82% Automated: What an AI Assistant Really Does for Your Dealership
Saturday, 4:40 pm. A customer wants to know whether the used estate car on your website is still available and whether he can come in for a test drive on Monday. Your sales team is busy, the phone rings into the void, the email gets answered Monday at noon. The customer? Inquired somewhere else on Sunday evening.
That exact inquiry costs zero staff time when a machine answers it. Not eventually — within three seconds.
The number that matters: 82% — measured, not promised
The AI industry is loud. Every second vendor promises “automation at the highest level” but mostly shows pitch decks when asked for proof. We prefer production data: at a leading mobility provider in the DACH region, our system has been answering 82% of all incoming customer inquiries fully automatically for months — across six channels, with response times down from 48 hours to 3 seconds. Read the full case study here.
That is the difference between a lab demo and a system serving real customers with real contracts.
What 82% means for a dealership — the honest math
Let’s be conservative. A mid-sized dealership receives 20 to 40 inquiries a day across website, phone and messengers: opening hours, “is that car still available?”, workshop appointments, test drives, financing and subscription questions, tyre changes. Say 600 per month at 5 to 8 minutes each:
| Item | Value |
|---|---|
| Inquiries per month | 600 |
| Time per inquiry | ~6 minutes |
| Staff time consumed | ~60 hours/month |
| That equals | a good third of a full-time role |
| Total cost of a service advisor | €4,500–5,000/month |
| Total cost of a receptionist | €2,700–3,500/month |
If an assistant takes over 80% of that, your team gets around 50 hours back every month — time that goes into sales conversations instead of “we’re open Monday to Friday, 8 to 6”. At a solution cost that is a fraction of your least expensive service role, this isn’t a marketing slide. It’s elementary school math.
And the part no spreadsheet shows: the assistant also works Saturday evening. The 4:40 pm inquiry gets answered at 4:40 pm — not Monday noon, when the customer is already sitting at your competitor’s desk.
What this looks like in practice
No portal, no app, no training. A discreet chat widget on your website — in your colors, with your logo:
- Frequent questions, answered instantly: a curated FAQ section shows your customers’ top questions, neatly sorted by topic. Every answer was reviewed before going live — the AI does not invent prices or commitments.
- Real conversations instead of click menus: anyone asking in their own words (“Do you have an estate under €25,000 with a tow bar?”) gets a real answer — in German, English, or the customer’s language.
- WhatsApp and Telegram included: customers write where they already are. Inbound WhatsApp conversations have been free and unlimited on Meta’s side since late 2024 — the most economical service channel ever.
- Handover to humans when it counts: price negotiation, complaint, edge case? The assistant hands over to your team — with the full conversation history.
- Feedback that improves the knowledge base: customers rate every answer thumbs up or down. Whatever doesn’t help goes on the review list.
Why not just “any chatbot”?
We tested the sales bots of major vendors ourselves. The result at one of the best-known: the simple question “what makes you different from others?” produced the same counter-question three times in a row — and the first reply to a German visitor came back in English. That’s the click-menu bot generation with a fresh coat of paint. Here is what that looks like to your customer:
The difference isn’t in the buzzword, it’s in the foundation: a curated, reviewed knowledge base of your dealership plus a language model that understands instead of guessing — operated on European servers, GDPR-compliant, without your customer data ever touching a US data center. What the AI doesn’t know for sure, it doesn’t say. What it’s not allowed to decide, it escalates to humans.
And yes, honesty is part of the pitch: an AI assistant doesn’t sell cars. It makes sure your salespeople talk to customers who already got an answer — instead of competing with your voicemail.
The next step
We are currently building the dealership package on exactly the system from the case study: web widget, WhatsApp, Telegram, curated knowledge base, transparent rental pricing instead of “price on request”. If you want to see what this looks like for your dealership — with your vehicles, your opening hours, your most frequent questions: talk to us. We’ll show you on your own example, not on slides.
