Exploring the intersection of Adobe and GenAI

For those who are not aware, Adobe Summit is an annual event hosted by Adobe, primarily focusing on digital marketing and customer experience management software and platforms. This year, their creative cloud services and new platforms like GenStudio took the main focus. Adobe’s new marketing platform for insights and creative development using GenAI services from Firefly was launched last year. The intersection of Adobe Suites and GenAI represents the integration of advanced technology with marketing strategies to deliver personalized and impactful customer experiences and content.

We have been fortunate enough to work with Adobe over the past few months on how GenStudio and future software will be critical to not only the production side of content and experiences but also on the insights, strategy and creative development side. Much like Adobe (and frankly some Forester research from last year), we believe that these tools are going to be part of the creative process which will continue to involve a significant amount of human input, insight and creativity.

When you think about this world broadly, most are talking about efficiency rather than talking about productivity and almost more importantly these tools as critical components in the human-driven creative process. Meaning, “AI helps creative people be more productive.”

Let’s unpack the differences for a second. According to Google AI – “Productivity is the amount of output or results generated in relation to the resources used. Efficiency is the optimal utilization of resources to accomplish a task or achieve a goal.” Most times, the word efficiency scares people because it implies a shift in jobs or skills. Productivity however can be considered more like a tool in a toolbox, which is what many of the features and new offerings from Adobe highlighted.

Let’s now look at some of the things we saw through this lens specifically related to content and creative strategy.

GenStudio: Adobe’s GenAI offering

We have known about this here at Valtech for some time but were excited to see it in action and some cases from Coke, IBM and Pfizer. Essentially, GenStudio uses models (both out of the box and custom) you can build for a brand based on Firefly. The “Five building blocks or steps” are:

• Workflow and planning

• Creation and production

• Asset management

• Delivery and activation

• Insights and reporting

Content is fueling the global economy. Digital content is really the way that enterprises need to engage their customers and drive and grow their business, and this is a single solution and application that they can do all of those five stages on.

- David Wadhwani from Adobe

One thing to clarify here is GenStudio does not replace Creative Cloud but is just more focused on the full marketing supply chain. Meaning that a model in Firefly can work inside multiple areas of the Adobe stack (it’s essentially an API). The idea of “model portability” is exciting not only from a workflow and communication perspective but from a single source of truth and sharing of responsibility perspective. We attended a round table discussion with some Adobe product owners about use cases where they create assets but then need to hand them over to an internal team or another agency. Now imagine that you can also hand over a model in a secure and IP-complaint manner just like you would a PSD file. Pretty cool! The other thing they announced which is not live yet is creation of audio, video and even 3D. Take note Eleven Labs, Sora Google Lumiere and others!

Improving productivity and performance with Adobe’s AI Assistant

Adobe’s AI assistant is positioned to help with productivity and performance, but the ability to inspire strategists and impact creative really clicked during the demo of the tools' ability to gain insights from data.

In one demonstration, we saw the AI Assistant embedded in GenStudio. We saw the tool give a marketer the ability to build a custom audience to help with remarketing based on purchase behaviors. The user would ask the AI Assistant questions about customer behavior from their existing database, and within seconds the Assistant would respond with visualizations and insights.

The quick turn in results allowed the marketer to ask follow-up questions that dug deeper to unlock more interesting insights. This all happened in minutes, and as Adobe described, with current processes it can take weeks for a team to get answers to one question—and even more when you have follow-up questions.

While the demo showcased how to help build a campaign audience for email targeting, the impacts upstream to briefing and creative inspiration were also evident. It’s difficult as a strategist to access consumer data without a lot of upfront discovery time. If we can learn behaviors and preferences at the snap of a finger, we can better brief our teams.

Understanding which products lead to a follow-up purchase (as shown by Adobe) can help understand what is important to consumers now. It can validate broader trends in the market or point out a unique attribute for a brand’s particular audience.

Data is just data without someone digging in and looking at contextual and interesting connections to uncover the next level of meaning to provide insights. The ability for a strategist (or anyone) to ask for insights using plain language prompts democratizes the ability to gain insights. But the ability to push in and ask the second, third and fourth questions is where the partnership of human instinct and AI speed and access really brings a new level of creativity.

Your AI journey starts here

With Firefly services, brands can access out of the box models to help produce content at scale while custom models can be developed to work within the whole suite of Adobe tools—such as GenStudio and Creative cloud. This shouldn’t be taken to suggest that the human element of the work is close to being eliminated. These announcements are new tools that need to be created, maintained and most importantly approved by humans. The feedback loop on the branded custom models is so important and the no code aspect lets anyone manage and create advertising, which is amazing, but it still involves human ingenuity and compliance.

It’s the ability to train a model based on provided assets. The primary intention is to train it on a specific brand with owned assets to provide a reference set for GenAI that is tied to brand elements (colors, photography styles, design elements, characters, etc).

But, through the demos and our 1:1 conversation with Ben Vanderberg (Principle Platform Evangelist at Adobe Firefly) there are more ways to look at this that can not only help with generating actual, usable assets, but also can impact earlier phases of the creative process (briefing, ideation, brainstorming, etc).

Say you have a custom model that is tied to your brand, you can have that and protect it and set it aside. But you can also take that model and create others that you can “nudge” to a campaign need (audience, theme, trend, design style, etc).

We could also train models that have nothing to do with a brand – but say we train a model based on Gen Z aesthetics and a specific product. We can use this in the briefing process to help generate inspiration that is more tied to elements of a brief but would never be published.

The impact of these changes is not lost on us. It has spurred many conversations both internally and with clients on how we should implement AI tools in our own work and how we should guide clients in their AI efforts.

A few things that really became clear were the needs to have end-to-end integration. You can’t just look at a tool in a silo, you need to look at entire processes and workflows to best make integration decisions. The other big takeaway was that we need to stop some of the homegrown AI efforts. The speed of innovation and investment needed to really make useful AI tools is impossible to manage unless it is your entire business. Now we always will need to customize how things are integrated, but in this specific case the days of building your own engine should be considered secondary.

We are excited to apply our initial learnings in clients' situations as we do more assessments and proofs of concepts on how AI can be applied for our clients. Reach out if you would like us to help you start on your AI journey.