Claude for Dealers / Learn / Build vs Buy
Strategy · Vendor Stack
Build vs Buy: Which Dealership Software Claude Can Replace and Which It Can't
The average store pays for dozens of software tools. Some of them are systems you could never replace. Some are a report and a login. AI changes the math on the second group only.
DMS · CRM · Inventory and pricing data · Desking and lender connections · E-sign and compliance · Payments · Phones
Reporting and dashboards · Content and descriptions · Internal calculators · Follow-up drafting · Q&A on your own documents · Vendor report reconciliation
Tools where you use 20% of the product · Add-on modules · Anything sold as "AI powered" that is a prompt and a login
Dealers should keep buying systems of record and anything with integrations, compliance liability, or uptime requirements: the DMS, CRM, inventory data, desking, e-sign, payments. Dealers can now build the layers that sit on top: reporting, dashboards, content generation, internal calculators, follow-up drafting, and question answering on their own documents. The test is five questions about data, liability, integrations, users, and who will own it. If nobody in the store will own a tool after it is built, buy it.
- Keep: anything that is a system of record
- Build: anything that is mostly a report, a document, or a draft
- A tool nobody owns will break and stay broken
- Build the simple version first. It tells you what to demand from vendors
- Never build anything that touches credit, payments, or e-sign
- Review the stack before each renewal, not after
Why this question is new
Until recently "build it ourselves" meant hiring developers, so the answer was always buy. Now an operator with Claude Code can produce a working internal tool in an afternoon. That does not mean you should fire your vendors. It means a specific category of spend, tools that mostly rearrange your own data or generate text, now has a real alternative.
The five-question test
Run every line of your vendor list through this. Two or more answers in the right-hand column means keep buying.
| Question | Points to build | Points to buy |
|---|---|---|
| 1. Where does the data live? | It works on exports of data you already own | It is the system where the data is created |
| 2. What happens if it is wrong? | A manager sees a bad number and asks about it | A customer, lender, or regulator is affected |
| 3. What does it connect to? | Nothing, or one file in and one file out | Certified integrations with OEM, lenders, DMS |
| 4. Who uses it? | A few managers | Every employee, or customers directly |
| 5. Who will own it? | A named person who wants to | Nobody. "IT" does not count |
What dealers are building
- 01ReportingDashboards from CRM, DMS, and inventory exports
- 02ContentDescriptions, emails, social, answer pages in your voice
- 03CalculatorsTrade, pay plan, ad budget tools you own
- 04KnowledgeQ&A on your policies, pay plans, and processes
- 05ReconciliationVendor reports checked against your own data
- Reporting: Build a Dealership Dashboard with Claude
- Content: AI for Dealership Marketing and AI for Used Car Managers
- Knowledge: Markdown Knowledge Files
- Reconciliation and rollups: Claude Cowork, Now Part of Claude
The vendor stack worksheet
Have Claude read last month's vendor invoices into one spreadsheet. Then add four columns by hand.
| Vendor | Monthly | What we use it for | Pile | Renewal |
|---|---|---|---|---|
| DMS | $____ | Everything | Keep | ____ |
| CRM | $____ | Leads, follow-up, desk log | Keep | ____ |
| Reporting add-on | $____ | One monthly report | Build | ____ |
| Description writer | $____ | VDP comments | Build | ____ |
| Reputation tool | $____ | Review replies only | Renegotiate | ____ |
We are not going to tell you a savings number. It depends entirely on your stack. The exercise usually surfaces a few tools where you are paying product prices for a single feature, and those are your candidates.
The real cost of owning what you build
- The tool does exactly what you want
- No per-rooftop, per-user, or per-lead fee
- Changes take minutes, not tickets
- Your data stays in your files
- Your team learns how the numbers work
- Someone has to own it and fix it
- Exports change and things break
- No support line, no SLA
- Knowledge walks out if that person leaves
- You are responsible for accuracy
Protect yourself against the last three. Keep every prompt, every rules file, and a one-page "how this works" note in a shared folder. A tool with documentation survives a resignation. A tool in one person's head does not.
Never build these
- Anything that collects or stores credit applications, SSNs, or payment data
- Anything that generates legally binding documents or disclosures
- Anything that quotes payments or rates to customers without your desking system
- Anything that must be up at 9pm on a Saturday with no one to fix it
Use the build as negotiating information
Even when you decide to keep a vendor, a working prototype changes the renewal conversation. You know what the feature is worth, because you know what it took to reproduce. Bring that to the table 90 days before renewal, not after the auto-renew date.
Common questions
Can Claude replace my dealership CRM or DMS?
No. Those are systems of record with certified integrations, compliance obligations, and every employee as a user. Claude works on top of them, using their exports or connectors. Do not attempt to build a replacement.
What dealership software can AI realistically replace?
Tools that mostly rearrange your own data or generate text: reporting add-ons, dashboards, description writers, simple content tools, internal calculators, and document Q&A. The five-question test on this page sorts any specific tool.
How much can a dealership save by building with AI?
It depends on your vendor stack, and anyone who quotes a number without seeing it is guessing. Build the worksheet on this page. Your candidates are tools where you pay full price and use one feature.
Who should own AI tools inside a dealership?
A named person who wants the job, usually an operator rather than IT: a marketing director, controller, BDC manager, or a sharp analyst. Give them time for it. See the 90-day rollout plan.
What is the biggest risk of building our own tools?
Abandonment. The tool works, the builder leaves or gets busy, an export format changes, and nobody can fix it. Documentation and a named backup owner solve this.
Build Real Tools You Own
No attendee is obligated to use any vendor technology. Two days of hands-on builds. Atlanta, January 5, 6, and 7, 2027.
Register Now →