Introduction
In the first part of this series, I walked through the initial stages of the Product Design Sprint (PDS) for a sample product idea: a marketplace for crowdfunding events. We focused on crafting the problem statement and mapping out the critical path. With the help of AI, specifically ChatGPT, I was able to quickly define the key challenges faced by both event organizers and attendees.
The central theme of this blog series is exploring whether AI can enhance the PDS process, accelerating key tasks, providing fresh insights, and ultimately making the design sprint more efficient and effective.
In this second installment, I’ll dive into the next crucial phase: Diverge, but first, I’ll wrap up the Assumptions Board, which we began in the Discovery phase and will continue to refine throughout the sprint. The Assumptions Board helps identify and validate the key assumptions underlying our product concept, ensuring that potential risks are addressed early on. After that, I’ll move on to Lightning Demos and Storyboarding. In these steps, we’ll visually map out the user journey and transform our ideas into more tangible concepts. By the end of this post, you’ll see whether AI can help in refining our approach and bringing our product vision to life.
For more insights into the different phases and exercises of a design sprint, you can also check out the Design Sprint Guide.
Listing out the assumptions
Throughout the sprint, we’ve naturally made numerous assumptions about our product and the problem space. To ensure future success, it’s crucial that we identify and track these assumptions using an Assumptions Board, which helps us plan for validation.
This was the prompt I used for ChatGPT to generate an Assumptions table:
Please create a table to be used as an assumptions board based on our critical path and problem statement. The table should have four columns: the first column should list the assumption, the second column should specify whether the test group is the Event Organizer or Attendee, the third column should indicate the method of testing (such as User Interviews, Surveys, or Prototype Testing), and the last column should define the specific result or outcome that would validate the assumption. Ensure that each assumption is clearly linked to its respective test group and validation criteria.
I was not satisfied with the limited table it generated for me at first, which seemed to miss a lot of assumptions. I simply asked ChatGPT again to include all the assumptions we were making based off our research together and it generated me a more complete list.
Below is a screenshot of my FigJam board where I took the output generated from ChatGPT and organized it into cards.
After reviewing, I noticed a couple more unaddressed assumptions and added them to the board. With the Problem Statement, Critical Path, HMW Questions, and Assumptions Board in a good place, we felt ready to move into the Diverge phase to explore potential solutions.
Lightning Demos
In the Diverge Phase of a sprint, we typically spark creativity with Lightning Demos, exploring competitors and relevant products that could inspire our approach.
For this example, I used ChatGPT to generate a list of potential competitors (both direct and indirect), along with their pros, cons, and insights on what we could learn from them.
This was the prompt I used:
Create a table listing potential competitors to our product, including those in related spaces such as crowdfunding and event planning. The table should have five columns: name of the product, device (Web/Mobile App), pros, cons, and what we can learn from it. Also include the URL of the product in the name column.
Having ChatGPT generate this list quickly, and with genuinely insightful data points like pros/cons and lessons from each product, was incredibly valuable. In a future design sprint with more participants, I can see this becoming an invaluable resource. It could be used alongside the examples brought by others to ensure we cover a wide range of insights and stay up to date with the latest trends and inspiration.
Crafting the Storyboard
With a solid foundation of competitor insights and inspiration from the Lightning Demos, it’s time to move into the next phase: crafting the storyboard. This step allows us to visually map the user journey, transforming our ideas into a concrete, actionable plan.
Focusing on key moments in the Critical Path, I began creating a Three-Step Storyboard. This exercise helped refine and organize our concepts into a cohesive narrative, illustrating how users might interact with our product to solve their problems.
Even though this phase usually depends on participants’ creativity, I was curious to see how AI could enhance the storyboard process. To explore this, I had both myself and the AI contribute, comparing our outputs side by side. I approached the three-step storyboards as I normally would, by sketching concepts loosely with a sharpie and paper.
Three-Step Storyboard for the Event Organizer
I had ChatGPT choose what they wanted to focus on for their storyboard for the Event Organizer and imported the output into FigJam for better presentation.
This was the prompt I used:
Create a three-step storyboard highlighting the most important steps in the Event Organizer’s critical path. Each step should include an image and accompanying text describing the action. Ensure that all the images are aesthetically consistent and aligned.
AI-Generated
ChatGPT uses DALL·E to generate images based on the prompts it receives. The results can vary, with some being less effective or even unusable. This is a current limitation of AI, not just with DALL·E but also with other AI image generators, especially when it comes to creating UI designs.
While the images aren’t exactly on point, they do capture the overall concept. Typically, the sketches we create during this exercise are rough and conceptual, so I was pleasantly surprised by how well these images convey the basic idea, despite knowing the image generator’s limitations.
That said, my own sketches, though less polished in presentation, offer more room for imagination and inspiration when informing future designs. For my storyboard, I chose to focus on the flow for Event Organizers backing an Attendee-created campaign.
Human Generated
In my storyboard, I placed greater emphasis on the text descriptions, delving deeper into the details of how this flow might operate.
Three-Step Storyboard for the Attendee
I used the same prompt but modified it slightly to generate a storyboard that focuses on the Attendee.
AI-Generated
Human-Generated
The results of the storyboard exercise allowed me to conclude on the pros and cons of using ChatGPT in this context.
Pros of AI-Generated Storyboard
- Speed and Efficiency: The AI-generated storyboard was produced quickly, allowing for rapid iteration and saving valuable time in the design process.
- Basic Concept Capture: Despite some limitations, the AI effectively captured the fundamental ideas, offering a starting point that could be refined further.
Cons of AI-Generated Storyboard
- Inconsistent Visuals: The images generated by ChatGPT lacked visual consistency, which can detract from the overall presentation and coherence of the storyboard.
- Surface-Level Understanding: While the AI provided a quick overview, it didn’t delve deeply into the details, missing some of the nuanced considerations that are critical in designing an effective user flow.
- Creativity Constraints: The AI’s approach tended to follow predictable patterns, limiting the potential for creative exploration and innovative solutions.
Both AI and human approaches offer unique strengths to the design sprint process. AI excels in speed and quickly capturing basic concepts, but it may lack visual consistency and depth. In contrast, human-generated storyboards, though more time-consuming and sometimes less polished, provide deeper insights and greater flexibility, fostering creativity and innovation. While AI isn’t likely to replace humans in the ideation and sketching process, combining the strengths of both can lead to well-rounded and creative results.
Stay tuned for Part 3
To wrap up this second part of our series on whether AI can supercharge a product design sprint, we’ve explored how AI can contribute to key activities like the Assumptions Board, Lightning Demos, and Storyboarding. We saw how AI can quickly generate ideas and visual concepts, offering valuable insights and saving time, but we also recognized the importance of human creativity and depth in the process. The combination of both approaches highlighted the potential for balanced and innovative results.
As we continue this exploration, the next post will delve into the final phases of the sprint: prototyping and testing. We’ll see how AI can assist in rapidly creating prototypes and creating an interview script.
