Knowledge Base in aiKO
Preserve engineering know-how at scale
Company
Aitomatic
Position
Head of Product & Design
Year
2024
Market
APAC & US, B2B
Problem
Industrial manufacturing teams rely on senior engineers to diagnose production issues, validate fixes, and keep production quality stable. Across PepsiCo, RHI Magnesita, and Tokyo Electron, critical knowledge was scattered across manuals, PDFs, machine events, troubleshooting logs, past issue history, sensor data, and individual engineers' memory.
When engineers searched for answers, they had to connect a symptom with the current machine state, past similar issues, possible root causes, known fixes, and supporting evidence. When that reasoning was hard to capture, teams repeated diagnosis, fixes took longer to validate, and new engineers needed more time before they could troubleshoot with confidence.
My role
I led product vision and design execution for aiKO Knowledge Base, working with two frontend engineers, two backend engineers, and one designer.
I defined the product model, core workflows, knowledge structure, and AI answer experience based on research and technical scoping with customer engineering teams.
Product model
The key product decision was to structure the Knowledge Base around how engineers troubleshoot: symptoms, causes, fixes, machine states, documents, sensor readings, and the relationships between them.
I shaped the Knowledge Base around three principles:
Capture engineering knowledge as connected entities.
Ground AI answers in machine context and prior issue history.
Make expert reasoning reusable across shifts, teams, and sites.
In practice, this meant treating machines, components, symptoms, causes, fixes, documents, and sensor readings as connected objects in the same knowledge graph.
Knowledge Hierarchy
Entities
What we build
aiKO Knowledge Base provided an interface for capturing, accessing, and applying specialized engineering know-how.
Experts could document troubleshooting knowledge, link related concepts, and structure proven fixes through hierarchical relationships. The system connected explicit knowledge from documents with structured expert reasoning from previous troubleshooting work.
An engineer could start from a symptom, review the related machine state, inspect similar past issues, and use aiKO to generate a source-backed troubleshooting path.
Impact
Based on customer validation and workflow modeling with PepsiCo, RHI Magnesita, and Tokyo Electron, aiKO Knowledge Base was projected to reduce troubleshooting time by 25% and new engineer onboarding time by 40%.
The Knowledge Base gave aiKO a reusable foundation for capturing engineering expertise and applying it during production troubleshooting workflows.
