their place in a world of round holes, with conversations on disruption, innovation, change, transformation, leadership and AI. Website: https://theinnovationshow.io Apple Podcasts: https://podcasts.apple.com/ie/podcast/the-innovation-show/id1148455669 Spotify: https://open.spotify.com/show/63nFKf4bsSWo3W72gWtOsK?si=b62d9614237c4450 YouTube: https://www.youtube.com/c/theinnovationshow LinkedIn: https://www.linkedin.com/in/aidanmccullen
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
Before we begin, thank you to our sponsor, Kyndryl. With a unique blend of AI-powered consultant built on unmatched managed service capability, Kyndryl helps leaders harness the power of technology for smarter decisions, faster innovation, and a lasting competitive edge. You can find out more about Kyndryl at K-Y-N-D-R-Y-L, kyndryl.com. A few weeks ago , we were joined by today's guest, and he told us about how we do not use AI when it comes to value propositions.
Don't use it to replace the human work. You need the human work. You need to go through that struggle. But now he has joined us again, but this time it's about how do we use AI to augment value propositions. This comes from the field. I love having this guy on the show. He never runs the same lap twice, so when he goes out there and he works with clients, he brings it back, he puts it into his frameworks, and boy, does he love a good framework. He is the author of, for those watching us, all those books on the shelf behind
me, and today's focus is how do you use AI in value proposition design. He is Alex Osterwalder. Welcome back, sir. Thanks for having me back. It's always a huge pleasure to be on your show Thank you for squeezing me in, man. I, I know you're traveling all over the world, just off to Riyadh, and I just got him, like, an hour before he's hitting the airport. So let's not waste our time. And I'm, I'm answering here, by the way, so many emails I've got saying, "Give us more tools. Give us more frameworks. We've had enough with all your knowledge that you have on the show." So we're
gonna give people a little bit of something different today, a shorter episode, which is also being called for, but from the horse's mouth, how do we use AI in value proposition? Maybe, Alex, you'll start with where this is coming from, what you're seeing in the field, what are clients looking for? So we've gotten a lot of demand around value proposition design, and we were brought in by one of the clients to really, on a large scale, support their product owners, product managers. And in that process, we just learn a ton every time.
So we bring in the frameworks. We actually design some new frameworks which will be out on the Strategyzer platform soon, like the buying decision map, value stack, different things. But what struck me most is how these established companies are not fully aware yet how to use AI to the max and how to augment the human work. So they're still kind of just improving processes when actually the work in particular of product owners and marketers are changing.
So I'll talk a little bit about that and, uh, zoom into it. But it's, it's really from the practice. So, so everything we do is informed by practice So let's get stuck in. I know you were working with a client here, and this, I love that you do this. So you'd identify a need, 'cause the, it's like that Heraclitus quote that nobody ever walks in the same river twice. It's not the same river, and they're not the same person. Things keep moving. Technologies keep coming on. That's why we're talking about AI a lot on the show. We're not jumping on the bandwagon.
It is here to stay. So clients are feeling this downstream. So what, what has come from that work? What have you talked, what have you felt from working with these latest clients? I'll quickly visualize and I'll scribble it, and for those of you listening, the voiceover should be clear enough, and if not, Aidan is gonna stop me, right? So I'm gonna talk about how modern companies develop, test, and iterate value propositions, and how AI actually plays into that, and what you need to do. You need to do a lot of homework so that you can use AI well.
And from everything we've seen in practice, the human still has an important role to play, even when at one point you are gonna use agents. But so everything starts with an opportunity. The opportunity could be in technology, in science, whatever, but also, you know, the assets that you have. Maybe you have a lot of a- customer access and you wanna come up with a new product. So you go through this search phase, and then you start to formulate your value propositions. But here's the key. this search phase, product development, you need to do quite a bit of homework
so that you can use AI afterwards to almost automatically generate webpages, brochures, anything you want, and even sales documents, sales decks. But if you don't do the homework first, gonna suck, right? Because it's just going to be l- you know, high-level stuff in and high-level stuff out, and if there's something that's terrible, it's this fuzzy value proposition with AI slop. So what do we need to do? We actually need to start with the basics, and the basics are very simple.
We need a good customer understanding here, and we use a customer profile, but then we need to support that with evidence. So map out your customer. Well, you select the customer segment you want to serve first, but then you test. You go and speak to customers. You can use AI to go faster with synthetic interviews or quick stuff you can do, but you need to do that legwork. You need to get the evidence. Then only do you go to formulating your value proposition the benefits. You don't build anything yet.
Again, you need to then go and test and find the evidence which benefits, which value actually matters most to customers. So again, it's not something you can shortcut. You have to do that. You'll be able to do it a lot faster with AI. So typically, you can put stuff on, you know, the internet very quickly, visualize it, and get feedback overnight. you have to do that homework. And then the third one is visualize your differentiation, not by looking
at just, you know, the competitors, by actually asking your customers, does this differentiation matter? Does this aspect-- You know, is this solved already by our competitors? So again, the second piece here is the evidence, not just the differentiation. now the fourth homework or artifact that we need to create is the buying decision map, this new tool I talked to you about, and then the evidence. So first you figure out how are customers buying, and then the evidence that they're actually willing to buy, willing to spend.
They have that budget. So in the product development process, you need to do this homework. You need to go through these four things, create these four artifacts, test them. And once you've done that, and again, you can do this much faster with AI, but once you've done that, now you have you need, these value proposition ingredients, use the magic of AI to create anything you want. Sales deck, marketing video.
Uh, you know, the power of AI is crazy good when you have the right data to feed in. So then, and I'll just give you two silly examples, and the one that I like a lot is a technique called value scene, a simple storyboard. If you have the right ingredients going into your AI, you can choose any LLM you want. You will have beautiful storyboards coming out. We use value scenes with three pictures. Or you could The second one, create paragraphs like a hero, you
know, title and a hero paragraph and maybe an image for that. So again, once you've done the homework on the left-hand side where you're developing your product and you have those ingredients, it becomes very easy to churn out anything you want, that's where AI is powerful. But the human still needs to iterate, right? So you get a first draft that's pretty powerful. So when we've done this, not just for clients, but we started obviously to use this for Strategyzer.
When we actually took our customer profiles on our platform and we took the value stacks and we said, "Hey, we need a hero line and a hero paragraph," it went so fast and we were blown away ourselves by the, the quality of what came out of that. So now what does that mean for marketers? It changes the job of the marketer. Your job is not to create a brochure anymore. Your job is to use AI to iterate, so you're really more an overseer. So it's very similar to what's happening in coding.
You know, software developers, they don't code anymore, but they need to oversee the process because coding agents alone can't do everything humans can. They can do a lot, but they still need the human oversight. And when you don't have human oversight, you get some of the stuff we're seeing in the press these days We were just talking about that on LinkedIn, for example, the amount of slop that, is out there. I, I just don't understand, Alex, how people don't know it's slop. I, I was do- I did a mock post recently about a kid that gets a gold star for
their writing from the teacher, and it was like, what happens when a kid gets access to LinkedIn and an LLM? And it was, like, going, "Absolutely honored to receive this gold star. It's not about the teaching, it's about the process." And it was all following all these patterns that you can, you can clearly spot now. But it, it goes… The reason I share that is it goes to your point about keeping the human in the loop, that you have to do the work. If you're gonna write something, I, I always was told this, Confucius said supposedly , speak because you have something to say, not because you have to
say something." And I think that that's valuable here in this point as well. There's a couple of things I just wanted to show, and, and maybe you'll show the, the slide behind you again. There's a few definitions in there, because we are gonna double-click on a couple of these and maybe do an episode on these - again in the future. You mentioned your new tool there, value stack. Maybe you'll describe what they are. So customer profile, I think that's self-explanatory. Value stack differentiation, and then willingness to pay. Maybe you'll take us through what they mean.
Yeah, so the value stack is something we added because what we realized is with the current value proposition canvas, which works really well, um, is not as powerful when you wanna formulate value proposition. So what happened with the value stack, it's actually a tool that came out of working with AI when I was working on this process. So right now, I'm pretty amazed how, uh, AI is helping us develop better conceptual tools. So the suggestion of adding some stuff came from AI, and then I realized,
"Hey, that's a tool." So what do we do? We put all the benefits that we have and that we're creating with, with our products and services, we put them into -- one of three boxes, so we kinda rank them. The top box is key benefit, the key value driver. It's the reason why customers will come to you. So I'll give you a silly example from two thousand and four, I think, or whenever, Google launched Gmail. needed something to radically disrupt Hotmail at the time because
they were new into the market. So what did they do? Some people might remember. They said, "We're gonna give you one gigabyte storage for email for free." That's the feature. And the benefit was never delete email again. Because back in the day, you had to delete email to get emails, as some of you might remember. So that was the key reason that attracted millions and millions of users. So you need to figure out what's that one benefit, that one value driver. And once you have that, well, you start with, like, that's the
crown of our value proposition. in the second layer, the second box, you put everything that supports the first one that is valuable, maybe even a strong differentiator and new, but is not the key reason why people will come to you. So if I stick to well, it was the fact that you had threaded emails first time ever, which is very powerful, but people didn't really go to Gmail for the threaded emails, right? Or Google level search in your email inbox. That was also very Goog-google kinda, you know, thing.
But people didn't go there because they didn't even know what that meant. But that was a key differentiator, but not the key benefit. So those are the two, first two boxes. And the third one is, is the one that came out of me working with AI The table stakes. So, you know, in product adoption, there's always been this thing of table stakes. But when we talk about value propositions, in particular new ones, you still have some benefits or features that are table stakes that customers will expect. And there, you know, let's take, you know, reliability of your
email, that would go into Gmail. Those are things that need to be there. But it sounds trivial, but here's what's interesting. When we were working with the client, using this the first time, we even did it with their marketing team. This is like a really, really big, European company. But when we did this the first time, when they had their differentiators and they needed to take the elements of the differentiators and put them into the value stack, of a sudden, some of the things that they said were differentiators, they pulled them down into the, into the table stakes.
So you can see how a simple tool can structure your thinking. now when we show that and the teams have to really rank the benefits, not just stack ranking, but put them into these three categories, of a sudden, the value proposition becomes sharper. So we just classify this, but then imagine now you have data structure, you have the key, value driver, you have the supporting benefits, and then you have table stakes. Now you put that into an LLM, it can much better handle,
okay, what actually matters. That's the reason why these structured tools that we create have so much value. So in the playbooks that we run on the Strategyzer platform, one of the things is while you're doing the work, you're already structuring the data, and that's very rich because then you can use it in AI. And it hit me, I didn't even realize we're actually doing the groundwork for great AI work by getting people into our Strategyzer playbooks. I think that's what people miss, that context is king. You need context for an LLM.
Also, something you said there, so I, see this increasingly happen where, say, for example, you have a team in an organization and somebody's given a task. Say it's a risk team, and you're Alex, you write us a policy on AI use, u- respectful use of AI, ethical use of AI for the risk team. You're, like, going, "Oh, I better say yes." You use AI, you create it, send it to everyone. You haven't written it. Sends it to everybody. Everybody then uses AI to summarize it and go, "Give me a few bullet points
here in case I'm asked a question at the meeting." They ratify it, then it goes up a level to the board to ratify. Nobody there reads it, and they sign off on it, and there's no- literally no human in the loop. And one of the concerns, I wondered what your thoughts are on this, were you still have to go through the work. You still have to go through the struggle. It can augment you, and this is the point of how you- you're, you're using it here now, but you don't replace the human entirely No, you don't replace the human entirely.
I do think, like in coding, in, more strategic work, we're gonna see something similar happen where some of the work will be done by AI agents. So what I just showed you was, like, a bit more manual than we'd hope it would be. So typically, let me say A- an AI agent could be a, you know, army of bots going out and doing customer interviews. Meaning, well, these aren't chatbots that, you know, contact you and then you answer a chatbot.
This is very effective. You could do 100 interviews within an hour. Now, you know, is that the best type of interview? Well, probably not, but it's still, you know, a very useful experiment. And then you can augment that with, you know, face-to-face or remote interviews where you're looking at people. So you, you can do things with agents. They can go and collect that, can synthesize it, can actually create value, you know, uh, uh, can create customer profiles for you. So some of that work will be done by AI agents, so they do the work, they
perform the work, but ultimately, they won't be able to make decisions. I don't believe we're there by far if I'm, you know, talking to the experts. Um, they-- they're, they're still saying that there's things humans can't be… You know, you can't replace the human for that judgment, that judgment call, because AI has no taste. AI has no understanding of like a human. They can do a lot of things. They can do a lot of work really fast. But when you're creating products, there's still a value judgment there,
and that is not replaceable yet. Because let's remember, LLMs are statistics. Like, it's quite amazing to see what stat-- what we can do with statistics. It looks like intelligence, but as you said, if we go a little bit deeper, you know, we can immediately see the limits of this artificial intelligence. So the human will have this role of an overseer, like in coding, and is going to orchestrate quite a lot of agents to do a lot of the work for them. But ultimately, for a lot of these things, in particular in product development,
the human will still have a role. But the profession is changing fundamentally. Let's be clear about that, right? So you're becoming an overseer of agents more and more. That's happened in coding, it will also happen in more strategic work. Okay, we're gonna, we're gonna wrap it there. T- couple of things. One is Alex is joining me here in Dublin in December. December 1st and 2nd. We are working with Enterprise Ireland and a group of companies for Enterprise Ireland. They're in for a real treat. We're gonna do workshops, half-day workshops, morning,
afternoon, over two days. But Alex is gonna be around as well, so if you're a company and looking for a keynote, he's in town, so never a better chance to grab the man, and we're gonna spend a bit of time together. The other thing is I have a copy of Value Proposition Design up for grabs, a digital copy. Here's the question. On Alex's canvas behind him, there's a mountain. What is at the top of the mountain? Email me in admin@theinnovationshow.io. Let us know, and you will be in the hat to win a copy of that book.
As always, Alex, always a pleasure. Author of all those b- I'm not gonna even say them anymore, all those books on the shelf behind me. We're gonna come back, and actually, we're gonna do an episode on the Value Stack, Alex's new tool, and share that with the world f- world first, right, Alex? World first, absolutely World first. Alex Osterwalder, thank you for joining us Real pleasure. Thanks for having me And thank you to our sponsor, Kyndryl, who have powered these book giveaways and much, much more with the Innovation Show.
That's why we're getting around the world a lot more this year. With a unique blend of AI-powered consulting built on unmatched managed service capability, Kyndryl helps leaders harness the power of technology for smarter decisions, faster innovation, and a lasting competitive edge. Don't forget, we have so many brilliant thinkers on the Kyndryl Institute. Many of them have been guests on the Innovation Show. Rita McGrath's on there. Sangeet Paul Choudary's on there. His article got picked up by the FT and so, so many more.
"Machines have no causes. They have only missions." Mohan Nair has worked in artificial intelligence since 1977, and he argues the people who stay unreachable by the machine are the ones who protect their own insight....
A surgeon changes her technique halfway through an operation. A mountain biker rebuilds a part in his garage. Both innovated, and the companies that made their equipment probably never knew. MIT Sloan's Eric von Hippel...
Most organisations have plenty of ideas. Scaling them is where innovation dies, says Harvard's Linda Hill. Linda Hill is a Harvard Business School professor, an ethnographer of leadership, and co-author of Genius at...
"Nobody ever measured management properly." Henry Mintzberg Henry Mintzberg has spent a lifetime taking management myths out of sight for the sake of insight. The McGill professor and Thinkers50 Lifetime Achievement...
Your company is spending real money on AI and the CFO wants to know what it is getting back. David Rogers calls the awkward silence that follows the AI value gap. David Rogers wrote the first book on digital...