Enhancing the visual design of an enterprise AI Assistant for the BMW Group
Elevating user experience by enhancing the visual design of collaborative virtual workspaces developed for global internal stakeholders of the BMW Group.
Role: Lead User Researcher
Methods & Tools: Semi-Structured User Interviews, Stakeholder Interviews User Stories, Archetypes, Concept Testing
Software: Miro, Powerpoint, Adobe Illustrator, Teams, After Effects, Unreal Engine
Contribution: As the lead User Researcher, I directed a team of 10 designers to execute a rapid research plan that informed design development throughout the project.
Timeline: 3 Months (September-November 2023)
Team: Team of 18 (Research team-10, Art team-8)
CLIENT BRIEF
To generate design concepts and integrate emotions into the visual experience of BMW’s future collaborative virtual workspace for global employees.
THE CHALLENGE
BMW’s initial attempt at designing their AI powered virtual collaborative lacked the emotional element. Our challenge was to evaluate the current design and conceptualize improvements to enhance the visual experience.
OUR APPROACH
Through a rapid generative research study, our team identified needs of key stakeholders in the car production process. Our research informed crucial design decisions with Archetypes helping build personalized workspace environments and design principles and experience framework informing the design of interactive AI Assistant form, sound and voice personality.
BACKGROUND
A vision for the future of work at the BMW Group
The BMW Group envisioned the future of it’s workplace through a central AI-powered platform that enables global stakeholders scattered across North America, Europe, and Asia involved in the car design process to enhance productivity and speed up the car production process. The complex vehicle development process involves multiple inter-departmental stakeholders scattered across geographies. Each stakeholder group has varied processes and behaviors in terms of collaboration and communication. According to the vision articulated by client stakeholders, a central virtual platform which is immersive and integrated with AR/VR technology can help simplify the communication chains and enable faster collaboration thereby speeding up the production process.
PROCESS & TIMELINE OVERVIEW
Our cross disciplinary team of design researchers, illustrators, interaction, motion, sound and technical 3D designers collaborated over a course of three months to research, design and present findings and design deliverables to clients.
UNDERSTANDING THE BRAND
Before the joining the project, I familiarized myself with BMW’s brand & mission, product range, recent launches and innovation ventures through news articles, company website and other credible sources about automotive industries. This helped me understand the values of the company that seep through the design of all their products, strong focus on brand, the emotional connect with their customers as expressed in their motto: We exist to move the Body, Heart & Mind.
View brand study report
PROJECT KICKOFF WITH STAKEHOLDERS
The brand, the client stakeholders and project expectations
Our team at SCADpro worked with the team from BMW Group, USA located at Greenville, SC for this project. The project started with a meeting with two client representatives at the Deloitte Foundry which is SCADpro’s Innovation Studio Headquarters in Savannah, GA.The clients presented a deck to help the team understand their mission, values and project goals, success criteria & expectations.
During the kickoff meeting, the team compiled questions to clarify context and understand project scope in detail
WHEN WE STARTED, BMW HAD…
A dummy user interface showcasing a generic user flow for a collaborative virtual meeting between car designers
A concept for a neutral 3D environment
A design concept for the form of the AI Assistant
HOWEVER…
“It has no emotions!’’
In the client’s own words, the AI Assistant was described as lacking "emotions." The clients were concerned about the potential impact of this lack of emotions on user engagement and adoption rates across their global employee groups.
RESEARCH PLANNING
Soon after the project kick-off meeting by the end of week 1, team leads were appointed as per the requirements of the project by the head of the project. With a clearer understanding of the project brief, I used the north star framework that helped me frame the project brief into a clear goal, a user-driven focus and a measurable business outcome. By presenting this framework to the extended team, I helped foster alignment across cross-functional team members to ensure a unified strategic vision and a shared understanding of the project’s intent.
Since the project focused on elevating the visual design, I structured a research plan to allot two weeks to in-depth research, followed by rapid secondary research to inform iterative design development. This provided sufficient time for the technical design team to develop initial design concepts ahead of the week 5 client check-in.
Secondary Research
In-depth secondary research was required to understand the AI space in detail. Along with AI, our research probed into understanding virtual assistants, existing research on emotions and AI and human-AI interaction.
Primary Research
After it was established that the root of the problem was inefficient communication & collaborative processes, I decided It was important to understand the intricacies of the work and collaborative styles of key stakeholders in the car design process from the users themselves. I found insights from primary research to be crucial to kickstart the ideation for design development, with deliverables like design criteria’s, principles & user archetypes.
Testing & Validation
Significant time needed to be allocated to the visual design team, as it was the primary focus of the project. It was imperative that the second half of the project be dedicated to design development, with research playing a supportive role through concept testing and ongoing secondary research to inform design decisions and validate concepts.
SECONDARY RESEARCH
Our secondary research explored relevant areas such as AI use cases, human-AI interaction, the role of AI assistants in the workplace, the future of AI, and BMW’s AI. The broad themes that were probed upon helped us understand the functioning of AI technology, its future, and its use cases and ways to integrate emotions into the AI assistant. We created four sub-teams involving the entire group of designers and researchers to tackle specific areas of research and collectively synthesized findings using the pains, gains, ideas and questions framework. Each team collaboratively clustered data points in an in-person session and later digitized it into the Miro board for documentation and further translation of insights.
INSIGHTS FROM ROUND I OF SECONDARY RESEARCH
After analyzing data from the round 1 of secondary research, I categorized insights to create a strategy for ways to integrate emotions into the visual design of the AI Assistant. View full slide deck here
How users perceive AI Assistants:
People value autonomy at work
The presence of an anthropomorphized helper like Clippy can undermine an individual's perceived autonomy. Prioritizing user autonomy in the assistant's design will ensure that the system values and respects the user's independence.
People treat robots like they’re alive.
User’s preferences and aversions stem from societal expectations. Users subconsciously treat automated technology like it’s alive.
Users find comfort and enjoyment in familiarity
Users feel at ease and enjoy experiences that are familiar to them. Introduce users to virtual environments gradually by incorporating elements from the real world. Strive for a balanced blend of reality and imagination.
Integrating emotions virtually:
I identified two types of elements that could be used to integrate emotions into the design of the AI Assistant:
Tangible elements: Physical or visible design elements that directly convey emotions, such as colors, form & shape, etc
Intangible elements: Subtle, underlying factors like trust & privacy, transparency, empathy driven conversation design, etc.
We decided to prioritize on tangible ways to integrate emotions to narrow the scope of the project.
Through tangible elements:
COLOR
Leveraging color psychology can help designers create more engaging user experiences. We looked into the "Wheel of Emotions", a model developed by psychologist Robert Plutchik to illustrate various basic emotions and their relationships.
FORM & SHAPE
The way users perceive and interact with virtual assistants is significantly influenced by their outward appearance. Users form expectations about the capabilities of virtual assistants based on their appearance. Biophilic design integrates natural elements into the design of objects, including robots. By avoiding overly human-like appearances, biophilic design helps reduce the "uncanny valley" effect, where people feel uncomfortable around robots that look too realistic.
VOICE & SOUND
Creating engaging, convenient and intuitive voice interactions without relying on visual aids. People are most comfortable when they get a response that matches the tone and formality of their request.
Sound Feedback for Alerts: Designing sound feedback for alerts, notifications, and error messages should consider the emotional responses these sounds elicit. Users' satisfaction and preferences can be influenced by the appropriateness of sound choices.
Volume and Intrusiveness: Research has highlighted the importance of maintaining appropriate sound levels and avoiding overly intrusive or disruptive sounds. Users tend to prefer sound feedback that is non-disruptive and adjustable.
MOTION
Motion can greatly enhance emotional engagement and the overall experience of a product through animated transitions, Micro-interactions, visual feedback, and voice visualizations. When utilizing motion, the ultimate goal is to provide users with a convenient, screen-free, and intuitive experience, meeting their needs and preferences.
DESIGN PRINCIPLES
Insights from round 1 of secondary research provided a direction for the design team to ideate on initial concepts for the AI Assistant form.
Adaptive
The design should be adaptable to be implemented across BMW Groups sub-brands and subsidiaries. Our clients emphasized on that they would like for the parent company branding to reflect in the designs. Integrating key elements from BMW’s brand identity to encourage a sense of active involvement and pride among employees.
Emotive
It should be interactive and emotive in its responses, conveying appropriate responsive feedback to users with colors, form, shape and motion to facilitate effective communication and teamwork among stakeholders.
Iconographic
The design should be iconographic, incorporating brand elements while maintaining the emotive ability. The ‘Iconographic’ principle helps avoid the uncanny effect of anthropomorphic and humanoid designs.
INITIAL DESIGN ITERATIONS-AI ASSISTANT FORM
Insights from the first round of secondary research informed the ideation of initial concepts for the AI Assistant's form. The concepts feature adaptable brand elements, such as the company's brand colors and logo, emotive shape-shifting forms, and biophilic elements, to ensure a sense of familiarity among users while avoiding anthropomorphic forms to mitigate the uncanny valley effect.
Several concepts were explored during the initial conceptualization stage that signified the essence of an ideal collaborative experience intertwined with the brand identity of BMW. Ideas ranging from company origins and historical references to variations in logo design. Furthermore, the team experimented with typography, transformative objects, natural forms, and origami shapes to develop a form that conveys emotion effectively. These concepts were presented to clients during the fifth-week mid-point check-in, where they received positive feedback for significantly shifting the design direction from the original concepts and introducing innovative approaches to designing the AI Assistant.
PRIMARY RESEARCH PLANNING
While the design team ideated on concepts for the AI Assistant form informed by insights from round 1 of secondary research, we continued our research process to conduct primary research to develop further concepts.
The car design process involves four primary stakeholders: Designers, Engineers, Production Planners and Decision Makers. User research was important to understand collaborative styles and pain points, so we could create designs that would appeal to users and increase adoption of the new technology across global employee groups of the company.
Initial primary research plan: I planned to conduct in-depth user interviews with 4-5 users under each user group, to get an overview of their roles and collaborations among cross functional teams in the car design process. Considering that our primary users—employees of the BMW Group—might be unable to dedicate time for a 30-45 minute interview session because of work commitments, along with potential time zone and language differences, I planned to include a quick survey. This survey would help gather supportive data in case we were unable to secure enough interviews.
View interview & survey questions
Navigating participant recruitment challenges: We could not connect directly with employees of the company for interviews or a survey. As external consultants, we were advised against directly contacting employees at global offices of the BMW group scattered across Asia and Europe. Without direct access to our users, strict NDA’s that and the need to get user perspective on the design to production process that wouldn’t be available in any secondary research material, I, along with the team lead pushed for the idea to leverage university connections to recruit user surrogates. We reached out to 4 professors who had extensive experiences working as designers in other leading automotive companies and gathered data on the workings of the automotive industry, their collaboration patterns with other stakeholders and their needs and pain points that hinder effective collaboration.
Interviewing car designers helped us understand details about the collaborative processes of not just designers, but other key stakeholders as well. The findings represent the point of view of the designer within the car production process and their perspectives on the collaborative environment in the industry.
DEFINING THE USERS
Lack of equal access to information: The fast-paced changes in management and outdated data hinder the designers' progress, leading to a lack of motivation and enthusiasm. There is a lack of transparency in information sharing, especially when information is selectively shared with certain teams.
“I think the most difficult thing is you'll be working with different studios. And then you know, like some bits of information are shared here and some are shared there, right? So that's a big problem.”
Senior Exterior Car Designer
Lack of clarity and direction: Designers feel lost without clear direction from management, and there is a disconnect between business leaders, design teams, and other stakeholders.
“The ideal collaboration setting will be one where things are done in the appropriate order, where the design directives are clear, are set collaboratively and agreed upon.”
Design Manager at Global Car company
Lack of understanding & slow information chain between stakeholders: There is a significant challenge in aligning cross-functional team goals with business objectives. The current design review process involves time-consuming steps, including passing information through multiple hands, leading to delays.
“The biggest problem is that business people have no idea what design is, and to be fair we have no idea where they are, we both come from different angles”
Exterior & Interior Car Designer
ARCHETYPES
I created the archetype, detailing characteristics of user groups and an overview of their primary needs, to help inform a more user centered approach towards the design of the collaborative virtual spaces. Research was conducted to inform use cases for the AI Assistant, I categorized insights to deliver keywords that would help inform themes for the design of virtual environments for collaborative activities between stakeholders. I derived insights from both primary and relied heavily on secondary research.
USER
NEEDS, CHARACTERISTICS
Designers
Designers primary goal is to push the boundaries of creativity to drive design innovation. Their processes are collaborative, highly creative and their primary needs are a sense of clarity and harmonious collaboration.
Engineers
The Engineer’s work revolves around practicality & precision and requires them to balance business goals, innovation with technical feasibility.
Production planners & decision makers
Both production planners and business decision makers work collaboratively to make high-impact decisions that shape companies. Visual and collaborative thinking, simplifying complexities and making strategic future focused decision are things that their roles demand.
SO WHAT?
A serene space that boosts creative thinking, helps bring clarity through the calmness of the environment.
A workspace designed to reflect the practical nature of engineers, and fosters precision and accuracy in their work.
An environment that allows for future focused planning, boosts focus to help them simplify systemic complexities and foster impactful decision making
KEYWORDS/MOOD/FEATURES
Keywords/mood: Serene, minimal, natural, calm
Features: boosts creativity, harmony
Keywords/Mood: Practicality, precision
Features: Ample space for technical accuracy
Keywords/Mood: Productive, complexity simplification, corporate
Features: Dedicated areas for decision making and information exchange
ENVIRONMENT CONCEPT IDEATION
With a clear direction to proceed with concept ideation, our team created three concepts for 3D virtual spaces, personalized to the needs of each user group. The sketches were conceptualized based on research about user preferences, color psychology, the future of the virtual workspace, and stakeholder archetypes. Each environment is a space designed to induce an efficient collaborative environment tailored to the needs of each stakeholder.
DESIGNERS
In the case of designers, consistency, empathy for users and collaborators, alignment across teams, and a creative atmosphere are crucial to fostering innovative thinking and creative collaboration. To fulfill those needs and help foster innovation and creative thinking, we designed a calming and serene space with minimal distractions. With natural lighting that complements foliage, emphasizing nature and an organic feel, we've created a space where creativity flourishes, ideas flow, and collaboration is seamless.
NATURAL/ORGANIC
ENGINEERS
Engineers, who are the architects and builders of products are required to balance business needs, innovation and technical feasibility. Their role demands precision and accuracy. The environment we crafted is heavily inspired by warehouse environments—a nod to the practical and purpose-driven nature of Engineers. With a large, open space, designed with a minimalistic approach, This space is intentionally devoid of distractions, allowing Engineers to work with precision and efficiency.
INDUSTRIAL/RUSTIC
DECISION MAKERS/PRODUCTION PLANNERS
The roles of production planners and decision-makers demand attributes such as complexity simplification, strategic foresight, resource optimization, a people-focused approach and resilience. The environment design features a sleek, eclectic layout, intelligently divided into three sections: a planning area, a discussion zone, and a dedicated space for displaying crucial information. This design caters to the multifaceted needs of both roles, fostering a harmonious and productive workspace.
CORPORATE/LUXURY
AI ASSISTANT FORM: ITERATION 2 & CONCEPT TESTING
After the initial concepts and research were presented to clients, few concepts were finalized for further development and iteration. Options that conveyed the brand Identity of BMW were well received. The further direction provided during the client presentation emphasized the combination of different elements from the many options available to create a more versatile and engaging solution.
The user testing involved refined designs for Mantis and Chime, both of which are combinations of elements merged from the different options presented earlier.
Mantis:
Mantis was designed with elements incorporated from Mini, Rolls Royce, and BMW as the AI Assistant will be scaled for usage across stakeholders under the BMW Group. The three petals embody the essence of the three brands, with the BMW logo at the core of the design. Upon activation, Mantis expresses itself through the fluidity of its liquid core, dynamically changing both shape and color. The transformation of colors and shapes in Mantis corresponds to the emotions expressed, serving as a visual cue to inform the user about the various interaction states.
Chime:
Based on the combination of two initial concepts, Chime is envisioned as a calming AI inspired by the aesthetics and tranquility associated with wind chimes. When activated, Chime peels off its sides to unveil a glowing core. The core undergoes a chromatic transformation, while the rings dynamically change shape, allowing Chime to emote on two distinct levels.
CONCEPT TESTING
Concept testing was conducted in two phases. In the initial round, we implemented A/B testing to evaluate various interactions of the Virtual Assistant in relation to the user's emotional responses. This initial testing involved 5 participants, including students and faculty members from SCAD, representing diverse disciplines such as user experience, animation, service design, and industrial design. The insights gathered from this round proved valuable in refining our design output. View data
Key Takeaway
After a thorough evaluation of the emotional and interactive states of AI Assistants, Mantis and Chime, and a comparison of their designs and animations, the key takeaway from user testing is that, despite limitations in expressing specific emotions like 'confusion' and 'error,' Mantis emerged as the preferred choice over Chime for both emotional and interactive states based on user preferences.
IMPROVEMENTS
Integrating Mental Models
Mental models serve as cognitive frameworks that individuals employ to comprehend, interpret, and respond to their surroundings, aiding in the navigation of complex situations, problem-solving, and decision-making. These models essentially represent simplified versions of reality, facilitating the processing of information and guiding behavior.
Upon conducting user testing, we identified a lack of understanding regarding some emotional and interactive states that were created. In response, we delved into the mental models associated with our chosen emotional and interaction states. For instance, a mental model for the "notification" state encompasses elements such as the red circle displayed on apps, pop-up dialog boxes, signs, vibration, and the color red. These components were then incorporated into our notification states.
The outcomes of our second round of user testing were markedly more successful compared to the initial round, underscoring the effectiveness of integrating insights on mental models into the designs.
SOUND & VOICE CONCEPTS
In the world of virtual assistants, the choice of personality plays a pivotal role in shaping user experiences. Each personality type serves a distinct purpose, aiming to resonate with users in unique ways. Our sound concepts were designed to cohesively integrate with the visual language of our environments and the AI assistant.
Voice Concept: Intelligent and Knowledgeable AI
As we delve into the specific personality type of Intelligent and Knowledgeable AI, our focus lies in presenting a virtual assistant that exudes expertise, providing users with comprehensive information. It's essential to recognize that these AI personalities are not one-size-fits-all; they vary based on application, context, and user requirements.
Our approach to sound design centers on creating an auditory landscape that intersects comfort and harmony, enhancing the overall user experience. In our exploration, we have honed in on the Intelligent and Knowledgeable AI to deliver a virtual assistant that aligns seamlessly with user expectations. Our focus was to create voice concept that would comfortably convey complex data and results to the user, but also would be professional and trustworthy without being playful or too friendly.
Sound design
The sound design for our AI virtual assistant lies on the intersection between comfort and harmony. The goal was to drives user engagement by instilling trust and providing an intuitive sound experience. We designed the AI to have a calming essence and the sounds created helps users intuitively understand the emotional and interactive state leading to lesser errors and confusion.
FINAL: AI ASSISTANT FORM & EMOTIVE STATES
MANTIS
INTERACTION STATES
3D ENVIRONMENTS: COLLABORATIVE VIRTUAL WORKSPACES
DESIGNERS
ENGINEERS
DECISION MAKERS/PRODUCTION PLANNERS
CONCLUSION
Our research highlights the core needs of our user stakeholders, centered around the pursuit of efficiency and productivity. At the deepest level, users seek to be understood by collaborators, align with team goals, be equipped with all necessary resources and information, and seamlessly collaborate with both internal and external stakeholders throughout the vehicle development process. To address these needs, we designed our virtual assistant to be intelligent and emotive, featuring a calm, intelligent, and comfortable voice, expressive movements, and colors that intuitively convey the nature of the information. Additionally, warm and soft sounds are seamlessly integrated to foster an environment conducive to productivity and efficiency. Moreover, our environments are crafted to be immersive, incorporating real-world elements to instill a sense of ease and familiarity. These spaces are personalized to each stakeholder's collective needs, contributing to elevated levels of efficiency and productivity.
VISION VIDEO
RESEARCH IMPACT
Aligned fragmented teams to drive cohesive efforts and data driven decision making
The research team utilized the north star framework, design thinking workshops to evaluate successes, failures and ideas with cross-functional teams in order to have constant communication with different teams, communicate findings and drive data driven decision making design development
Identified key issues with the current design and recommended a direction that incorporated emotional design, which was successfully implemented and led to the desired outcome, securing client buy-in for implementation
Our research resulted in strategies and recommendations that offered a structured approach to integrating emotions into the visual design through tangible design elements like form, color, motion. Insights were communicated visually through archetypes, visual frameworks and design principles.
MY CONTRIBUTION
RESEARCH PLAN
Developed a rapid research plan for a generative research study. Defined research goals and narrowed the scope of the project to ensure project completion on time and for client satisfaction.
RESEARCH, ANALYSIS & SYNTHESIS
Collaborated with cross-functional team members to drive research activities such as secondary research, participant recruitment and user interviews.
Utilized affinity mapping and thematic analysis to analyse and translate user insights.
Designed the ideal experience framework and 4 user archetypes. Co-designed the design principles.
TEAM ALIGNMENT
Defined the vision for the project utilizing the North Star framework, an enterprise design thinking method to align cross-functional stakeholders on a unified vision for the project.
DESIGNING & PRESENTING PITCH DECKS
Collaborated with the visual design team and researchers to lead content development for 4 client decks.
Planned presentation scripts and presented findings to client stakeholders.
Wrote final client report to document entire process and outcome.
LEARNINGS
Don’t discount the value of secondary research: Primary research is an essential part of user research, but secondary research can be just as impactful and even critical to a project's outcome. It is quicker and served the needs of this project. This project heavily relied on secondary research to shape recommendations, while primary research served as a supportive element—rather than the other way around, which is the typical approach in most design research projects.
Speak the stakeholder’s language: One of the biggest takeaways from the project was learning the importance of curating insights in a format that is both digestible and tailored to each stakeholder, depending on the purpose of the communication. Crucial team bonding time should not be taken for granted no matter how fast paced the project is. It is important to get to know team members and their styles of communication, strengths, weaknesses and collaboration styles within a short timeframe for highly productive output.
Have a Plan B and C: I learned the importance of planning for worst-case scenarios. Things can always go wrong, especially in research, where participant recruitment challenges are a common hindrance. Planning for such scenarios helps ensure the project runs more smoothly.
