1. Inbound extract + draft first reply
- Value: high
- Difficulty: medium
- Risk: medium
Agent reads the inbound message, pulls vehicle / location / contact, drafts a first reply. A named person sends it. Recommended first build.
Resources
A sample AI Workflow Audit.
EXAMPLE / COMPOSITE. This is what the deliverable looks like — not a case study, and not a named client.
- Fixed scope
- Human approval
- Built around your tools
- Documentation you own
Situation
A multi-location dealer group. Inbound sales email, a shared inbox, an owner who still forwards the good ones, and a CRM that is only as current as whoever remembered to type.
Current workflow map
Public form or manufacturer lead → shared inbox → whoever is watching → optional forward to a closer → optional CRM row, often later, sometimes never → first reply from whoever has a minute.
Time-leak / bottleneck notes
Assumptions, not fake exact hours: the owner is in this pile most weekdays; first replies often wait until evening; CRM rows for the same lead show up hours to a day later. I would measure a two-week sample before I quoted a build, not invent a weekly-hour number here.
Opportunities
2. CRM draft after approval
- Value: high
- Difficulty: medium
- Risk: medium
Same extract writes a CRM note. Nothing saves until the reviewer says yes.
3. Location routing
- Value: medium
- Difficulty: low
- Risk: low
If the message names a store, draft the handoff to that closer. Still reviewed.
4. Manufacturer lead cleanup
- Value: medium
- Difficulty: medium
- Risk: low
Normalize the fields before a person looks. Do not auto-call or auto-text.
5. Leave the angry / legal / wholesale threads alone
- Value: n/a
- Difficulty: n/a
- Risk: high if automated
Those stay with a person. Automating them is how you get a story you do not want.
Rough weekly hours currently consumed
Owner attention on this pile: a range I would confirm from a two-week sample, not a marketing number. The useful fact is concentration — one person is the bottleneck — not a precise hour count.
Estimated implementation cost range
If this became a Business build after the Audit: typically $3,750–$7,500 for one scoped workflow. Audit credit −$1,250. Typical remaining $2,500–$6,250. 50% to start on that remaining balance.
Recommended first build
Inbound extract + draft first reply, with a named approver. CRM write as a second step once the reply path is boring.
Integration / access diagram
Inbox (read) → agent (draft) → approver (send / edit / discard) → CRM (write only after approval). No send permission on the agent. No unsupervised CRM write.
| Permission | Allowed |
|---|---|
| Can read | ✓ |
| Can draft | ✓ |
| Can send | ✕ |
| Can write after approval | ✓ |
Approval / risk plan
Customer-facing text waits. CRM writes wait. Money, credit applications, and anything that looks like a promise wait. If I cannot name the reviewer, I will not build it.
60–90 day automation roadmap
Days 1–14: first-reply draft in review. Days 15–45: CRM draft after the same approval. Days 46–90: location routing if the first two are boring. Stop if the reviewer is not actually reviewing.
Written Business proposal
In a real Audit this is a one-page quote: scope, range, what is out, and the credit math. This example is not a quote.
30-minute delivery / debrief
In a real Audit I walk the map, the scores, and the first-build recommendation. You can take the document elsewhere.
That is the shape. If you want one of these for your stack, the Audit is $1,250.
Get your automation plan — $1,250