SaadBayahia.

The problem

Before buying or maintaining a vehicle, a non-specialist needs to know what to inspect, how to record the facts and when to seek professional help. Scattered observations, fault codes and documents make that decision harder.

Personal engine restoration, rally safety work and simracing made one problem tangible: recognising a symptom does not tell a non-specialist what to check next. Cardiag turns that field interest into a guided, documented inspection journey.

My role

Product Owner: frame the buyer, owner and workshop journeys; define the inspection backlog, prioritise the first usable local workflow and translate the scope into user stories and observable acceptance criteria. Technical choices support continuity in the field and a readable vehicle record.

The delivery

The responsive web app is available on desktop and mobile. A shared Expo codebase targets Android and iOS; native distribution is separate from availability in a mobile browser. Local inspections, verified-account sync and PDF reports connect field observations to the vehicle record.

Result

Public release since September 2026, with approximately ten users (owner-reported). The web app offers a 12-point quick inspection, 33 full checks and a catalogue of 201 makes.

Product decisions

33 checks, with a 12-point quick path

Seven inspection areas organise documents, engine, chassis, bodywork, equipment, road test and the final summary. The full 33-check path keeps that coverage explicit; the 12-point selection lowers the effort of a first pass. A quick inspection is not presented as equivalent to a complete one.

Local first, sync second

Keep recording usable beside the vehicle when connectivity varies. Inspection data is stored on the device; verified-account sync can then share supported records between devices. Photos remain local. The Gemini assistant requires an online connection.

Vehicle context before generic advice

A catalogue of 201 makes links the selected model, generation and engine to documented recurring faults. That context helps prioritise the next checks; it does not establish that a known fault is present on the inspected vehicle.

Prioritisation · MoSCoW

• Must: identify the vehicle, complete quick/full checks, preserve local observations and export an understandable PDF.
• Should: enrich the record with model-specific risks, guided symptom analysis and verified-account synchronisation.
• Outside the current integrated scope: Stripe collection, automatic purchases from parts suppliers and included third-party history reports. These remain configuration work or external services.

Acceptance criteria

• Each observation, comment and photo stays attached to the selected vehicle and inspection check; saved local data can be resumed.
• The selected mode exposes 12 or 33 checks. Progress represents completion, with unanswered checks remaining visible; it is not a reliability certificate.
• The PDF identifies the vehicle and distinguishes recorded observations, photos, completed checks and points requiring attention.

One vehicle, different needs

• Buyers: inspect a used vehicle step by step before purchase, identify points to check and keep a detailed inspection sheet.
• Owners: bring together observations, inspections and OBD readings to track the vehicle’s condition over time.
• Mechanics: structure diagnosis, document checks and prepare an intervention using the vehicle record and workshop workflow.

A complete, accessible inspection journey

Choose a 12-point quick inspection or 33 checks across seven areas. Plain-language steps help non-specialists know what to observe and record. Completion measures the checks entered, not vehicle reliability.

Model-specific knowledge, not a generic checklist

The extensive catalogue covers 201 makes and 2,697 model entries, with generations and engine variants. Documented recurring problems highlight symptoms to look for, affected components, usual mileage ranges and indicative costs, with guidance on the next checks for the selected vehicle.

Describe a symptom, build a diagnosis

The conversational agent, powered by the Google Gemini API, asks targeted questions and combines the vehicle context with symptoms, observations and fault codes. It helps narrow down likely root causes and prepare checks, with indicative repair costs and labour times refined through measurements and workshop inspection.

A detailed PDF, before purchase and over time

Generate a detailed vehicle inspection sheet in PDF format, bringing together vehicle identification, completed checks, recorded observations, photos and points requiring attention. Use it as a practical pre-purchase checklist or keep it with the vehicle record to preserve traceability of its condition and subsequent inspections.

A connected workshop workflow

• Client → vehicle → repair order → quote → recorded approval → intervention → invoice → recorded payment → maintenance reminder. Stock is consumed when work is completed.
• Web, Android and iOS share an Expo codebase. Local records work before sign-in; verified-account sync handles version conflicts. Photos remain on the originating device.
• Quotes and invoices can be printed or exported as PDFs. Customer approvals, supplier orders and payment records are currently manual; a PDF export does not process a payment.

Vehicle history, parts & payment

• carVertical — External history link
• Oscaro — External parts link
• Ovoko — External parts link
• AUTODOC — External parts link
• Stripe — Setup pending

History reports and parts sourcing currently use external links and manual records. Partner API access and Stripe activation remain configuration milestones.

Application foundation

• Expo — Web & mobile
• Firebase — Identity & sync

Conversational diagnosis · Gemini API