AI-Assisted Smart Factory Operations

Overview: 

Modern manufacturing facilities depend on continuously operating systems that move materials between equipment, storage areas and production stages.

Operators, engineers and technicians need to understand equipment status, detect interruptions and make safe decisions without slowing production.

The existing experience distributed critical information across multiple screens, nested dialogs and data-heavy views. This increased cognitive load precisely when users needed clarity and speed.

My Role:

As Senior Product Designer, I contributed across the end-to-end product-design process:

• Discovery and contextual research
• Research synthesis and opportunity framing
• Product and UX strategy
• Information architecture
• Multi-device workflows
• Data visualization
• Interaction and visual design
• Enterprise design-system direction
• Human-AI interaction concepts
• Cross-functional design reviews

I collaborated with product managers, software engineers, factory-automation specialists, visited factory chambers and domain experts.

Rather than asking users only what features they wanted, I studied the sequence of information they looked for when beginning a shift, investigating an alert and escalating an issue.

The research led to six guiding principles.

Simplify

• Remove duplicate information and unnecessary interface noise.

Prioritize

• Surface critical equipment conditions, operational impact and required actions first.

Preserve Context

• Keep location, equipment and surrounding system status visible during investigation.

Progressive Disclosure

• Present a clear operational summary first, then allow users to inspect detailed telemetry and configuration when required.

Design for Mobility

• Adapt the experience for tablet use with larger touch targets, focused workflows and appropriate data density.

Keep Humans in Control

• Use automation to assist judgment without removing operator authority.

Where AI Should - and Should Not - Act

• A key product decision was not to apply AI to every operational task.

Deterministic System Responsibilities

• Safety interlocks, equipment-state tracking, compliance controls and execution boundaries remain governed by predictable system rules.

AI-Assisted Responsibilities- AI can help:

• Prioritize anomalies by operational impact
• Correlate signals from multiple sources
• Suggest possible causes
• Summarize supporting evidence
• Recommend next actions
• Generate shift-handoff summaries
• Support natural-language information retrieval

Human Responsibilities

Operators retain authority to:

• Approve a recommendation
• Modify the proposed action
• Reject or override it
• Escalate the incident
• Confirm recovery

Human-AI Decision Flow

• The system detects an abnormal condition.
• The AI layer correlates available evidence.
• The interface presents the recommendation, confidence and rationale.
• The operator reviews alternatives and operational impact.
• The operator approves, modifies or rejects the recommendation.
• The deterministic system executes only an authorized action.
• The outcome is recorded for audit and future learning.

I reorganized information around the equipment or device being investigated.

A selected device became the persistent context for:

• Current status
• Location
• Active alarms
• Related material
• Connected ports
• Recent activity
• Available actions
• Supporting diagnostic information

This device-centered model reduced the need to repeatedly move between unrelated system sections.

Unified Operations Overview:

A single operational view brought together equipment health, spatial context, prioritized alerts and active material movement.

Design decision: Present critical status and exceptions first while keeping deeper details available on demand.

Detail Panel:

Recommendations displayed:

• What the system detected
• Why it mattered
• Supporting evidence
• Confidence level
• Possible operational impact
• Recommended next action
• Alternative actions
• Approve, modify and reject controls

Design decision: Allocate additional screen space to explanation because trust and safe decision-making were more important than interface compactness.

Mobile Incident Triage:

Tablet and mobile concepts allowed technicians and supervisors to review urgent conditions away from fixed terminals.

Design decision: Reduce chart density and emphasize priority, location, status and immediate actions.

Shift-Handoff Support:

Structured summaries preserved the investigation context, actions taken and unresolved issues across shifts.

Design decision: Support automated drafting while requiring human review before handoff completion.

Enterprise Design System:

The experience required a reusable visual language capable of supporting dense enterprise information without sacrificing clarity.

The system included patterns for:

• Equipment and status cards
• Critical alerts and warning states
• Data tables
• Contextual side panels
• Operational charts
• Recommendation and confidence states
• Action confirmations
• Empty, loading and disconnected states
• Desktop and tablet layouts

Status was communicated through labels, iconography and hierarchy—not color alone.