The support team's reality
Support teams operate in a state of managed overwhelm. Tickets arrive faster than they can be read. Priority is guessed, not calculated. Context is scattered across five tools.
Dispatch is a study in what happens when AI handles the triage layer — not the resolution, the triage.
AI that earns trust
The challenge is not technical. AI can classify tickets. The challenge is trust: will a support agent act on an AI's priority recommendation without understanding why it was made?
The study explores how to make AI reasoning legible without making it verbose.
The control room model
The strategic model is the control room: a space where operators have a complete, real-time picture of a complex system, and can intervene with precision.
Dispatch applies this model to support operations. The agent sees the full queue, the AI's reasoning, and the recommended action — simultaneously.
Graphite and signal lime
The visual language is industrial: deep graphite backgrounds, monospaced typography, signal lime for AI-generated content. The palette is designed to feel like a tool, not a product.
Every colour decision has a functional meaning. Lime is AI. White is human. Amber is escalation.
The triage architecture
The signature feature is the triage pipeline: a visual representation of how a ticket moves from arrival to assignment. Each stage shows the AI's confidence, the signals it used, and the human override point.
The system is designed so that agents understand the AI's reasoning before they act on it.
Real-time without noise
The prototype explores real-time ticket updates without the anxiety of constant change. New tickets appear with a controlled animation — visible but not alarming. Priority changes are communicated through colour shift, not notification.
The system is designed to feel like a live dashboard, not a fire alarm.
The reasoning surface
The most considered detail is the reasoning surface: the panel that shows why the AI made a decision. It is designed to be scannable in under three seconds — a list of signals, not a paragraph of explanation.
The study explores how to make AI reasoning feel like a colleague's note, not a system log.
What the study produced
The study produced a complete product design: information architecture, component system, interaction specification, and a working prototype of the triage pipeline.
The prototype explores how AI-assisted tooling can increase an operator's sense of control rather than diminishing it.
This is a self-initiated ZELQOR concept study — not a client engagement. Concept studies are published to demonstrate how we think, not to imply a client relationship. No revenue figures, conversion data, or business outcomes are claimed.