HadilBack to projects
build case studyPrototype to product concept

Speedo AI

Speedo AI translates computer vision signals into a workflow inspectors can review, trust, and act on.

My role

AI product UX, dashboard design, prototype build, workflow design

Case study
corrosionmedium

Overview

Speedo AI explores how AI-assisted inspection can support teams reviewing pipeline footage. The interface brings together video timelines, detection markers, severity states, and report generation in one review flow.

My Role

AI product UX, dashboard design, prototype build, workflow design

What I Built

  • A timeline-based inspection interface with scan markers and severity labels.
  • Detection review cards for corrosion, confidence, and location context.
  • A report-generation flow that turns reviewed findings into structured output.
  • A prototype dashboard for navigating video evidence and inspection status.

Key Features

  • Animated scanning timeline with detection events.
  • Severity and confidence indicators for faster triage.
  • Review states for accepted, rejected, and needs-follow-up findings.
  • Report-ready summaries for inspection handoff.

Product Decisions

  • Treat AI output as reviewable evidence, not a final answer.
  • Keep video, timeline, and finding details visible in the same mental workspace.
  • Use severity language that supports operational decisions without overclaiming certainty.

UX Decisions

  • Use color sparingly for risk, status, and attention.
  • Make the timeline the anchor so users always know where a finding occurred.
  • Design dashboard density for repeated professional use rather than a marketing-style demo.

Why I Built It

Inspection workflows are high-stakes and information-heavy. I built the concept to explore how AI can reduce review friction while keeping human judgment central.

AI-Assisted Workflow

  • Used AI to scaffold dashboard components and simulate inspection data states.
  • Explored alternate report formats before narrowing to a review-first workflow.
  • Refined interaction states manually to keep the experience calm and trustworthy.

Tools / Stack

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • Computer vision UX patterns
  • AI coding tools

What I Learned

  • AI interfaces need strong evidence trails.
  • Domain workflows benefit from restrained motion and clear state language.
  • The best AI product moments often happen around review, correction, and handoff.