Behind the design: Adobe Brand Concierge

The process of designing a personalized and immersive AI-powered conversational experience

That effort was eventually absorbed into the Experience Cloud organization, where the product is being further refined and launched as an AI-powered conversational companion for brands.

Brand Concierge guides consumers from exploration to confident purchase through immersive, self-serve and highly personalized experiences. With context-aware personalization and brand-controlled AI, it ensures seamless, engaging interactions and takes intelligent action based on real-time activity with each consumer. It empowers companies to create digital brand experiences, where every customer interaction is intuitive, immersive, and effortless.

Principal designer Rebecca West, who’s worked on the application since its early-stage innovation efforts, looks back on the design process that simultaneously considered the end customer journey and the interface for the people using the application.

What was the primary goal when you set out to design Brand Concierge?

Rebecca West: When we began looking at digital experiences in the age of generative AI, we envisioned a future where they were more immersive, personalized, and conversational—with the ability for companies to interact back-and-forth with their customers in a meaningful way.

Our main priority for Brand Concierge was to make sure it was low code and easy to set up using natural language, which meant ensuring that a business user, who may not be technical, could set up the experiences and feel confident about their impact. We also wanted to balance the dynamic nature of the generative technology (where large language models have some degree of agency to interact with and generate responses to customers in the moment) and control elements dictated by the needs of the business user (such as ensuring that elements like product details and pricing would remain consistent and couldn’t be dynamically changed by models).

We also kept in mind that we were designing for two sets of users: our enterprise customers (who needed brand control and management) and their customers (who were on the receiving end of the output). It required us to be empathetic to an extreme while designing to ensure we were accurately considering each of their needs.

For the core industries we focused on to start (retail, travel, hospitality) the enterprise customers are ultimately driving towards conversion and brand loyalty, while their customers are focused on solving their shopping or consumer needs. There’s a happy medium to be found between an enterprise customer and their customers, where both can benefit from the experience.

What user insights did you leverage to help inform your design solution?

Rebecca: For Brand Concierge, the foundational insights that informed our design approach came from the power of our multidisciplinary team. We partnered with many teams, each with insights to contribute—corporate strategy, user research, product management, machine learning, engineering, and of course, our customers.

The early partnership with corporate strategy provided an outside view of the market and an educated analysis of trends. Machine learning and engineering teams building AI powered experiences within Adobe were able to share deep knowledge—ultimately, we’ve been able to leverage quite a bit of the technology. Adobe Design’s Research & Strategy team was also an extremely valuable partner. We ran many user research rounds (and continue to do so now) to understand people’s mental models, expectations, and how we could help them achieve their goals (which is extremely important when you're trying to create confidence and conviction in your design approach.

Because every industry has its own expectations we tested different versions of Brand Concierge, with different industries so we could distill feedback into our design strategy. At the same time, we were also researching the end consumer experience and trying to answer questions about what people were expecting in terms of the amount of engagement, personality, and response types.

What was the most unique aspect of the design process?

Rebecca: How quickly generative AI is evolving and along with it, societal understanding and expectations. When we started exploring the future of experiences, a conversational experience was quite impressive, but now with the wide use of tools like Microsoft Copilot and ChatGPT, AI-powered conversations are becoming commonplace. Adapting and evolving with that meant watching how consumers are starting to respond to those experiences so we could better understand what their expectations are now and what they might be further down the road.

Additionally, since we knew that creating a high-quality experience would require “real” content, data, and insights and that the quality of the experiences would be inextricably tied to the generated experiences of this foundation, we invested heavily in a design partner program. The program allows our multi-disciplinary team (product management, engineering, machine learning, user research, and design) to co-innovate with customers so we could deeply understand their needs and the needs of their customers, and prototype with generative experiences based on their actual content, data, and insights.

It helped us understand the business needs of different industries to scope out the biggest pain points and discuss options for addressing them. As an example, one design partner was focused on the Brand Concierge providing a personalized app recommendation. Specific approved data sources were used to generate real responses and check the quality of the answers and customer interactions. In this instance we learned how many questions to ask before giving a recommendation that felt personalized. We also learned, through user research of actual responses, that there’s a sweet spot of interactions: too few interactions and it wouldn’t seem personalized; too many interactions and it felt cumbersome. We also explored how dynamically changing the length of responses to mirror individual users might resonate with people, and how much personalization would be meaningful based on context and user preferences.

These design partners are allowing us to, quite literally, test how Brand Concierge will work in the real world before it gets to the real world.

What was the biggest design hurdle?

Rebecca: Many current enterprise applications of generative technology heavily rely on human review before being exposed to customers or are internally facing and provide a forgiving environment for scenarios where responses are less than ideal. With Brand Concierge, the generated responses are experienced directly by a brand’s customer, so the biggest design hurdle was designing a customer facing experience, based on generative technology, while enabling companies to maintain their brand standards.

When we spoke with our customers, they understood that the potential for individualized personalization was huge, but launching a customer-facing experience that would generate unique responses for everyone also required a level of trust. AI principles told us that humans need a level of control, visibility, and transparency with AI, so we knew there was a need for them to be able to configure and control elements for a concierge, but for a brand to feel confident to launch would require a lot more.

Our driving hypothesis was that we’d need to instill confidence throughout the experience, in the configuration, in the ability to test at scale, iterate, and improve, and to measure and improve the experience once it launched.

By listening to and observing our enterprise customers we learned what they’d tried on their own and what had failed. And as we launched various AI features with internal teams, we also saw the need to do stringent testing, to add guardrails, and to explore both automated and manual testing methods at scale. Since these customer experiences would serve as ambassadors for a brand, there was a level of quality we needed to achieve, and a level of confidence we needed to instill in both users and their stakeholders.

We continually pushed to provide business users with that confidence (in quality, control, with feedback loops, and the ability to achieve business goals) which trickled down to specifics in the business user experience: the ability to dictate scenario handling, to control voice, brand styling, and guardrails, to test, monitor insights, and measure results, and to iterate and improve the consumer experience.

With the goal of driving confidence, we created an incremental path to a north star fully guided conversational experience. By starting with a minimally viable and delightful product (with the basic ability to control goals, scenarios and brand elements) we can drive toward that guided conversational experience as we learn from our design partners and adapt the large language model. Our thinking was, if a user is launching a conversational experience, having that same type of conversational experience in the configuration would build confidence in the technology and empathy for what customers experience.

How did Brand Concierge solve customer needs?

Rebecca: To launch these generative experiences, companies generally must build them from scratch (a huge feat to ensure enterprise readiness) or go with a vendor. Then to get them to a truly personalized level, they must be integrated with existing content, customer data, and insights. The Brand Concierge experience—natively integrated into the Adobe ecosystem—enables speed to market and the ability to personalize experiences.

Our solution is a low code, natively integrated product experience that empowers brands to deliver personalized, immersive, intent driven experiences to their customers. The product experience accesses data from Adobe Experience Manager, for expert knowledge content, Adobe Experience Platform, for personalization and user preferences, and Adobe Customer Journey, for insights.

Teams using Brand Concierge will have control of setting up a goal-driven experience that can handle specific scenarios and brand expression. Testing is built into the experience, along with the ability to simulate different audiences to do scalable testing, to ensure it’s enterprise ready with strong guardrails. Then, once launched, business users have a comprehensive overview of key activity, opportunities for improvement, and can review real customer conversations (that human-in-the-loop feedback will continue to improve the quality of the experience).

Brand Concierge is also a new avenue for brands to gain direct insights into what their customers are asking for: By exploring new opportunities from insights, a brand can create a true two-way interaction with its customers.

What did you learn from this design process?

Rebecca: I was reminded time and time again of the power of collaboration, particularly when it brings together core teams with the trust to bounce ideas openly, push on each other, and riff on other’s takes. Having multi-disciplinary teams openly discuss every design decision was hugely beneficial, especially during incubations where we had frequent check-ins and guidance from the head of corporate strategy and, Scott Belsky, Adobe’s chief strategy officer.

I started as a single designer working with corporate strategy and user research and have expanded to have a design squad of extremely talented designers (Erica Fasoli and Roza Atarod), amazing new product partners (Winnie Wu and Loni Stark), and extremely motivated and intelligent machine learning and engineering teams. We’ve been able to push one another and push for a better experience.

What’s next for Brand Concierge?

Rebecca: Our end goal has always been to empower brands to create immersive, personalized experiences that resonate and provide value for their end customers and deepen brand loyalty. And with Brand Concierge, brands will be able to deliver true personalization at scale, across different surfaces, where each individual relationship can be unique and scalable.

As the landscape evolves (whether through technology, user expectations, or shifts in brand/customer relationships), we'll continue to work with customers to learn and grow the experience with design guided by Adobe’s core ethical AI principles.

A special thanks to Sanchit Ladha, Victoria Lu, Catherine Chiodo, Erica Fasoli, Roza Atarod, Winnie Wu, Loni Stark, Archana Thiagarajan, Ryan Cobourn and so many others without whom building this product would not be possible.