Meridian is a fictional flagship dealership we built to show what an AI sales employee actually feels like — one that answers inventory questions, values a trade-in, runs the finance math and books a test drive at 2am, without a human awake. It's an interactive concept, honestly labelled. What is not pretend is the engine underneath: every number you'll see is computed live by real models, never scripted.
↑ That's the real experience rendering live inside this page — the conversation, the trade valuation and the payment math are all computing right now.
A luxury car buyer fills in an enquiry at 11pm. At most dealerships, nothing happens until a salesperson logs in the next morning — and by then the buyer has messaged two other showrooms. The gap between "I'm interested" and "someone replied" is where deals quietly die.
The industry data is blunt about it. Mystery-shop studies put dealer response rates in a harsh light — and the reward for closing that gap is well documented.
Figures from published automotive mystery-shop and lead-response studies (Velocify, 2026 industry reports). Shown here as industry context for the concept — not results from a named client.
Ask for "a white electric SUV under 80k" and it searches a real 18-car luxury inventory — Porsche, Mercedes, BMW, Audi, Lexus, Range Rover and more — matching body, fuel, colour and budget, then talks back about what it found.
Give it a make, year, mileage and condition and it returns a real valuation range — a depreciation model, not a lookup: brand tier, age decay and a mileage penalty all factor in, with an honest low–high band instead of a single fake number.
Buyers think in monthly payments, not sticker prices. Tell it a monthly budget and a deposit and it solves for the car you can actually afford over a 60-month term — the same payment math a finance manager would run.
It offers real open slots across the next few days, takes the booking, and scores the lead by intent, budget, trade and finance readiness — so the morning briefing to Sofia, the human closer, is ranked hottest-first. The night's work is done before the showroom lights come on.
Parses free-text enquiries into body, fuel, colour and budget filters, then queries a live 18-unit luxury inventory for genuine matches.
A depreciation model — brand tier, age decay and mileage penalty — returns a defensible low–high range, not a made-up figure.
Payment-to-budget math over a 60-month term: give it a monthly figure and deposit, it solves for the reachable car price.
Generates real open slots across the coming days and confirms the appointment inside the conversation.
Scores each lead on intent, budget, trade and finance signals so the human team gets a ranked, hottest-first queue.
Compiles the night's conversations into a ranked handover for Sofia — who talked, what they want, what to do next.
The scroll-story and the live chat call the exact same valuation and finance functions — the numbers can't disagree.
A pinned narrative of one night, 23:47 to 08:56, with the trade counter computed by the same engine driving the chat.
Meridian the dealership is invented — a stage to show the work on. We don't present it as a paying client, and there are no real customer names or numbers anywhere in it. That line matters to us.
The engine is real, though. The inventory search, the depreciation valuation, the finance solver, the slot booking and the lead scoring are working code you can poke at until it breaks. When we deploy this for an actual dealership, only the inventory, branding and CRM wiring change — the intelligence is already built.
Ask for a car that doesn't exist. Value a wreck. Give it a tiny budget. Every answer is computed live — see how far the concept holds up.
Open the experience