Guide · AI emails for car dealers
AI emails for car customers: context = customer, car and deal
AI email writers are everywhere. The problem is that most of them know nothing about the customer, the car, the test drive, the trade-in, or what you agreed on Tuesday. The result is well-written spam. An AI email that actually moves a car deal knows the context: customer, car of interest, prior messages, trade-in, owning salesperson, next step. This guide walks through how AI emails work when built into the CRM rather than living as a separate microservice, with concrete templates and guardrails.
Last updated:
Short answer (TL;DR)
A generic AI writer (ChatGPT, Outlook AI) knows nothing about your customers, cars or pipeline. It writes nicely – but generically. An AI email in CARRUSLiNK pulls the customer card, car, lead source, last message and next step into the prompt, so the mail reads like a salesperson who remembers the customer. Salesperson reviews and sends, AI never sends on its own. Result: 3-5x reply rate vs. generic templates, without feeling like spam.
Why AI emails are often worse than no email
- ChatGPT mails without context sound like a press release from a consultant
- The salesperson spends 5 minutes copy-pasting name and car into the prompt
- The AI invents price, model year or trim because it has no data – and the customer notices
- Outlook's built-in AI doesn't use your CRM, so the customer history is gone
- Customer gets three auto-mails in a row from three different tools – sales, marketing, service
- Nobody can see in the CRM what was sent, when, and by whom (or which bot)
- Generic AI mails get 0.5-2% reply rates, and salespeople stop using them after 2 weeks
- Compliance risk: a hallucinated price or trim in an email can cost the deal and a complaint
How context-AI emails work in CARRUSLiNK
1. Salesperson opens the customer card and clicks 'AI draft'
From the customer card or lead, the salesperson picks 'Write draft with AI' and selects the type: post-test-drive follow-up, portal enquiry reply, offer draft, ready-notification, service contract renewal, or free text. The AI gets customer, car, last message and next step automatically.
2. AI builds the prompt with your data
Behind the scenes the system collects name, car (make, model, year, price, trim), test drives, trade-in, recent mails/SMS, lead source, owning salesperson, and your tone guide. The salesperson doesn't see the prompt – but the result is a mail that mentions exactly the car the customer looked at.
3. Salesperson reviews and edits
The draft is shown in the mail window with the salesperson's own signature. They can tweak tone, delete a sentence, add a personal detail, or send. The AI never sends on its own – the salesperson always has the final word.
4. Mail logs on the customer
When sent, the mail logs to the customer card with timestamp, content and 'generated with AI, edited by [salesperson]'. The next salesperson or manager sees the whole thread without guessing.
5. AI uses the reply as new context
When the customer replies, the AI has the reply as context next time. Follow-up doesn't repeat itself and doesn't contradict anything earlier – because it all lives in the same system.
6. Tone and guardrails per account
Under Account settings you set tone (formal/informal), length, language (DK/EN/SE/NO/FI) and no-go topics (price guarantees, delivery times you can't hold). The AI stays inside your guardrails, not OpenAI's defaults.
7. Pre-approved templates for high-volume
For mails sent many times (ready-notification, day-before-drive, offer-attached), you create AI templates the manager approves once. Salesperson clicks send, AI personalises per customer inside the approved frame.
8. Measure: reply rate, not open rate
We measure reply rate, drive-booked-from-mail and deal-from-mail – not open rate (it lies). Salespeople see their own reply rate on AI-mails vs manual – and tune tone based on data.
What you get from context AI mails
Personal mails in 30 seconds
Instead of 5-10 minutes. The salesperson spends time on conversations, not on rewriting the same eight times a day.
The AI knows the customer
Each mail references the actual car, the actual conversation, the actual next step. No hallucinated price, no 'dear customer'.
Consistent tone across salespeople
Junior and senior staff send mails that sound like the same dealership. No raw mail in 8 different styles.
Guardrails against hallucination
AI may only use data from the customer card. It does not invent delivery dates, warranties or prices. Deliberate.
Everything logs on the customer
Each AI mail logs with 'generated by AI, edited by [salesperson]'. Manager and next salesperson see the whole thread – no shadow inboxes.
Faster response = higher conversion
Industry benchmark: leads answered within 5 min convert 4-9x higher. AI drafts make hitting the 5-min window realistic, even on a busy Wednesday.
AI mails in CRM vs. ChatGPT/Outlook AI
| Area | ChatGPT / Outlook AI | AI mails in CARRUSLiNK |
|---|---|---|
| Knows the customer? | No, salesperson copy-pastes | Yes, customer card in prompt |
| Knows the car? | No | Yes – make, model, price, trim, stock days |
| Knows last conversation? | No | Yes – last 90 days of mails/SMS |
| Logs on customer? | No | Yes – 'AI draft, edited by X' |
| Hallucinates price/trim? | Often | No, only CRM data allowed |
| Measures reply rate per rep? | No | Yes – AI vs manual |
| Follows dealer tone? | Generic OpenAI style | Tone defined per account |
| Pre-approved templates? | No | Yes – manager approves once |
Generic AI is fine for writing a postcard. It is not fine for writing a mail about a specific car to a specific customer who was on the lot today.
Frequently asked questions about AI emails
See AI mails written on your own customers
30-minute demo. We import 3-5 anonymised leads and show how the AI draft is built from customer card, car and pipeline – and how the salesperson reviews and sends in 30 seconds.