aiVA
AI Reasoning for Manufacturing
Overview: Empowering Field Engineers with Intelligent AI
In manufacturing, field service engineers are often under immense pressure to troubleshoot machinery and fix problems quickly. aiVA was designed to address these challenges by acting as a virtual advisor, helping engineers rapidly diagnose issues and access domain-specific knowledge in real time.
I led the design of aiVA, focusing on creating a tool that would not only improve efficiency but also enhance decision-making for engineers in high-pressure environments.
The Problem: High Stakes, Limited Time
Field engineers deal with several obstacles that hinder their ability to resolve issues efficiently:
Time Pressure: Engineers are required to fix machinery on-site quickly to minimize downtime, often under tight deadlines.
Complexity: Diagnosing equipment problems is often a manual, trial-and-error process, leading to wasted time and effort.
Lack of Accessible Knowledge: Engineers frequently lack immediate access to the specific knowledge needed to address complex problems.
We needed to create a system that could support engineers in quickly identifying the right solutions, with an intuitive interface that allowed them to focus on fixing, not searching.
Approach: Crafting AI that Thinks Like an Engineer
To develop aiVA, we began with thorough research into the workflows of field engineers. By interviewing teams at Petronas and Miura Boiler, we gained a clear understanding of the unique pressures they faced on-site. This research helped identify three core principles that guided our design:
Engineer-Centric UI: The interface had to mirror the way engineers think and work, minimizing cognitive overload during high-stakes problem-solving.
Solution-Oriented AI: aiVA needed to go beyond knowledge retrieval to offer practical, context-aware solutions based on the specific issue at hand.
Adaptable Design: The system had to be flexible enough to handle the varying use cases engineers face in diverse environments, from boilers to semiconductor equipment.
Execution: Designing Intuitive AI for Complex Systems
I created user personas to represent field engineers, taking into account their technical expertise and on-site needs. The design emphasized simplicity, focusing on clear, actionable diagnostics that were intuitive for engineers with varying levels of tech-savviness.
Our prototypes, developed with feedback from Petronas and Miura Boiler, were designed to streamline the troubleshooting process. By working closely with these clients, we tested aiVA’s reasoning capabilities and refined the UX based on real-world use cases.
Impact: Revamped Troubleshooting for Field Engineers
The deployment of aiVA made an immediate impact on field operations:
Engineers reduced time-to-repair, thanks to aiVA’s diagnosis and solution recommendations, improving first-time fix rates.
The project helped secure PoCs with clients like Petronas and Miura Boiler, resulting in $500K in revenue and attracting $500K in additional investments.
aiVA’s streamlined troubleshooting process eliminated the need for trial-and-error, allowing engineers to focus on delivering faster, more effective service.
Reflection: Crafting AI That Adapts to Each Engineer
This project taught me that good AI design is about enhancing human skill with precision. One aspect I’d refine further is the adaptability of aiVA, making it more responsive to engineers’ unique workflows.
I learned that it’s the details—how the AI communicates, how its recommendations are delivered—that define trust in high-pressure situations. In the future, I’d focus on refining these micro-interactions, ensuring that the AI feels like an intuitive extension of the engineer’s expertise.
Next work
