aiVA
AI advisor for maintenance scheduling
Company
Aitomatic
Position
Founding Product Designer
Year
2023
Market
APAC, B2B
Problem
Gas refinery downtime cost millions every hour. Maintenance planners at Petronas struggled with slow, reactive tools.
Solution
Led the design of aiVA, an AI advisor that delivered clear insights and adaptive, real-time recalibration for task scheduling.
Impact
Boosted operator confidence and decision speed
Shifted planning from reactive chaos to proactive control
Problem
Every downtime hour in gas refinery operations can cost millions. Maintenance planners at Petronas coordinated hundreds of interdependent tasks using traditional maintenance management tools. Yet those softwares overloaded them with data and lacked signals. In the mean time, for each operator, they need to make each decision right because every decision on what to fix and what can wait carried risk and cost.
Design principles
Planners told me: "We don’t need more data; we need explicit clarity. Tell us what to prioritize, why it matters, and what the operational and financial impacts will be."
These direct user requests became my core principles:
Provide immediately actionable insights.
Transparently explain reasoning behind every recommendation.
Enable learning from planner decisions to accelerate future planning.
Unlike traditional CMMS, the core of aiVA's functionality was transparency, predictive reasoning, and clear business logic behind each recommendation.
Design process
Alongside our head of product, I conducted contextual research with workflow mapping sessions with planners. We identified task-priority methods, critical trade-off factors, and areas causing planners friction.
From that foundation, I collaborated closely with our product-engineering team through concept design and prototyping iterations. Design sprints with validation by Petronas planners ensured aiVA matched real-world decision-processes and complex operational scenarios. Each iteration evolved beyond traditional reporting outputs or simpler AI indexing, explicitly focusing on robust predictive reasoning and clear visual prioritization.
What we built
aiVA was a proactive advisor for maintenance planners, first at Petronas, then more enterprises that have the same high-stake problems.
aiVA highlighted its recommendations by reasons and projected business impacts: financial, operational, and logistical, such as, "A valve defect at plant section 3 could reduce capacity by 50 MMSCFD within 12 hours, resulting in an estimated $100,000 in immediate losses." Such quantified reasoning far exceeded standard maintenance management systems and simpler AI tools which provided static retrievals or basic keyword-based advice disconnected from real-time predictive impacts.
aiVA performed real-time task recalibration, visualizing how each adjustment would impact the overall maintenance plan, production yield, projected downtime, and resulting financial risks. This allowed planners to adjust priorities within changing operational contexts.
To enhance long-term accuracy, we incorporated human-in-the-loop functionality. Maintenance planners could correct and provide reasons for AI, driving continuous learning.
Impact
Early testing showed measurable impacts like planners' recalibration time reduced by 50%, planners expressed more confidence and attributed to aiVA’s transparent reasoning.
This integration transformed Petronas’ maintenance processes from pure cost centers into important assets capable of driving production yields, efficiency, and profitability.
