“Just Go For It!”, and Other Stories from the Front Lines of AI Imagery, with Cliff Crosbie, VP, AI Consultancy at Bynder

Transcript

Our transcripts are generated by AI. Please excuse any typos and if you have any specific questions please email info@digitalshelfinstitute.org.

Lauren Livak Gilbert (00:00):

Welcome to Unpacking the Digital Shelf, where industry leaders share insights, strategies, and stories to help brands win in the ever-changing world of commerce.

 

Peter Crosby (00:22):

Hey everyone. Peter Crosby here from the Digital Shelf Institute with 4,500 customers across every industry creating digital assets inside their platform. Binder has a bird's eye view of the rapid transformation of the process to make imagery and drive it to market, and how AI is transforming that cycle. Our guest, Cliff Crosbie, no relation, is VP AI consultancy at Binder, and he brings both the stories and data that illuminate the strategies and best practices that are working to drive massively greater efficiency and better results on the digital shelf and the AG Agentic Shelf. Welcome to the podcast Cliff, and thank you so much for being here. And thank you for your superior name, Crosbie, although with the IE spelling, which is I think superior, but I was so delighted to have you on.

 

Cliff Crosbie (01:17):

Yeah, thank you. Thanks Peter. It's great to be here.

 

Peter Crosby (01:21):

You spend so much of your time thinking about this little thing called AI and how it's going to change the way consumers shop and discover products, which is really where kind of all of these conversations should begin, that that's shifting. And so how do we take advantage of those opportunities and challenges? And you've worked with a lot of brands that are thinking about that. So how can they implement ai? How can it help them? So I'd love to dive into that and start really just to see what your definition of agenda commerce is because all over the place and also what you were seeing in the market.

 

Cliff Crosbie (01:57):

Yeah, it is a really big question and it's where to start with it. As binder, we're working now with four and a half thousand customers, so we have a lot of people and in the last year and a half we've had over 1300 of the biggest customers have signed up to the AI programs basically in search and all that kind of thing. And then AgTech, our proper AgTech program launched just in September last year. So we're only that few months into really putting agents into that market in a big way. And I think we're seeing three really big trends that we see at the moment. One is just a massive, massive growth of content. There's just a huge growth of content and that AI is spurring the need for companies to really elevate their thinking, but probably have elevate their digital strategies because with all this coming in, they really have to look at what they're doing because if they're not using it, likely their main competitor is using it.

 

(03:10):

So there's that danger. And also it's transforming content creation, distribution, what operation looks like it can improve speed, execution, productivity. So that's where everyone is really looking and how that's shown up in our world is that 50% of all the assets we have in binder we're created in the last 18 months. And that's crazy because we've been going for 13 years. So it's just proof of that proliferation. It's crazy. And second is that our users have grown by over 40%, nearly 50% now in the last 18 months as well, which shows that more people are touching all the content than ever before as well. So that's really the main trends we see, if you like at the moment. What we do about it is another question I guess you're going to ask me next.

 

Lauren Livak Gilbert (04:08):

Well, just to go back to the agentic commerce piece for a second. When you think about agent at commerce, are you thinking about it in the bucket of autonomous shopping? Are you thinking about it in content creation? I feel like people use that term interchangeably, so would love to hear in your world, how do you define it? What is the way that you think about it in your bubble?

 

Cliff Crosbie (04:34):

Yeah, it is a really good part because if you're a marketeer now, you're not just seeing incremental change, you're seeing, I dunno, a revolution or something like that. It's complete change. And so it's going all the way from content creation, personalization that you'll put in there, campaign execution, performance optimization, all of that in there. So AI is fundamentally transforming how a marketing team work end to end. I think that's the whole thing and that's exciting and then super challenging because while productivity and creativity gains are huge, it introduces complexity and especially if you haven't got the right system in place, when we are talking about AI making a difference, we're looking at this system of record, if you like. So how does a company work with all its assets? It needs a clear system of record and that's no longer optional. It's kind of essential or foundational I would say even because with all the things that are coming on with ai, you have to have a grip on everything you have in the first place, and that's all the way through creation and delivery all the way through.

 

(05:56):

So you are in an era now where what was impossible is now becoming scalable even so not even doable, but doable and scalable. And so it's not about just storing content about anymore, it's about activating all that content in a smart fast way with absolute intention and hopefully huge impact as well. So to answer your question, we're talking about everywhere you touch from creation through to delivery and we're trying to be the system of record and all the tools you need in that process to make that deliverable, it still comes down to that company and its creativity, but then everything we're delivering is tools to try and make that faster, better, all these smarter, all these kinds of things. Does that answer

 

Peter Crosby (06:54):

Question? Yeah. Well yes. Lauren is nodding because it certainly answered it for me, so glad we're on the same page, Lauren. But one of the things that fascinated me about the data that you shared in terms of amount of content and number of users is one of the things we've been talking about a lot is sort of that iceberg of data that's needed for agent commerce in that sort of, there's the e-commerce data and at the very tip of the top, the logistical shipping data, all the physical and digital shelf requirements, and then you have this whole rest of the iceberg that is what do you need to feed the right conversation with the exact person context use case? I still struggle with it or call that personalization or not, that has such a fraught history. But in any event, it seems to me that the number of users getting involved and the amount of content is being produced, I would imagine some good portion of that is around trying to meet that need to have more specific content for more use cases and across more products. Do you have some sense of if that's true, and does that reflect sort of where you think this is going?

 

Cliff Crosbie (08:09):

Yeah, I think it's true. I mean that's the reason there's so much off it now. And of course they're going to cross every single touch point that they can work with as a company. And we see it because I think the need now to play with content in the way that not just have that one asset but use that asset multiple times in multiple ways and get the best value out of it. Where we see it starts is just in the ability to find these assets because we know that when companies go in and use this stuff, the first problem all the time is have I got the right stuff in the right place? That's what I mean by that system of record. That's what we are trying to get into, that when someone really comes in and needs to find a campaign and all the assets associated with it, all that stuff is available, then whatever work they've got to do with it, Peter, we don't get too involved with that.

 

(09:13):

We provide the tools to do that, but we're not too involved in what they want to do to we try and provide the tools to let them do whatever that happens to be because it'll be many different things. They might want to personalize it for certain campaigns, they might want to just take an old asset and change the background, put the asset on if it's say shoes or clothes, put that on a different model and do all that within an AI world instead of having to go and get a studio together and pull in models and do all that kind of stuff. So make that content workable in whatever way you need without spending the kind of money you have to spend sometimes on photo shoots and taking the time you have to do these shoots and all that kind of thing as well. So it's kind of making all that possible. It is kind of what I was saying about what was not possible before is now very doable and scalable. And so I think all we are doing is facilitating that piece to make people able to do that, if that makes sense. So many use cases, there's so many different things that we see. So you just try and make sure that the ability is there within your world to help them to do that piece.

 

Lauren Livak Gilbert (10:39):

And Cliff, we know that examples always bring all of this to life. So you have a bunch of examples to get into. Let's start with you worked with a tea company to implement ai. Can you used it for and how it affected their imagery?

 

Cliff Crosbie (10:53):

Well, it is a big global tea company and there's always discussions going on there, so I'll keep that confidential at the moment, but large company gone for over a hundred years, way over a hundred years. So one of the big companies and there was questions about could you even identify tea by training an AI to even look at let's say just a pile of tea leaves and be able to identify what these tea leaves are instantly for them or other things about looking at heritage imagery and being able to say where this is and what it is. And that's the beauty of agents. We've developed four agents and one of them we call enrichment. And all we mean by enrichment is when you have assets working to enrich that asset with whatever information is going to be useful to you as a brand. And of course if you've got a pile of pictures of tea leaves, useful would be to know exactly what that tea leaf is and what it does and what its properties are and all that kind of thing.

 

(12:04):

So with that, we're able to take an agent and just to give an example, be able to write the prompt that is saying you as an agent or you as an AI are an expert in this industry, you're an expert for that tea company and all about the heritage and what's going on with these tea company. And we can refer it to the website of that company, we can refer it to just online. It will go and look at resources so that it behaves as an expert and we want to know certain things, tell us about this. T tell us about the aroma of this T, tell us where it came from, what it is, what it does, what its properties are, all this kind of thing. And do that from this image and show it that simple. And I'm simplifying dramatically here, but look at that image and tell me what the T is as though you were a human expert in T who knew exactly what this is and that is exactly what an AI does.

 

(13:01):

And they were amazed and we were kind of amazed when we wrote it that oh my god, this really does work. But the ability to just give it that image and do it. And that sets you on this course with prompting and with AI agents that you realize that when you say enrichment, it can be anything. You've suddenly created this T expert that can look at T and say, oh yeah, this is lapsing tea from this part of China and blah blah blah and just goes on and tells you all about it. But that is quite a skill to put into that kind of business and to be able to do this, just to put this in perspective, that means you can create that image, that agent, you can run it against, I dunno, 10,000 images of tea leafs and then you can just say, update all these images and within seconds, all my images of tea leafs have been enriched with all the knowledge of what this tea is, what it does for you, I can say to create alt text to go with that so it's searchable and then give me that in all these languages as well and add all that to the image as well.

 

(14:15):

And suddenly I've got these assets that have every single piece of information I could possibly want about the tea leafs or whatever the subject might be. And it's all searchable and it's all there on the image. And of course alt text in parts of Europe now is becoming the law that has to be there for people with disabilities that want to know what is in an image. And you have to be able to do that in multiple languages as well. So alt text is something that a lot of companies are putting a lot of time and energy into with copywriters and translators, but we could do 10,000 images in five seconds or maybe less and done. So it's that ability to do things like that and we're referring to a T brand, but this applies to anything you can make an agent be an expert in any field you wish it to be an expert in and make it enrich assets through that. So it's a super, super powerful tool.

 

Lauren Livak Gilbert (15:19):

And that all text also feeds answer engine optimization, which I think is a really great call out. So if we think about a EO or GEO or insert acronym here, the industry hasn't decided that agents are scraping alt texts and looking at imagery on text. So that's really impactful. But the thing that really struck me, cliff when you were talking is if I'm a large organization, I probably have assets that I've created for the past how many years, and they may not have been tagged with alt texts, they might not have metadata and you might have images that you could use that you spent a lot of money on, but that live in your dam or any other system that you've never been able to find before. So this really enables organizations to get visibility to the images that they can use for their PDPs, for their brand websites. I mean, I see a cost reduction opportunity here, right, because you have visibility to all of the content that's created. I used to call it creative stretch where you now know one image exists that you can use on multiple channels. So it's just such a big opportunity I think for brands to have access to all of this imagery that they could probably never have found before this.

 

Cliff Crosbie (16:29):

Exactly. Yeah, and I mean that's exactly it. And we have found that the agent we're talking about that enrichment and certainly that alt text use case is the one that's come forward in the last few months has been one of the biggest use cases. And a job that you say is something that probably no human really enjoys doing alt text, I can't imagine. No thank

 

Lauren Livak Gilbert (16:56):

You. No

 

Cliff Crosbie (16:57):

Human enjoys translating alt text either. So when you can automate jobs like that and do it really at scale, and I think that's where we've seen the biggest interest for ours. I can do a hundred images or whatever, we can do thousands of images at once, and you never take the human out the loop when that old text comes through, you look at it and you can approve that if you want. I mean, in reality, what people will do is look at several and say, yeah, this looks great. And then I'll select all, and I think we've talked before about this that does AI make mistakes? And yes, everything, there are certain mistakes from it. It is a tool. But if you had a human doing 10,000 images and an ai, 10,000 images, I'm pretty damn sure where the better result is going to come from and more of the consistency rather than the human. There'll be humans who are great at it and humans not so great at it, but an AI will consistently do it in the same way right across. And I think that's the difference. But then you can have still the human who takes a look and just make sure that what I'm going to accept on my alt text in with the AI is in good shape to use.

 

Peter Crosby (18:19):

I love that you brought the humans into it because one thing I was interested in hearing about this and also your earlier comments about instead of using photographers and setting up photo shoots, and I started to think of the changes in people's careers that are happening because of this stuff. And I'm wondering what are the human reactions you're getting from people? And I'm sure it's a range of things, but in this case I'm hearing, oh my God, this work would be impossible to do under any circumstances. It's boring, I hate it beret, but I'm wondering what's the range of reactions that you're getting to how this is changing their

 

Cliff Crosbie (19:01):

Theme? Our aim going into this was that we're freeing up the humans to be creative. And I think, dunno if it's a misunderstanding of this, but we hear a lot that this has taken away work, but it's taken away a certain type of work. And even if you are a photographer and you want to create a photograph and we've got a new AI that we're just launching right now that is in testing and will come out soon where you literally, you can take for example, I dunno, a headphone manufacturer and I have a set of headphones in an image, but I want that set of headphones in and I've just got a set of headphones with a white background for example. But I want to see these headphones on a, I dunno, 25-year-old girl playing music in her room in a certain whatever kind of place I want to put that in and I want that image created and I can prompt the AI to create that whole image now from scratch and then take away whatever and repro it to get to where I want to be or anything.

 

(20:20):

I'm bouton and I want to see that pair of shoes on a model in Paris on a catwalk or whatever. So I can prompt it to do that and I can use images as input to get to where I want to be. So I still need a very skillful person to put that all together that I think as we've talked before, the AI is not creative. The AI is following you tell it to do, and so it's pulling from your instruction. The creativity comes in my ability to tell it exactly what I want, the same as my ability to set up a scene on a camera and take that picture. So you need that skill level to be able to do this. And so I think jobs are not going to go away, it's just things are going to change because there now needs to be a person who's a prompt engineer, whatever we're going to call it, visual engineer or whatever, who can prompt in that way. And before it might have been a cameraman and it probably still will be in many cases, but it's a different set of skills and a slightly different set of jobs, but the same skills it takes to take that image are the skills it takes to prompt to get that image in the right way as well. So yeah, it is changing. So I hope it's always difficult to describe

 

Peter Crosby (21:46):

This. No, and we were talking about this just before we went live, this is always changing and that's terrifying and fascinating at the same time. And so let's jump into another example from T to automotive, not as big a leap as you can get. Tell us about that use case.

 

Cliff Crosbie (22:07):

Well, that's what I love about my job in four and a half thousand companies, we cover every single industry, every single product. So it's kind of one minute you are doing cars in the next minute you're doing T, which is kind of fun. But this was the ability to, this company again is a long history of car manufacturer and it was about the ability to identify models. And when I mean not just looking at a picture and saying that is the whatever model, but suddenly there's a picture of a dashboard or a picture of a seat or a picture of a view into the boot or under the bonnet or hood of the car and can you identify the car from these? And the answer is yes. When you again train the model as an expert in that brand of car, it can identify everything from everywhere, even down to small details in maybe taillights that you need for Europe versus taillights you need for America or for another market.

 

(23:12):

So even these market details, and of course even a human would have to really what we found is that even where a human was looking at that image, they would be referring to, oh my god, is this the right one? And the time it would take with one image, and I actually remember this from when I worked at Apple and you would have images of iPhones, but was that the iPhone 15 or the 14 or the 13? And when you were putting out into the market, and especially products like Apple as well, very subtle differences between some products and an AI that's able to identify these small details and identify that product is so much more efficient than unfortunately us as humans are to be able to do that. So being able to just quickly say, here's a thousand images of these cars, do all this to identify every model and exactly what model it is in year of manufacture and country of distribution is just huge.

 

(24:18):

And yes, maybe at the time some people can take these, but you know what happens, A company does a photo shoot and not every image is labeled or as you said, Lauren, older assets that I need to get labeled up. So I don't know, you can't even compare the value. If I put some humans in, it would take, I dunno, six months to go through that project and I can throw it in an AI and it takes me six hours, six minutes, whatever to do that. It is huge advantage. So again, that's that enrichment model. Although saying that with the other one, we've touched on this kind of visual models as well, but it's that ability to look into images and I mean, I saw images from this manufacturers of cars under a tree away in the distance on a site and I'm thinking, God, we need a magnifying glass of something to try and see what this car is. And it's just going, yep, that's that model. And that was the year. And it's fascinating stuff.

 

Lauren Livak Gilbert (25:23):

The way I look at that too is then it's faster for you to create seasonal content or to update your content where in the past, to your point at scale, we could not do that because we were just met data tagging the ones that were from the photo shoot from last week and that would've taken months. So it can also accelerate a lot of that personalization, which I think is exciting.

 

Cliff Crosbie (25:43):

You don't need to do any of that tagging at all now because we can automate what I've just said. The moment you upload that set of images, it's all done. So you can automate these processes to make sure any new assets coming into your dam are just automatically tagged like this. You just don't have to do that job anymore. You can just entirely leave it to ai.

 

Peter Crosby (26:10):

And automotive specifically is such a cool example because it's really one of the most, if not the most complex industry in terms of this particular part was only in this model year. And so that complexity and being able to do that, I remember going into car parts stores and they'd have these enormous books where they'd sort of try to page through it and find the exact, it was just a nightmare. I couldn't imagine how they managed that and now I just snapped if that didn't make it into my mic. In an

 

Lauren Livak Gilbert (26:48):

Instant.

 

Peter Crosby (26:48):

Yes, in an instant, I'm very interested. Cliff, in the backend of your system of record must be tapping into so many new sources to be able to, or if not new, at least sort of really expanded integrations and things like that to be able to give the AI what it needs to do this. Can you talk a little bit about how that's

 

Cliff Crosbie (27:15):

Yeah, I mean it's kind of the secret source of the company, but it is, we are using, I think currently, and don't hold me to this a hundred percent, but I think we're using at least seven or eight models right now, integrated some of them, Amazon, some other models that we use in there. I think all of these things, it's how you use these models and how you put it all together. And we have little bits on our image. Similarity is more a bespoke type software, but putting together these models to work is the key thing to it. Now what we are trying to do is take away that pain for the customer. So we will update to, we are just updating to a newer version of one of the text models right now because here's a new one that works even better. So for the customer, they should see a seamless experience from their end that should just improve over time as we update these models.

 

(28:19):

So we have a team that are just looking at what is the latest and greatest and what's delivering, and then we test that and if it's better equal or better or has better potential than the model we use, we then bring that one in and move on to the latest model. So we're trying to make it from a customer point of view, you're going to get all that functionality and great ai, but you're not going to have to be the one that worries about, oh, is there a better one now? Is there another one? Is there a newer one? We are trying to do that out the box as you're using our software. So it does mean a lot of work at our end. And as I say, something like seven or eight models, and I know there's a new one with this image generation one that we're doing right now that allows you to generate from all sorts of content. But yeah, super exciting. But yeah, yeah, you're right, pier, there's no one model does all this. You have to patch it all together to make it work.

 

Lauren Livak Gilbert (29:16):

And Cliff, you talked a lot about prompting and if people are listening, a lot of them are probably saying, well, I'm not a prompt engineer, I'm not an expert. And we've talked about this before and I really like your perspective on this. What is your advice for any brand or retailer that's working with prompts or working with agents? How should they be thinking through this?

 

Cliff Crosbie (29:36):

Yeah, well, I mean, prompt engineering is becoming almost like a job. I think it probably is the new job where you have to define what you're trying to do, you have to draft up initial prompts, you have to value how it works, you have to refine and all this kind of thing. So there's a lot of work in the prompting, but I think when we were speaking before, my take on it now from putting on my old marketeer hat is think of it like briefing an agency, but the agency now is your agent. And what you're doing there is putting together a brief of what you want to do. And as we all know, as marketeers or Rex marketeers, the quality of that brief determines the quality of the output that you get.

 

(30:29):

And you know, have a back and forward with an agency on a brief and you have a back and forward with the agent, no, you didn't get it right this time. No, what can I do? What can I change? So I feel it's like that. And I often talk about telling it, like talking to a child. This is not an intelligence if you like, it's got this ability to draw from everywhere to get what it needs to do, but it doesn't know until you tell it. So you briefing it and putting the right prompt in is absolutely key or you won't get the right result. So that's that creativity point from a human. You have to tell it exactly what to do. It's not thinking on its own and thinking, oh, I know what Cliff wants, I'll just go and do that. It won't get that idea.

 

(31:21):

It really needs you to write it very clearly and define what you want. And of course that is the difficult but also enjoyable part when it gets it right, which it does very often now. And of course the new way of prompting now, you don't need to be a coder or anything at all. This is writing in plain English, so it's as good as your language skills are, I would say, to be able to write the prompt and be very specific about what you want. But yeah, it kind of makes you think differently because it makes you think clearly about what you're trying to get out.

 

Peter Crosby (32:01):

I was at a conference recently and a Forrester analyst referred to AI as really a psychopath. It just does the next thing that comes through. Its mind based on what's happening. Exactly. It has no sense of really morality, Phil, no emotion. No, it's just doing its thing. And so really prompting is controlling the psychopath. It's like putting the humanity, putting the specificity, putting the guardrails that make AI do what it does best and the other. So I love the idea of it as a creative brief. It feels like computer programming to me sometimes, and it isn't so that I think I'm imagining Binder is investing in trying to educate because it's all about upskilling. Lauren talks about this a ton that HR really needs to be involved in all of this to make sure, and I'm wondering what a binder and then also what you see your customers doing to upskill their people in these areas.

 

Cliff Crosbie (33:06):

I think you can know if you go in places like Coursera or all these kind of educational sites, there's just a mass now and you can do them through Google Gemini, and there's all sorts of courses and YouTube is absolutely full of them now. So I think everything now, the trick is what should I read and what should I not read about this? And it kind of depends again, back on what you really want to do inside Binder. Yeah, I mean our HR team put together amazing courses to help us with this. So we've got a whole education piece inside the company, but I see that's happening everywhere in companies. The very few companies we talked to now don't have someone who's thinking about AI or is responsible for AI or has a whole team and has a budget is more than Norm I would say now.

 

(34:03):

So that is all there. And prompting is something that's very much on people's minds. And we have this part, I haven't talked about it yet on here, but as we went into, I think it was a year and a half ago, we created a team called Binder Labs. So we have a team inside, and that was where all the AI agents came from, binder Labs. And that team, which I'm part of, works with customers to see what their needs are and what we can do. Because really what we are looking for is true use cases, things that companies really need because creating AI for the sake of creating AI is crazy. We are creating AI for return on investment faster, more efficient, more productive, all these kinds of things. So to do that, you have to work closely with your customers. So we see inside that, but as we build all these prompts, we of course are building a prompt library within that and we're giving customers access to that library as well. And if a customer says, I need a prompt to do this, yeah, we will jump on it and we'll work at creating that with them or for them. So that really is going to be the big thing nowadays because you want them to get the best result and not everyone is there yet when it comes to prompt engineering. So we've spent a lot of time on it, so we provide that help as well.

 

Peter Crosby (35:35):

That's great. I mean that's always been a hallmark of this industry is just we're all trying to figure it out. Nobody has all the answers and you just try and put it out there and together, come up with, I mean that's sort of been the whole reason of being the

 

Lauren Livak Gilbert (35:50):

Dsi. The

 

Peter Crosby (35:51):

Dsi. Exactly. Yeah. I mean it's kind of where that idea came from is just the hunger of everyone to try and figure it out together. So to close out Cliff, we try to be a no hype zone as much as we can be. And there's a lot of people out there that are screaming that everything's on fire and oh my God, you're getting left behind. Then there's also the reality of how quickly this stuff is moving. I'm wondering, when you talk to customers, how do you sort of express what the urgency of this shift is? This kind of strategic shift really, and how you think about your imagery and the power that it has in this new environment and on the new agent shelf and then how your organizations need to transform to do that. What's your timeline in your head of if you were running the shop where you would want to be and get your company to,

 

Cliff Crosbie (36:52):

It's how long is a piece of string? I think you have to get into this really urgently is our message because it's, like I said before, we've got four and a half thousand. You really big brands work with it. So I think you just go on the website, you'll find who we work with and you can guarantee that all these bigger names that you see on there are already into this. And if you are in competition with any of these guys, boy you better get moving because the efficiencies are huge. It's like I said about when you're talking about a brand who would normally take six months to do something and is taking a very short period of time. I know there was a big liquor company that we work with, a big global, global brand, and they had such a backlog of images that they really needed.

 

(37:52):

They had all kinds of problems with these issues and they had seen that the work would take them something like six months to do. It really was six months plus project. In fact, they had even boarded line on saying it was impossible because they had hundreds and hundreds of thousands of assets. And with some of the software we've talked about here, they had an intern do it in nine days, everything was done and how can you put the cost saving there and the time saving and then the fact that it was all done and everything was running more efficiently within that time. So save six of inefficiency with assets that need sorted. And it's kind of priceless on some of these things to get this done. My advice to a lot of people is this isn't going away. This is only going to get bigger and better. And do you want to be that kind of pioneer and get ahead of the curve here or do you want to just wait until everyone in your industry is already on it and it might be 100% reliable by then if anything ever is and then you'll be lost.

 

Lauren Livak Gilbert (39:15):

I remember it takes so much time to change. I just think that, remember, think about e-commerce in general, right? It's been around for 31 years and there's a lot of organizations that still don't have their digital transformation completed. Just start learning and testing now because it's moving so much faster than any other transformation we've ever seen.

 

Cliff Crosbie (39:34):

Yeah. Yeah, I agree.

 

Peter Crosby (39:37):

Besides that, it's kind of fun. I got to say I've seen that it's, I'm going to put just my salsify half of my brain, I'm just going to light it up for a second, which is similar to what you were talking about. Salsify put out something called intelligence suite. It injects AI into existing workflows. And one of our early customers came back at the end of last year having tried it with a couple of things and came back with 50 more use cases for what they wanted to do to pick up the advantage here. And that's not a plug necessarily, but just how much, just trying it can unlock the possibility of it. And so the best thing to do is to get that if you haven't yet already, get that test and learn together. Choose your outcome, not your process, but what's your outcome and drive towards it. And it will unlock a lot for everyone, both the company and the individuals involved in where they're taking their career.

 

Cliff Crosbie (40:40):

And it is the human, I mean, on another manufacturer of big equipment I was dealing with, they had first come on saying, would it be able to recognize this forklift truck? And we're like, well, yeah. And we put it in and it can recognize the forklift truck. And by the time we got down to, oh my God, it can recognize the individual parts of the forklift truck and tell us this bearing and everything else. And of course they kept asking questions that if you had been sitting here six months ago, a year ago, you wouldn't have even thought this was possible. But of course, the moment you give a human something, we can identify the truck. Yeah. Can you identify the wheels? Yeah, we can. Can you identify the engine parts? Yeah. And go all this way down. We just naturally want it to do more and more and more.

 

(41:28):

So of course we are part of the push and people keep asking for something they wouldn't even dreamed of asking for 12 months ago. And of course we can do it. And I think that's what pushes it, because there's a constant need for people to say, oh my God, we can do all of this. And I think one of the things I remember when I worked at Ikea and I had the privilege of working with Invar, kra, IKEA on many occasions, and one of the things he used to say was, and maybe this is part of why IKEA is so successful as well, was he was always

the thing that if you get your idea in place and you're 70% there or something around that, just go for it because that last 30% takes you so long. Fix that while you're moving. But if the 70% feels good and in the right direction, just do it.

 

(42:21):

And then of course you'll fix the rest as you go and you'll adjust. But don't wait until you get near to a hundred percent or wait until everything's perfect before you go. The world doesn't move like that these days. And he was saying there's nothing to do with ai. I mean, this guy didn't even carry a mobile phone. He was just saying this about business. And I think, but that's still relevant. That's what AI is, it's business. And I think that philosophy applies to this as much as it did to everything else. For companies, it's get going before you're the last one doing it,

 

Peter Crosby (42:57):

Jess. Go for it. I love it. Great way to close. Cliff, thank you so much for sharing these stories and your point of view from us in a really fast changing and exciting piece of commerce. We really appreciate it.

 

Cliff Crosbie (43:10):

Great. I've enjoyed it. It's been great to meet you, boss.

 

Peter Crosby (43:13):

Thanks to Cliff for sharing his wisdom with us. Just go for it. Speaking of which, just go and sign up right now for the Digital Shelf Summit, where all the best conversations in commerce will be happening in May. Digitalshelfsummit.com. Thanks for being part of our community.