One in (Seven) Million: Linking AI with Design To Get Products To Market Faster and More Easily

Alarice Rajagopal: Good afternoon everyone and welcome to our webinar: “Accelerating The Package Design Process With AI.” Please note that while the materials in the webcast have been reviewed by our editorial staff, the views of the speakers and the organizations are their own and do not necessarily reflect the opinions of CGT or its owner EnsembleIQ.

From new product innovations to new materials and rebrand, product change can create a need for new packaging, technical and creative information collides with file formats whenever packaging is involved and packaging designers are called on to localize designs for multiple markets and change specific elements, creating many iterations of the same basic package design.

That's a lot of moving parts to manage and coordinate. That is why we're going to explore how product asset management combined with artificial intelligence and machine learning models can help by accelerating ideas to market for consumer goods companies. To help us discuss this further, I am delighted to introduce our subject matter experts on this topic.

Our first speaker today will be Cheryl Perkins, founder and president of Innovationedge. Cheryl is a thought leader in innovation and the creative catalyst in brand building initiatives for companies looking for that innovative edge with over 20 years experience directing growth and innovation. Cheryl most recently served as the senior vice president and chief innovation officer for Kimberly Clark. She ran the company's innovation and enterprise growth organizations including R&D, engineering, design, new business, global strategic alliances, environment, safety and regulatory affairs, and oversaw innovation processes, systems and tools. Cheryl brings her expertise to Innovationedge, creating new strategic business opportunities for companies seeking a competitive advantage across key businesses to ensure longterm growth and deliver a continuum of sustainable innovative solutions. In 2006 Businessweek magazine chose Cheryl as one of the top 25 champions of innovation in the world.

Cheryl will be joined today by Alan Porter. Alan is an industry leading content strategist and Nuxeo's director of product marketing, he is the author of The Content Pool and a regular conference speaker, workshop leader, and writer on content marketing, content strategy, customer experience, brand management and content and localization strategy. Named one of the top 25 content strategy influencers and a digital strategy thought leader in 2016 and 2017, Alan is a catalyst for change with a strong track record in developing new ideas, embracing emerging technologies and improvements. When he's not at work Alan makes even more creative content like comic books and novels, including the official Cars graphic novels for Pixar.

So as you can see, we have two very creative and innovative experts on hand today for this discussion. So Cheryl and Alan, thank you both for joining us. I'd like to go ahead and hand things over to Cheryl to get started.

Cheryl Perkins: Thank you, Alarice. I'm really pleased to be here with you and Alan today, to talk about two of my favorite topics, the power of AI coupled with packaging design. And what we'll do today is talk about some leading trends that we're seeing in packaging. I'll talk about a case study that taps those trends and really helps us understand the co-partnering that's going on between AI and design. And then turn it over to Alan and he's going to talk specifics of how AI can help accelerate that whole package design process.

So those of you out there that are designers or work with designers know that there's never been a more exciting time outside of the challenges we have right now with COVID. But packaging is playing such a critical role in helping to create the experience that our customers and end users see. And we see trends in color, form, all these elements coming together to give us some unexpected product packaging and AI playing a huge role with that.

And it's helping brands really break through the clutter of the competition. And that's what I want to talk about the brief time we are here together. So what I'm going to do is hit three trends. These are trends that we've been tracking for a while and we really feel that again pre-COVID, they've been playing a role. And I'll talk a little bit about what might happen post-COVID.

So the first trend here is in this e-commerce space. And e-commerce is really hard because we don't have the luxury of touching or feeling that package in real life. And so you have to leverage whatever we can to really tell the story. And what's really interesting is knowing the trend was subscription boxes and there's more and more out there. I want to use the example of the Cause Box, but those of you that don't know about the Cause Box, it's a quarterly subscription and it really features brands and items dedicated to doing good all over the world.

And many of the featured brands inside the box donate a portion of their proceeds to charity, and they even employ local artisan type creativity and crafts, which is really excellent. But you can imagine this box, this branded box, how can they get their message across? How can they in a single box knowing the box is filled with many different brands and many different products, tell their story? How can they communicate their key message and still build a strong connection to make these linkages with their ideal consumers?

So it does start with that outside packaging, that primary packaging, it's the start of that journey or experience, and they've done an excellent job. If you have time, you should go out and look how every season for $54.95, a box arrived full of these different brands and product opportunities. But the box itself symbolizes and creates unity among everything within that package. And so the role of branded packaging, in e-commerce will continue to be something that we're going to keep our eye on. And it's something that all of you should as well.

The second area is another exciting area we call metamorphosis. And basically what we have is where the company sends multiple messages by the imagery that they use on that outside package. So first off on the left, there's a Chinese company that has won many awards. The Pentawards gold for 2019, the iF DESIGN award in 2019 and the Red Dot award in 2018. It's a mineral water product company and they use these saline mountains as the place where the water sources, but the designs if you have a chance to look at them is that mountain morphed into different shapes and figures, so again, metamorphosis at its best. The one on the right is a really cool example of a fruit and jam company. This company, again, Beak Pick, they know that there's a challenge out there with consumption of sugary jams and the hurdle to healthy lifestyle.

So they know birds consume small amounts of food and they use this to create these images. These images are an infusion of the bird and the fruit, in this natural modern packaging that houses their jams and fruit preservative line. Now you have to imagine how hard it was to find the right bird with the right fruit, a pear, an apricot, in line. But they have amazing anatomical specificity for each fruit. And what you see then is this metamorphosis where the fruit being the head then takes the shape of a body and so each of them in their daily fruit tree is really representing, it's just a small amount to provide that experience, and so it's very interesting because we've heard consumers say that even after they've completed using the jams or jellies, they like leaving these out, showing their style or the things that they liked as part of their lifestyle.

The next trend, and this will be the one that we have to really watch post-COVID, is on eco-friendly packaging. We all know there's been more and more brands starting to explore using eco-friendly packaging. These materials again are trying to easily be recyclable, minimize the amount of material necessary for the package design, and even for going packaging altogether.

Now the question is, based on again, our behaviors and attitudes post-COVID, what will happen with this, but this is still an area we think it's important and we will continue to keep our eye on it. So those are some trends that are really influencing the design area for packaging.

Now what I'd like to do to set the stage for Alan is talk about AI and design. And I put that in where they're very specifically, this isn't AI being leveraged by design, with design, it's in. And the reason for that is that we have seen again the last 10 years, the Internet of Things moving forward into the intelligence of things.

If any of you attended the Consumer Electronics Show in Vegas in January, you saw how important AI and the role it plays. It's here to stay. It's very exciting to think about it, but it is a true partnership between AI and design, and that's what you'll see. We're starting to see the emergence of co-creation.

It's important to note that we don't ever think AI is going to replace designers, but more importantly we're going to see co-creation opportunities that will open up all these new possibilities. And I'll actually share a case study with you. What we do is think about the designer who then chooses the parameters that they want to operate within. And this smart AI system that you'll see as Alan takes you through this, will generate a series of design alternatives to give back to the designer. The designer still is the decision maker.

We don't see that role moving away, but can you imagine the system generating millions of different opportunities, co-creating directly with the designer what those opportunities are. Then the designer can choose the best design according to his or her preference. So it's amazing to think about the reduction in cycle time to get hundreds or even millions, which I'll show you shortly of design opportunities to choose from.

And the reason we're talking about this being co-creation is if the system, the smart system, and the designer, being in the crosshairs of art, science, engineering, and design. And that's pretty exciting. And so we're really pleased to be thinking about this, and we really believe that these algorithms will extract the colors and patterns of the design, and create thousands of variants within a specified range. So the designer still set some boundaries, be it the color, be it the pattern, and then allow the algorithms to do what it can do to create the potential.

So a designer now can have many, many opportunities for possibilities at his or her fingertips. And that's why we're saying it's AI and design, it actually is a co-creation. And I want to just give you a case study, and this was in Italy a while back. I don't know how many of you've had the hazelnut spread Nutella, but it's delicious if you haven't. The manufacturer Ferrero, partnered with Ogilvy & Mather in Italy. And they really wanted to define designs that would represent, again, be expressive to the Italian people.

They wanted to create a piece of art. This glass jar with this amazing packaging that they would leave it out and it would symbolize something to them about the experience they had when they actually used the Nutella. And so there were thousands of color combinations, but they all had to fit with the brand, look and feel, for hazelnut spread, for Nutella.

And what happened through the AI work that was done, again, a partnering between the designer and AI, they created over 7 million one-of-a-kind jars. These designs sold out in 30 days. They were over 3 million different posts about these designs, people showing off this design, and it reflects me in this way. People placing it in a prominent place within their home. There were 10,000 videos and very positive.

So again, imagine an individual designer trying to create 7 million different designs; it would have taken a very long time. But again, they were able to charge a pretty good price, have a margin, and have 7 million things sell out within 30 days. So it just goes to emphasize that the possibilities are endless, and as we learn more and more, and machines and computers start developing these human-like capabilities to complement our own, we're fundamentally going to see a shift in the relationship between our tools, and with the design process.

And this is the first sign that we're seeing in terms of accelerating that process. And what we're going to see is that humans become more like mentors. We're actually coaching our tools, and guiding them through tasks that we want to accomplish. So it's a much different relationship, because we're a mentor or a coach, especially when you have tools like deep reinforcement learning to reinforce the principles and create these efficient algorithms.

And this isn't just happening in packaging, we're seeing in robotics, video games, the world of finance, healthcare and transportation. So you're going to see more and more of this and that's why I'm excited for you to hear from Alan in terms of how this actually can happen.

The one last thing I suggest, if you want to think about the world of possibilities, you should go out to the TED Talk by Michael Hans Meyer. It's incredible what he's done. It's been inspired by biology and cell division, but he's been able to write these incredible algorithms that creates these fascinating shapes with variants of facets on them. And he's in Zurich and utilizes this to create complex structures.

And I think the reason I wanted to end with this is that the possibilities are limitless. It's unimaginable these shapes that you'll see, and you see this self-experience coming out. And again, it's just the world of possibilities is endless. And with that, I'd like to turn it over with Alan because he's going to take you now through how this can actually decrease the overall cycle time.

Alan Porter: Thank you Cheryl. That's great introduction and some great insight into some of the trends and some practical use cases there. So I'm going to talk around the idea of reducing time to market. How would you do that? How AI can play into that.

So one of the things that we really find and Cheryl touched on, particularly in the digital world, is the idea that images and video and rich media is becoming more and more important in the customer experience and actually in some cases it's exceeding the physical handling of the product and the packaging.

And we're finding that as we're talking to more and more companies, the idea of really putting the content first and foremost is having an impact not just in marketing but throughout the whole digital supply chain. From the idea, of the product's idea, from really the ideation stage all the way through to design. And then actually design the packaging, design the product design through to the customer experience.

We call it product asset management lifecycle, defining a really efficient digital supply chain. And this is an area that the growth in AI can really help speed this process. I want to lead in with that with a quote that actually comes from the cosmetics industry, but I think it applies equally as well in CPG in general and packaging specifically.

And probably most of us, or in our current situation, people do more and more exchanges on online. That's really the trustworthy brand of those that will be the obvious when it's uncertain times, are those aware then they are really putting themselves forward. And the people who do provide I think will be rewarded.

And so we're like customers. I mentioned the product life cycle. At next year, working with our clients and based on our experience, we've really seen seven touch points in the product life cycle where we believe connecting the content with the product information and using technologies like AI can really provide the most value and speed up the plan for delivery.

And those were around things like building materials to design, actually designing the product, taking imagery of the products, doing the campaign development. Obviously when it comes to actually putting the products out there, the packaging design is key and central.

Even the people at the point of sale be that actually a digital website or even in a physical store, giving them access to the content and the product knowledge that's being generated throughout the product life cycle. So we believe these seven touch points really can help with understanding information about the products and where the AI fits in and how does that really come together. It can be understanding and using technology to get access to things like suppliers and pricing and availability, even down to imagery of materials watches and design.

It can be pulling together the product specifications and building 3D virtual prototypes of packaging and labeling, linking that with the photo studio and the models and the photographers and the imagery coming out of there. Then where they can actually build campaigns around it, understanding the product that they're trying to sell, packaging, taking certificates using images, taking some of the images, doing also cropping, labeling, localizing it, and then understanding where that content is being generated across the organization and feeding it to the people actually on sale point.

So they have on sale the right SKUs, the right products, options. So really pulling all this content and data and information together is key. But that's a lot of information to handle. And really that's where things like AI can help scale.

What we're finding particularly with folks working in design is some of the trends that we're finding with our customers, is that more and more they're looking to use turn key components in the packaging, be that physical packaging or be that asset such as imagery or video and they want to go see what's been used before. They want to reuse it, they want to create derivations of it. So they're really looking to find out how they can use it and what's been done before. Often just finding it in an organization can be quite difficult.

So again, using AI to add metadata around information can be really useful. They want to be able to preview and access customized designs throughout the product life cycle. They want to localize things to different markets. I love the Nutella thing where maybe that you want to bring it down to personalizing an asset or personalizing a product design.

And what we've found with most of our clients is that a lot of the processes are really hampered by very complex workflows that really need multiple levels of approval and so forth. So just to give an idea as it was mentioned at the beginning, the new planning, new products, gradient changes, rebranding, really every product changes is causing a need for new packaging, and bringing that together with the technical and creative information, different file formats, packaging design, localization, multiple approvals, it can really take quite a while.

So here's an example from a cosmetics company again, on that packaging where really say we're finding that, all that moving parts that they were having to deal with and the approval process was actually costing a lot of business lives. Things were going around, the approval cycle. The packaging design wasn't kept up to date with the changes that we're counting, so they were having to repeat things over and over again.

By applying a more product asset management streamlined process and using things like AI and plugging it into a more global view of the overall product value life cycle actually managed to reduce the approval process from weeks down to days, by just thinking about how it fits in, how the approval process fits in, what's needed to actually drive the packaging, what assets are needed upstream of that.

And then also how they could use AI and asset management and connecting content and the data together give to people in the approval process the information they needed at the time they need it.

And one of the areas that keeps coming up over and over again is really thinking about derivatives. If you're going to create 7 million derivatives of a basic design, or you're going to create five or six derivatives of a basic design, be that for a different market or it can be to different physical product is going on or maybe needs to meet different legal requirements in different markets. It’s actually keeping track of that.

So again, building the information and the content around the assets used to correct packaging can be a really complex issue. So if you'd go to the way of actually linking those derivatives to a parent asset, that can be really helpful and really help reduce waste because you're not going off and recreating stuff that's already been created or something needs to be changed, you know how that change will be reflected in all the derivatives downstream.

So again just keeping this flow of information across the product life cycle from upstream through design and down to downstream changes and making sure that the impacts of any changes are actually minimum.

One of the lessons that we've really learned is that by applying machine learning for developing the metadata models of around the design assets, around video, it can really help drive a more accurate search experience as well as add value to the images and videos and other assets to use within the design process.

So when we're talking about AI and machine learning, really, and I'd say we consider there are really two types of AI or artificial intelligence and we consider machine learning as being a subset of AI. And what we're really talking about here is the idea of using machine learning where we're running algorithms to understand the information, understand the data that we're working with.

And there is two basic types, generics of where you're really connected to a broad set of public services for business specific which are around what you actually do within your individual organization, that you're actually finding and then reaching to specific business use cases or specific taxonomies or specific domains within your organization.

And like taking an AI practice approach, you can develop three ways, you can look at how you actually want to enrich your information, and use that information to decide to start making decisions based on AI, based on error detection and things like that. And then you can actually use it to help find places that you can optimize and improve your processes. So when we talk about generic AI, we're really talking about plugging into things like recognitions for amateur video, transcribing, extracting text from videos, comprehension, maybe symbols, micro translation. We think that this is really as a first layer of AI. You can plug into generic models that don't need to be trained.

They are based on a larger focus. They would pick out that this is a bird, but they wouldn't necessarily pick out what breed of bird it is that you want to use on your packaging. Or it would pick out that this is a red truck, it wouldn't necessarily let you know which make of red truck it was or things like that. So those are really a good first leg to start with, they can really help you enrich and get some of the basics.

If you need for your content and for the design, files and so forth that you want to run through it. But what we think real value comes is when you get to business specific AI or ML. This can really to the point that Cheryl was making help with reinforcing, particularly where you've got something that will actually give continuous trainings.

One of the things we do with our product, the next year we can make sure the content bots are continuously evolving, that are continuously learning. They're learning each time you run a model it learns from it and apply those learnings to the next run of the model and so forth. If you can take a look at what models are returning what results — it's not something you can do with the generic system but with your own models that you've built for business specific machine learning, you can then pull back if you need to. And can look at performance monitoring of what things have been very quickly identified and you can track all your machine generated content and the metadata and make connections.

We found that the training of business-specific machine learning really takes very few objects to actually get really high results. And generally when we're looking at a human being applying information, or adding metadata or going in and looking at things like imagery and trying to identify what's in an image and so forth, a typical human tends to get around an 80% success rate. And you can get up to 90%, 95% and above with machine learning pretty quickly.

So generally they are more efficient than a human actually identifying what things that are in an image that you want to use that you want to pull out. Just to give an example of that, here's a use case where we're looking at a couple of images. One, more of a lifestyle image of a product being in use and one of the actual packaging itself and product image.

You may want to be using AI. Is this a single product shot, multi-product shot? Is it a shot of the front, the back of the products and the packaging, where is this being used, what's the geography, the location, then maybe even seasonable. And one of the things that we're finding really useful in CPG packaging retail environment is the ability of AI to relate to connected, related products. So start to understand that this may be a shot of one particular product, but this could be something else in the background.

Maybe if it is something like packaging for a cell phone, it would actually also link to the fact, recognize the fact that there's a cell phone case in there and know what a cell phone case SKU is or the design around that. So we're really finding that these things, if you think about the bird example, how long would you need to train somebody to recognize the right bird to go with the right fruit? But once you've actually got that relationship set up and that's so in the AI, it would actually automatically bring that up whenever that is needed. So it can really speed up that process and give you the scale of the things that Cheryl was talking about.

And one of the things that we found as we've been working with clients and really bringing them into looking at applying this product asset management approach and thinking about packaging, this being one of the central points to those seven touch points is the fact that, if it can get the final packaging design and done early enough and into the process, the digital product life cycle early enough, it really does actually speed up things.

And by using AI to connect the content and the data, it helps with that getting the packaging, getting the packages design done earlier in the overall product life cycle. And in doing that you can get to market quicker because you're not going through all the iterations that we've seen people go through before in terms of design changes impacting information.

So again, just to reinforce these seven areas that we fit, we've seen across the product life cycle and packaging design I think is central and one of those key areas where we actually see that by applying AI is a real improvement and reduction in the time it takes to market.

And if you think about, just thinking another example in terms of allowing the scale by connecting your products and your product information with your assets, using AI, connecting that to your business systems, your design systems, it really allows you to do the ambitious scaling that we saw with 7 million job which I think is a great example.

So really what we're talking about is taking a product asset management approach and enriching your assets with AI. And it really becomes the connective tissue across the product systems that adds value throughout your product life cycle. Our AI service at Nuxeo, it's called it Nuxeo Insight and it really allows you to build those custom business driven specific models within apps, it also works with the generic AI as well.

So you can start with that first layout and then build your own algorithms and content models for AI specific to your business and what we feel. That's where you can really get real value is by doing that which will then give you a consistent view of your product related assets and associated data across the product value chain, which really helps you reduce the time to market and hopefully that will drive some revenue growth as well.

Rajagopal: Thanks Alan, and thank you Cheryl. I appreciate your insights there and I especially love the very detailed case studies and use cases for this because I feel like that's where a lot of the attendees get so much value. So yes, we do have some time for questions.

So the first question I'm going to start with: With all the variables and packaging and all the decision makers, who or what functions do you typically serve? Packaging or R&D as an example? Why don't I start with you, Alan?

Porter: That's a very good question. We usually find thoughts with folks in design, and often as you said packaging R&D is that we often find that that also leads to conversations with the folks who are immediately upstream and downstream of themselves. We usually end up working with a combination of packaging, the photo studio, and the Nuxeo people together.

But what we like to do is try and connect across to at least two or three stages of that product life cycle. So can depend on whether the initial entry point in the initial business cases but for instance the cosmetics company I mentioned earlier that was roughly we were focused with working with the packaging R&D folks there. And then that led to conversations with the department on either side of them.

Perkins: Just to build off what Alan said, a lot applying to stuff starts with R&D and package design. But what we're seeing more and more is that we try to engage brand and marketing because we're trying to make sure that especially with the role AI and design play together, that no matter what the iterations are, thousands or in the example of Nutella, millions, we want to make sure that we're taking into account not only the look and feel of the brand today, but where they might be going. So it's really important that that brand piece is also brought in sooner versus later.

Porter: Yeah, the sooner you can get everybody talking together and understanding what the brand story is and how they actually want to communicate that and build it, the better.

Rajagopal: And this one is I think is a follow-up to that: Do you bring ideas or do you serve as a researcher?

Perkins: I've seen it both. But the value I've seen in the cross-functional interaction is that people end up thinking of different ideas building off each other and then evaluating them together to see which ones make the most sense in terms of the patterns or the colors that then you feed into the smart AI system to generate the variants.

Porter: Yeah, just to clarify, Nuxeo was a software platform and we come in ready to help build a platform that allows us to manage the assets, the metadata and move that content across the organization and build by connective tissue that I was talking about, connecting content and data together. We have our AI services on top of that. But because we're doing that we are very much building ongoing longterm partnerships with our clients, so now they're, "Hey, here's a tool to..." We walk away.

So we do end up working with our clients to help both in a consultative and service mode as well. As well as partnering with industry experts in which have a particular industry we're working in, with folks like Cheryl or what if they help with consulting. So we like to be that catalysts of change that we talked about. I think we both had to use that gracing our introductions. And that we helped spark those new ideas and spark those conversations and open up some of those cross functional conversations as well.

Rajagopal: I think you touched on it a little bit, but again just along those same lines, how do you convince key stakeholders to invest in AI and data collection? And I'll add how do you get started? I think that's usually what's most challenging for most is where do you start? So Cheryl, can I start with you?

Perkins: I think one thing that the whole thing you have to do here with AI and design is really try to shift the mindset. So it's not about the cost of the investment in doing this, but it's more the investment of what you get out of it. And so in getting started, it's really the case for change or why you need something like this. And basically, if you think about a designer spending how many days to get the five designs versus a designer, setting up the colored patterns and boundaries and using AI to generate thousands or millions of variants. So you can talk about the time and translate that to money, you can talk about the cost of not doing something like this and having limited variants to choose from.

And I think the whole thing of defining as Alan talked about in the examples, the amount of cycle time saved and the lack of choices versus more choices and minimizing risk. So bring on risk, cycle time, investment dollars and really try to put that case for change together. That's the way to get started is to everybody including all the key stakeholders are on board in terms of what you're trying to do and how you're trying to go about it.