CarX

Problem

Buying a used car here means trusting a stranger's account of a vehicle's past. Odometer tampering, undisclosed write-offs, outstanding finance and stolen-vehicle flags are all matters of public and commercial record — but that record sits fragmented across eNaTIS, SAPS, SAIA and the insurers, effectively out of reach of the person who needs it most.

What I built

The Challenge

Those sales run against roughly 180,000 new ones a year, on a national fleet of about 12 million. Only around 3 million of those carry insurance, so a large share of damage history never generates a claim record at all.

CarVertical is the closest thing to a competitor. It proved the model in Europe and the US: aggregate the fragmented record, sell the buyer certainty. Its South African page exists, but its depth on locally registered cars is thin — it is strongest on imports carrying European or US histories, which is not what most South African buyers are standing in front of on a Saturday morning.

The gap isn't the business model. It's local data.

Objectives

• Make a fragmented record legible to a non-expert in under a minute.
• Deliver a verdict that reads instantly — before any detail is read.
• Build for local reality — eNaTIS, SAPS, SAIA, Code 2 classifications, not imported equivalents.
• Prove it as something you can use, not a static mockup.

Decoding the incumbent

CarVertical launched in 2017 by selling a data product as consumer protection, and entered each market the same way: aggregate whatever records that country actually held, then treat every market as its own negotiation.

The instructive move was the US, where it became a listed NMVTIS-approved provider and unlocked legal access to state title records. The South African equivalent — formal RTMC and eNaTIS recognition — is the highest-leverage move any local entrant could make.

Defining the local checks

Six checks, each designed to pull from a South African source and matched to a real way buyers get caught out.

The core design problem

Vehicle history data is dense, technical and boring. The person who needs it is not a data analyst — they're standing in a driveway with a seller watching them, trying to decide.

So the verdict comes first, and everything else sits underneath it.

The traffic light

A physical traffic light on the results screen — red, amber, green — carries the outcome before a single word is read, with a score out of 100 beneath it and the detailed checks below that. The metaphor needs no explanation and no legend.

The example vehicle rates amber: five clean checks, one Code 2 write-off flag. Not a rejection, not a green light. Proceed with caution, and go look at these specifics.

Visual direction

Deep navy with a green accent — trustworthy without being generic, and deliberately more considered than a classifieds listing, since the product is asking to be trusted with a decision worth tens of thousands of rand.

Native, not wrapped

iOS conventions throughout: status bar, card hierarchy, sheet transitions, system-native motion. It should feel like something already installed on the phone, not a web app in a wrapper.

Coded, not clicked through

This is not a clickable Figma file. The prototype runs as real code:

• Real state management
• Directional screen transitions
• Expanding check rows
• A sequenced loading animation

A static mockup shows what a product might look like. This shows how it behaves.

How it was built

The idea is my own: the market thesis, the six checks, the traffic-light system.

The prototype was implemented with AI, which freed up the time to test whether the idea actually held up.

Current state

A validated concept with a workable prototype. The gap is real: CarVertical's local depth is thin and the South African records exist.

The route to viability runs through RTMC and eNaTIS partnership, which is a commercial and regulatory problem rather than a design one.

Next: put the prototype in front of used-car buyers and watch where their eyes land on the results screen. The traffic-light hypothesis is confident but untested. Then the drill-down and dispute flows.

Lessons

Research before pixels, even on a personal project. The competitive teardown produced the sharpest insight in the whole concept: the gap is not the business model, it is local data depth.

AI is a tool that shortens the distance. It will not find a market gap or decide that a traffic light beats a data table — that thinking is still the designer's. What it does is close the distance between having an idea and being able to show it.