About ADIS
This project focuses on developing an AI-driven solution, ADIS (AI-Powered Downtime Intelligence Solution), aimed at minimizing unplanned equipment downtime in the manufacturing industry. By leveraging predictive maintenance, ADIS seeks to enhance operational efficiency, reduce financial losses, and foster a proactive maintenance culture.
Problem
Manufacturers face significant challenges due to unexpected equipment failures, leading to financial losses, operational disruptions, and reactive maintenance practices. On average, manufacturers experience 800 hours of downtime annually, resulting in approximately $80 million in losses per company. Unplanned downtime causes supply chain interruptions and decreased productivity. Current maintenance approaches are often reactive, addressing issues post-failure rather than preventing them.
Solution
We developed ADIS, an accessible AI-powered mobile application designed to predict equipment failures, provide real-time support, automate maintenance scheduling, and enhance data security. Utilizing IoT sensor data and advanced AI models, ADIS foresees potential issues before they occur. The application incorporates a chatbot feature for instant troubleshooting and maintenance assistance. It optimizes maintenance plans based on historical and real-time data to reduce unnecessary interventions. Ensuring compliance with global standards like GDPR and ISO 27001, ADIS protects sensitive information.
Process
User Research & Synthesize
Ideation
Sketching
Wireframing
UI Design & Prototype
Role
Sole UX/UI Designer + Developer
Project Details
Project Duration: 6 weeks
Platform
Mobile and Web
Tools
Miro
Figma
ChatGPT
Lovable + GitHub
1. Research
2. Ideation
3. Design
4. Reflection
1. Research
1.1 Secondary Research
For my secondary research, I analyzed industry reports, technical papers, and market studies to understand the financial and operational impact of unplanned downtime in manufacturing. The research showed that manufacturers experience significant losses due to inefficient maintenance strategies. According to an article in Forbes, unplanned downtime costs the manufacturing sector $50 billion annually, with an average of 800 hours of downtime per year per company. A separate study by Siemens Digital Industries highlights that predictive maintenance can reduce these losses by up to 40% through early fault detection and optimized scheduling.
Additionally, a report by McKinsey & Company stated that 80% of maintenance work is either reactive or unnecessary, meaning companies are either fixing problems after they occur or over-maintaining equipment, leading to wasted resources. By leveraging AI-powered predictive maintenance, businesses can transition from reactive to proactive maintenance, preventing failures before they happen.
According to an article in Manufacturing Today, manufacturers that integrate AI-driven maintenance solutions see 25% lower maintenance costs and 15% higher operational efficiency. These statistics reinforce the need for ADIS, an AI-driven predictive maintenance system designed to minimize downtime and optimize industrial efficiency.
1.2 Primary Research
I recruited participants through industry forums, LinkedIn, and direct outreach to manufacturing professionals. My goal was to understand how maintenance teams currently handle equipment failures and how they perceive AI-driven predictive maintenance solutions. Through a screener survey, I gathered insights from factory operators, maintenance engineers, plant managers, IT specialists, and manufacturing executives.
The survey revealed that 30% of respondents were factory operators, followed by 25% maintenance engineers, 20% plant managers, and 15% IT & data analysts. A smaller percentage of manufacturing executives and other roles participated, reflecting the decision-making hierarchy in industrial settings.
After conducting in-depth interviews and collecting qualitative data, I synthesized my findings into affinity groups, empathy maps, and user personas. These insights shaped ADIS’s feature set, ensuring that the AI system aligns with the real-world needs of maintenance professionals.
1.3 Screener Surveys
A total of 50 participants were surveyed, all of whom were professionals working in manufacturing, maintenance, and industrial operations. Participants were recruited through industry-specific forums, LinkedIn outreach, and direct referrals. The screener survey was designed to identify common challenges in equipment maintenance and understand the adoption barriers for AI-driven predictive maintenance solutions.
The primary goal was to pinpoint why companies struggle with effective maintenance strategies. The results showed that 50% of respondents cited the lack of predictive maintenance tools as their biggest challenge, while 45% reported high unplanned downtime costs as a major concern. Additionally, 40% faced difficulties in maintenance scheduling, 35% relied heavily on reactive maintenance, and 20% expressed concerns about AI implementation.
By analyzing this data, I was able to identify key user pain points and validate the need for ADIS, ensuring that its AI-powered predictive maintenance capabilities addressed real-world industry needs. The insights gathered from this screener survey informed the design decisions, guiding the development of an intuitive and efficient AI-driven solution for maintenance teams.
1.4 Interviews
I conducted five in-depth user interviews with professionals in the manufacturing and maintenance sectors to gain a deeper understanding of their challenges and experiences with equipment maintenance. The goal was to identify pain points in current maintenance workflows, understand how teams approach problem-solving and assess their openness to AI-powered predictive maintenance solutions.
Major Takeaways:
Many professionals expressed frustration with unpredictable equipment failures and the inefficiency of reactive maintenance approaches. Some of the most common themes that emerged included:
Lack of predictive insights: Maintenance teams often rely on outdated, reactive methods without early warning signs for failures.
High operational costs: Unplanned downtime leads to increased labor and parts costs, creating financial strain.
Ineffective maintenance schedules: Many organizations struggle with optimizing maintenance intervals, either over-servicing equipment or waiting until failures occur.
Data overload: While IoT sensors collect vast amounts of data, maintenance teams lack tools to effectively analyze and act on it.
Skepticism about AI adoption: Some professionals worry about AI accuracy and whether it can fully replace human expertise.
Limited training and resources: Many companies lack the in-house expertise needed to implement AI-driven maintenance solutions.
These findings helped shape ADIS by ensuring the platform addressed real-world challenges, particularly by offering actionable AI-driven insights, automated maintenance scheduling, and seamless IoT integration to assist maintenance teams rather than replace them.
1.5 Affinity Mapping
After conducting interviews, I documented insights and organized them into key themes to identify recurring pain points and user needs. By grouping these notes, I was able to create an affinity map that categorized the qualitative data into actionable insights.
The mapping process helped me uncover five core themes from the research:
Predictive Maintenance Gaps
Operational Efficiency Concerns
User Trust in AI
Data Overload & Accessibility
Integration Challenges
By synthesizing these insights, I ensured that ADIS directly addresses user concerns by offering real-time AI-driven maintenance recommendations, predictive failure detection, and intuitive dashboards that simplify data interpretation for technicians and plant managers.
1.6 Empathy Mapping
I developed a core persona, The Planner, from my interviews and synthesis. These helped us to better understand the end users and form the user Persona, which would be our lens into how users would interact with our product.
1.7 Personas
After a deeper look into the empathy map, the personas were formulated from a results review, idea, empathy & affinity mapping process. They formed the foundation for how we would go about generating ideas, evaluating design decisions, and a view through which we'd conduct spot functionality testing of our designs.
Kyle: a program coordinator who is extremely busy and needs to meet people to be more active, build community, have better habits, and be healthier
1.8 How Might We
The final piece of my research was synthesizing all gathered insights into How Might We (HMW) statements, helping to reframe user challenges as opportunities for innovation. Writing HMW statements allowed me to clearly define the most pressing user pain points while keeping my focus on problem-solving. These statements serve as a foundation for ideation and solution development in predictive maintenance and AI-driven maintenance optimization.
How might we help maintenance teams transition from reactive to predictive maintenance?
How might we simplify AI-generated insights so technicians can take action without extensive training?
How might we reduce unplanned downtime by providing real-time failure predictions?
How might we integrate AI-powered maintenance seamlessly into existing legacy systems?
How might we increase user trust in AI by making predictions more explainable and transparent?
How might we provide maintenance managers with a clear, data-driven view of asset health and performance?
How might we ensure maintenance schedules are optimized to reduce costs without compromising efficiency?
How might we help organizations overcome the challenge of data overload by delivering only the most critical alerts?
2. Ideation
After focusing on HMW’s, it was time to ideate and began sketching screens to find possible solutions to the user problems. All the research and data provided me with enough insights to start sketching to generate ideas and establish concepts to tackle the user’s problems.
2.1 User Stories
Before I could begin designing, I needed to identify the most essential functions for the AI-powered maintenance solution. To prioritize the most critical features, I categorized user stories into three levels of importance—high, medium, and low. This helped structure the necessary elements that needed to be implemented for a seamless user experience.
High - Must Have
As a maintenance manager, I want to receive real-time AI predictions about potential equipment failures so that I can prevent unexpected downtime.
As a technician, I want to have clear, step-by-step troubleshooting guides based on AI recommendations so that I can efficiently fix issues.
As a plant operator, I want to track asset performance and maintenance history so that I can ensure machines are running optimally.
As an executive, I want to see AI-driven reports with actionable insights so that I can make data-driven decisions to optimize costs.
As a maintenance planner, I want the AI to automatically suggest the best time for maintenance so that I can reduce unnecessary servicing.
As a user, I want an easy-to-use dashboard that simplifies AI predictions into clear recommendations instead of overwhelming data.
Medium - Nice To Have
As a technician, I want to receive mobile notifications when a machine is at risk of failing so that I can respond quickly.
As a maintenance manager, I want the ability to simulate different maintenance scenarios so that I can test AI recommendations before applying them.
As a plant operator, I want to filter AI-generated insights by priority so that I can focus on the most critical issues first.
As a decision-maker, I want the AI system to provide explanations for its recommendations so that I can trust the predictions.
As a maintenance planner, I want the AI to predict the estimated lifespan of machine parts so that I can plan ahead for replacements.
As a team lead, I want the AI system to provide performance insights on my technicians so that I can identify areas for improvement.
Low - OK Not To Have
As a technician, I want to compare maintenance trends across different facilities to see how my plant performs relative to others.
As a user, I want the AI system to gamify maintenance efficiency, awarding points for quick fixes and accurate troubleshooting.
As an executive, I want to forecast long-term AI-driven maintenance cost savings based on historical data.
As a maintenance manager, I want to export AI-driven maintenance reports in multiple formats for external audits.
As a technician, I want a chatbot interface where I can ask AI for troubleshooting advice in real time.
As a user, I want the AI to suggest third-party suppliers for replacement parts.
2.2 Sitemap
Now that I clearly understood the user's needs and goals, I structured the ADIS Industrial Monitor software sitemap to provide a smooth and intuitive navigation experience. The main objective was to ensure that maintenance teams, technicians, and managers could quickly access critical information without unnecessary complexity.
The sitemap was designed with hierarchical navigation, organizing features based on their importance and usage frequency. Given that predictive maintenance requires real-time monitoring, historical insights, and actionable recommendations, the software structure was developed to offer clarity, efficiency, and ease of use.
2.3 User Flow
Understanding how users navigate through ADIS was essential in optimizing the system for efficient monitoring and predictive maintenance. By mapping out key user interactions, we ensured that critical alerts, system diagnostics, and maintenance scheduling were intuitive and accessible.
2.4 Sketches
Sketching played a crucial role in shaping the user flows for ADIS, helping me visualize the navigation and interactions needed for real-time monitoring and predictive maintenance. I quickly realized that the interface needed to be clean, intuitive, and data-driven, with a focus on critical alerts and system efficiency.
2.5 Low Fidelity Wireframes
Using my sketches as a guide, I began working in Figma and brought those sketches into low-fidelity wireframes; doing so helped clarify how I wanted functions spaced out on screen and gave me a much better insight into how I could simplify and prioritize functions.
3. Design
Now that the software has a structured user flow, the focus shifted to refining its design. The goal was to keep the interface clean and minimal, ensuring that data visualization and alerts remained the primary focus. Given that ADIS is a real-time industrial monitoring system, I opted for a structured layout with a professional, intuitive feel, avoiding unnecessary visual clutter.
The interface prioritizes clarity, with a balanced use of color to differentiate between normal, warning, and critical alerts. The design ensures that technicians and operators can quickly interpret system health and take necessary actions without distractions.
3.1 Moodboard
I started by creating a moodboard to define the aesthetic direction. The goal was to establish a functional, data-driven, and highly legible interface while maintaining a sense of control and efficiency for users.
I chose a color scheme that reinforces urgency and clarity—greens for stable operations, yellows for warnings, and reds for critical alerts. The typography and iconography are selected to ensure high readability and quick scanning, crucial for industrial environments where time-sensitive decisions are made. The design of ADIS is built to instill confidence, ensure efficiency, and support real-time decision-making with a structured yet visually intuitive approach.
3.3 Style Guide
I developed a style guide to create a cohesive and intuitive user experience. The chosen colors reflect industrial precision and reliability, ensuring that data remains highly readable in complex dashboards.
The interface prioritizes clarity and efficiency, using strong contrasts and structured typography for quick information retrieval. Minimal shadows and color-coded alerts enhance usability, making critical alerts immediately recognizable.
3.4 High Fidelity Wireframes
With the use of my style guide, I began to build my high-fidelity screens. I focused on making the content easily accessible and easy to understand. By giving descriptive information, categories and intuitive flow.
4. Reflection
4.1 Conclusion
Designing ADIS was a valuable learning experience, allowing me to explore real-time monitoring and predictive maintenance from a UX/UI perspective. Throughout this project, I refined my skills in designing data-driven interfaces, balancing functionality with clarity to ensure that complex industrial insights were accessible and actionable.
One of the biggest challenges was the testing phase. Due to the constraints of this project being part of an academic course, we were unable to conduct formal usability testing with real users in an industrial environment. While testing is a crucial phase of UX design, factors such as limited access to actual industrial settings and time constraints prevented us from gathering direct user feedback.
Despite this, we made design decisions based on UX best practices, industry research, and heuristic evaluations to ensure usability. The interface was structured to align with standard monitoring dashboards, ensuring familiarity for industrial professionals. If given the opportunity in a real-world scenario, usability testing would focus on validating navigation efficiency, alert readability, and user interactions with predictive analytics.
This project reinforced the importance of designing for clarity and actionability in high-stakes environments. It also highlighted the need for iterative design, where testing and feedback loops play a critical role in refining interfaces. Moving forward, I aim to incorporate real-world usability studies to validate and enhance future iterations of ADIS, ensuring it meets the needs of industrial professionals effectively.