In AI design sprints, boundary objects are artifacts or concepts that help different stakeholders (designers, developers, users, and domain experts) bridge knowledge gaps, create shared understanding, and align on goals. These objects don’t belong to any one particular group but are flexible enough to be interpreted in different ways by various teams. They play a critical role in ensuring communication and collaboration remain fluid, even when participants have different perspectives.
Here’s how to use boundary objects effectively in AI design sprints:
1. Identify the Stakeholders’ Diverse Needs
Begin by understanding the diverse perspectives that will come into the sprint. AI design sprints often involve multidisciplinary teams, including engineers, data scientists, product managers, and subject-matter experts. The challenge is to reconcile their different languages and goals.
Example: If you’re working on a healthcare AI system, doctors may be focused on clinical accuracy, while engineers are concerned with performance, and product managers are worried about user experience.
Boundary Object Use: Use something like user personas as a boundary object. Personas can be modified or adjusted depending on the specific focus of each team. Designers might use personas to understand user needs, while engineers may use them to design relevant features or algorithms.
2. Create Prototypes as Tangible Boundary Objects
Prototypes serve as concrete representations of ideas and can help different stakeholders align on the final product. During design sprints, creating low-fidelity prototypes allows rapid testing and feedback loops.
Example: A wireframe or a low-fidelity prototype of a user interface for an AI product, such as an AI-powered chatbot, can be modified based on feedback from the development team, while the user team focuses on refining it based on real-world user interactions.
Boundary Object Use: These prototypes are boundary objects because they can be understood and manipulated differently by designers (for interaction design), engineers (for technical feasibility), and users (for usability testing).
3. Use Stories and Scenarios
User stories and use cases are essential boundary objects that provide a context in which different teams can evaluate the functionality of an AI system.
Example: A user story could read, “As a teacher, I want an AI assistant to recommend personalized learning paths based on students’ progress.” This boundary object is flexible enough that a data scientist might focus on the AI’s ability to personalize, while a UX designer might focus on how this recommendation system is presented to the teacher.
Boundary Object Use: This allows teams to focus on different elements of the design (e.g., data inputs, algorithm design, and user interface) while keeping the overall goal in sight.
4. Map Out Journey Maps or Service Blueprints
Journey maps or service blueprints help visualize the user experience and how AI components fit within that experience. These artifacts are useful as boundary objects because they are highly adaptable and can be used by different groups in the sprint.
Example: A service blueprint for an AI-powered mental health app might highlight the user’s emotional state and interactions with the app, such as AI-driven prompts for journaling. The design team focuses on the user interface, while the data team works on how to classify user emotions, and the engineering team integrates the AI model.
Boundary Object Use: This shared artifact allows everyone to focus on the same process flow while still addressing their individual expertise. It’s flexible enough for both macro (product vision) and micro (technical features) discussions.
5. Leverage Data Representations as Boundary Objects
Data plays a central role in AI, and it can be a boundary object that unites different perspectives. For example, a data schema or dataset visualization can serve as a tool for multiple groups.
Example: A data scientist can use a dataset visualization to discuss how data is structured, while a designer might use the same visualization to discuss how the data will be presented to the user.
Boundary Object Use: This helps bridge the technical aspects of data handling with the human-centric aspects of AI design.
6. Use Collaborative Tools for Real-time Feedback
Tools like Miro, Figma, or Mural are great for creating and sharing boundary objects that evolve in real time. These platforms allow different teams to contribute simultaneously and interact with shared materials like user flows, wireframes, and other design elements.
Example: If you’re designing a conversational AI, the dialogue flow can be a shared boundary object. While designers focus on the conversational UI, developers assess the technical feasibility, and product managers monitor how it aligns with the user’s needs.
Boundary Object Use: These shared, dynamic tools help teams collaborate on the same artifacts, ensuring alignment and creating a space where inputs from various disciplines are synthesized.
7. Create Role-Specific Metrics
In AI design sprints, defining success metrics for different roles can serve as a boundary object. These metrics translate broad goals (e.g., user engagement, system performance, accuracy of predictions) into tangible data points that can guide the design process.
Example: For an AI-based recommendation system, designers might focus on how well users interact with the recommendations (usability), data scientists might focus on the accuracy of the algorithm (performance), and the business team might focus on conversion rates (ROI).
Boundary Object Use: Each team can interpret these metrics according to their needs, but they all contribute to the overarching project goals.
8. Foster Cross-Disciplinary Conversations
Boundary objects can also act as conversation starters that encourage cross-disciplinary thinking and collaboration. Regular discussions around these shared objects allow teams to explore differing perspectives, refine concepts, and align goals.
Example: After a design sprint session, you might have a visual model of a machine learning process. Different stakeholders can assess the model from their viewpoints, leading to cross-disciplinary solutions to potential issues.
Boundary Object Use: The model itself serves as a platform for structured conversations, ensuring that each team’s concerns are addressed while fostering a collaborative environment.
Conclusion
Incorporating boundary objects into AI design sprints is crucial for creating a shared understanding among multidisciplinary teams. By leveraging flexible artifacts like prototypes, user stories, journey maps, data visualizations, and more, you ensure that all stakeholders can engage with the project from their unique perspective. This approach not only promotes collaboration but also accelerates the development of AI systems that are user-centric, functional, and efficient.