Valyant AI
Valyant AI develops a proprietary conversational AI platform for the restaurant, retail, and service-based industries.
Problem
Valyant AI developed a platform to address the robot's lack of complete accuracy, which involves human agents, known as auditors, who correct the AI's errors. However, the platform's engineering-driven design and chaotic construction have resulted in an inefficient and user-unfriendly experience for auditors. The platform is excessively colorful, has disorganized information, and provides a poor search engine experience, significantly increasing processing time for auditors.
With the COVID-19 pandemic causing labor costs to soar in the service industry, businesses are struggling to maintain profitability. Therefore, Valyant plans to reduce the number of auditors to help these businesses cut costs.
Solution
I devised a novel method for the POS to place orders through the platform, which involved streamlining the platform from three sections down to two. This optimization resulted in a reduction of over 70% in per-order audit performance time.
My Role
I worked as the sole product designer on the Holly Platform team, I was responsible for overseeing projects that involved both developer experience and content experience. I would love to chat about the full process for each project with you, so please don’t hesitate to get in touch.
Time
Sep 2022 - Jan 2023
Initial Research
As with any great design adventure, I started with the discovery phase of unexpectedly being trained as an auditor in the first few weeks in order to understand the stresses of order-taking. In addition to addressing the product’s goal to reduce the burden of labor costs, I learned to take fast-food orders as an auditor to help the robot correct the orders. After reflecting on several ordering experiences, I identified commonalities in the way that people place fast-food orders.
User Interviews
In total, I conducted 10 hours of qualitative interviews with 5 individuals. Throughout the interviews, I posed questions and delved deeper into specific functions and features based on their responses.
After the interviews, I made a report to include the research methodology used, the participants and the QA details, the findings of the interviews, and my recommendations based on the insights gained from the interviews.
Insights
Lack of a clear hierarchy and structure: It led to the system turning into a lengthy map with everything possibly available, offering an overwhelming number of options that proved challenging to navigate.
Out-dated visual design: A good UI design has the potential to simplify the product and ease the user's experience. However, the old product has the opposite effect, instead of reducing complexity and alleviating the emotional burden of the user.
Redundant features: a whole section of the transcript appears to lack a justifiable purpose.
Overview
Referencing the identified insights, I eliminated many complexities from the existing experience. Using a funnel approach to narrow down the food options created more efficiencies. For example, if you order one type of menu, you will only see that type of food. If you want to change the food category, then you would change the category of food. Essentially, a user sees what they need to see, not everything all at once.
Menu Section
The menu section on the right is essential for the whole platform. I calculated the average ratio for this part took 75% of the whole order processing time, so it’s important for me to reduce the complicity of this section in order to shorten the processing time.
I decided to implement an automated menu system. This would streamline the ordering process by allowing auditors to quickly select items from a simple menu with few clicks, reducing both the time and effort needed for auditors to place orders. Additionally, I recommend introducing advanced filtering options such as sorting by categories, frequency, and availability to help auditors find what they need more quickly. Finally, optimizing the search bar allowed auditors to input the first few letters and sort the options in seconds. These changes help reduced the average ratio of order processing time total by 70%.
Eliminate All Unnecessary Features
When it comes to designing and maintaining your website, it’s important to eliminate all unnecessary features. Especially in the AI industry, this is, even more important as you don’t want to overwhelm your users with a barrage of features. Focus on creating a streamlined experience for the user and make sure that each feature serves a purpose. This means taking the time to analyze each element and deciding if it will actually be beneficial or if it's just cluttering up the page.
After conducting user interviews and research, we discovered that transcripts offered no real value in a conversational AI platform. As a result, we made the decision to remove them. This required some convincing of stakeholders and the user training team. This brings me to my next point: the AI industry is still in its infancy, and many individuals continue to rely on outdated knowledge that may not always be the best solution. To truly succeed, we need to prioritize design thinking and innovation.
Iterations
We underwent multiple iterations and user testing during the redesign process to achieve the ideal balance. Our efforts resulted in a strong foundation that will enable the team to incorporate new features in the future while preserving the user experience and shortening the processing time.
All iterations should be the result of testing, on the surface level and under the hood. That's why it is important to measure success in terms of user feedback, raw data, and customer satisfaction. Though content changes may be slight from iteration to iteration, they should always be purposeful and measurable. The goal is to create a unified experience that customers can trust, by making sure that each step in the process works coherently.
Reflections
Feedback so far has been so positive from the company. 99% of my idea was very quickly adopted by the team, and I felt proud of the resulting changes from its initial chaotic state to a more polished system.
This project is amongst my proudest work as I identified a significant opportunity, collaborated with many multi-faceted colleagues, and delivered impactful changes to a product that would help the existing fast food experience. (I worked in an ice cream shop when I was a student, best experience ever :] )
The research process, particularly the user interviews, provided me with the most valuable insight from this project. I thoroughly enjoyed conversing with the auditors, as they were both friendly and professional. Although I initially scheduled an hour for each meeting, they consistently ran over. While collaborating with junior designers, they would typically rely on a questionnaire when engaging with users. However, I always advised against this method, stressing the importance of first having a comprehensive understanding of the product before engaging with users. Anyway, I will expand on this topic in a future blog post.