Case Study
Buying the Right Car
Vehicle replacement research, three-engine AI synthesis.
Problem
A UK driver needed to replace a Dacia Jogger with a larger vehicle. The requirements were specific: significantly more cargo space for a dog, diving gear and computer equipment; frequent 500-mile motorway runs between two homes; petrol or self-charging hybrid only, because one home is off-grid with no mains charging available. Value-conscious, open to mid-range pricing.
She had a brand loyalty question too: was there a Dacia large enough to replace the Jogger, or did she need to look elsewhere?
What happened
The question was put to three AI engines simultaneously, each assigned a different role: one to research official specifications and confirm what the Dacia range actually offered; one to analyse the broader market and assess practicality for this specific use case; one to calculate real-world fuel economy and running costs.
All three engines independently converged on the same recommendation: the Skoda Superb Estate. Named What Car? Estate of the Year 2026, it offers 690 litres of boot space, an optional height-adjustable flat load floor suited to repeated heavy loading, genuine motorway refinement, and a mild-hybrid petrol engine delivering over 400 miles per tank. It undercuts its nearest equivalent, the VW Passat Estate, by approximately £5,600.
The Dacia question got a straight answer: the Bigster is the only meaningful upgrade within the range, but it loses the Jogger’s seven-seat flexibility and offers less cargo space than the Superb or the Skoda Kodiaq. If significantly more space is the priority, the step outside the Dacia range is the right move.
On the hybrid question: self-charging full hybrids require no mains connection and work entirely through the engine and regenerative braking, making them fully compatible with an off-grid home. For motorway-dominated use specifically, a full hybrid saves approximately 15% on fuel over petrol — real, but modest. The financial payback on the hybrid premium at sustained motorway speeds requires high annual mileage. The Superb’s mild hybrid is standard equipment rather than a costly option, which is why it sits well in this brief.
One engine disclosed that it had encountered prompt injection attempts — hidden instructions embedded in car review website content designed to manipulate AI behaviour — and had ignored them. This is worth noting as an example of responsible AI operation under adversarial conditions.
Outcome
A complete purchase recommendation with confidence ratings, a comparison table of eight vehicles, ruled-out options with reasons, a practical note on the 500-mile run, and a portable prompt for independent verification — all from a single structured research session.
The user came away knowing which car to buy, why, and what to look for on a test drive.
Why it matters
Car buying research is time-consuming, opinionated, and full of content designed to influence rather than inform. The three-engine approach cuts through it: each engine works from the same brief but approaches it differently, and convergence across independent outputs is a meaningful signal of reliability.
The prompt injection disclosure illustrates something important about how this methodology works. The engine flagged the manipulation attempt and noted it in its output. That kind of transparency is not available from a standard web search.
A detailed write-up of this case — including the comparison table of eight vehicles, the ruled-out options with reasons, and the portable prompt for independent verification — is available as a PDF.