Introduction

Condor AI builds real-world intelligence systems by combining computer vision models, edge devices, and scalable software to help teams monitor and manage complex environments across smart cities, industrial automation, and fleet operations.

Role

User research, wireframes, prototype

Team

1 Designer (me)

Project type

Concept

Problem

Background

This project designs a web platform to manage AI vision devices across 200+ city intersections. Each smart camera runs edge AI models and connects to a cloud dashboard. The goal was to create a high-level monitoring dashboard and a detailed configuration screen for assigning, tuning, and updating AI models while tracking performance and events.

Primary User

Managing city traffic shouldn’t feel like debugging software.

Process

Research

The industry landscape was reviewed to understand common monitoring workflows and interaction patterns. This helped define user goals (monitor, respond, manage models) and business goals (scale, reduce cost, improve AI performance)

Alternative's of the wireframe

Opportunity Area

Make bulk actions feel deliberate than risky

Improve how alerts guide users from awareness to resolution

Reduce cognitive load while monitoring system health

Key task flows

These flows illustrate how operators monitor system health, respond to incidents, and safely configure AI models at city scale

Outcome

Opportunity 1: Make bulk actions feel deliberate than risky

  • Monitor deployment status (live/offline devices, alerts.)

  • Configure and assign AI models to selected devices

The dashboard provides a clear overview of system health across all devices, with a map highlighting areas with the most alerts to help operators act quickly. Complex data is simplified through visual hierarchy and color, making it easier to scan, understand, and take action at a glance.

  • Map view to show device health and issues across different regions.

  • Alerts requiring attention

Opportunity 2: Improve how operators understand and act on device health

The screen reduces mental effort by visualizing system behavior directly through live footage, model status, and categorized events, though further hierarchy could strengthen at-a-glance monitoring.

  • Models can be assigned, tuned, or updated

  • Real-time events show model output

  • Actions are separated from monitoring to reduce risk

  • Live camera view shows what the device sees

  • Detections are highlighted for quick validation

  • Active AI models are listed clearly

Opportunity 3: Improve how alerts guide users from awareness to resolution

Quick status tags showing device health, activity, and alerts at a glance

Alerts help operators quickly identify problems and understand their impact. Each alert provides context, affected devices, and clear actions to resolve the issue.

Operators can review what happened, see which devices are affected, and take action to resolve the issue from one place.

Map

Map view to show which areas are working well and which need attention.

Learning's & takeaways

01

Clarity matters more than adding features

02

Guide users through decisions instead of exposing raw system data

03

Alerts should guide decisions, not just notify problems