What we build

The foundation first. Then everything that runs on it.

A staircase, not a leap. Step one is owning the data; every layer above it gets easier because the foundation exists. Fifteen systems, five layers, all in production - explore them on the Stack.

Start here

/00  The Warehouse — data integration

The data already exists. Ownership is what is missing - it sits in other companies' systems, and the day a group leaves them it returns to step one.

# the staircase
1 DMS     → transactions, gross, service
2 CRM     → intent + identity
3 Website → first-party GA4 events
4 APIs    → recall, equity, income
5 Activate → mail, SMS, AI

one warehouse. one customer record.

1. DMS - the foundation, where money changes hands. Programmatic access to the group's own transactions means everything downstream ties back to real dollars.

2. CRM - intent and identity, tied to the DMS as one real customer across both. Without that connection, a customer's actual value cannot be determined.

3. Website - GA4 events exported raw and owned. Server-side tagging places the dealership in the middle rather than handing every click to Google and Meta.

4. Third-party APIs - owned enrichment rather than rented lists. Recall data direct from NHTSA is free; it is the same feed resold back to dealers as an outreach product.

5. Activation - audiences and suppression, done properly. Most dealers begin here, at step five, without the foundation, which is why a customer receives new-car marketing two weeks after buying one.

"A CDP is just a data warehouse with all the data in it. It's not some special product."

How it flows

Everything in. Everything back out — in the group's definitions.

DMSdeals · service · GL CRMleads · notes · history SERVICE CRMappointments · check-ins WEBSITEclicks · forms · first-party MANUFACTUREROEM portals · allocation MARKET + APIScomps · recalls · book values /00 THE WAREHOUSE one customer. one record. yours. TEXTS + EMAILoutreach, receipted DASHBOARDSone live picture ASKplain-English answers AUTOMATIONSwork that runs itself AI AGENTSthe data, safely AI PLATFORMfed real context

Six sources already paid for, one warehouse the group owns, and every output in its own definitions — texts, dashboards, answers, automations, agents, and the vendors the group keeps.

What's already in there

The answers are already in the database. Nobody has asked for them.

Ask an operator what the database is worth and the answer is consistent: we are scratching the surface. The money is not in acquiring more data.

The question nobody ranWhat the guess wasWhat the data said
What do we pay over on a trade?"About $2,000 a car"$300 - a $1.5M/year gap group-wide
Are we priced right in service?Set by feel, held for yearsMoney "just going to the bottom line" once measured
Was the deal structured right?Assumed cleanA couple thousand in gross, without charging the customer more

Real cases, on the record from operators on the Car Dealership Guy podcast. None required new data - only the ability to query what the group already held.

28%lift in service-appointment completion after a 22-store group unified its datavia Car Dealership Guy
40-55%email open rates after another group unified its records, against 1-3% beforeidentical list

That is the argument for the warehouse: these answers already sit in the DMS and service history, and cannot be asked across until they occupy one place.

"183 deals, in two minutes"

A principal asks how many 84-month deals the group wrote this year. The old answer was "give me a couple of days." Connected to its own data it is 183, in two minutes - and the next question, how many carried gap, is another two.

Intelligence // /01–/04

See everything, instantly.

One operating picture across every store, and a plain-English way to query it. A live P&L while the month is still happening. A CRM funnel cross-checked against actual deliveries. Ad spend next to outcomes.

Activation // /05–/08

Act on what the group owns.

Segments described in plain English, pushed to mail, SMS, email or the AI platform already in place - with no CDP intermediary. An opportunity engine gives every store a defensible reason to call. Recall outreach runs off the free federal feed.

Your vendors can only build for the average dealer. We hand the AI already paid for clean, ranked audiences and the real context it never gets - the last mile that turns a canceled-in-three-months tool into one that closes cars.

Can an LLM read the dealership's site?

If a large language model cannot access the website, $5,000 a month of optimization will not matter. Vespoke checks it, fixes it, and owns the tracking so the data comes back to the dealership.

Customer-facing // /09–/11

Own the touchpoints.

Trade-in, lead-capture and campaign pages on the dealership's own site, delivering clean leads to the CRM by ADF. The web-activity data stays home.

The same widgets that rent for $1,000-$2,000 a month per store, built to fit how the group actually operates.

Autonomous // /12–/14

Runs while you sleep.

The group number lands in every GM's inbox at 6am with no manual entry. Competitor inventory is crawled nightly. A governed agent layer lets approved AI read the warehouse safely.

Move it before you buy it

Order, hold, move or stop - day-supply, competitor pricing and inter-store transfer signals, so a group sells what it already owns before ordering more. The market side runs nightly today; the stocking-signal engine is in build.

The honest version

Three paths. Most operators have already lived two of them.

PathWhat actually happensEnds in
Rent everythingSound for the platforms. The long tail keeps stacking - per-rooftop pricing penalizes smaller stores, and the data stays in someone else's silo.A cost line that only rises
Build aloneThe code rarely fails. Adoption does. The tool works and sits on a server nobody opens.A science project
Build with an operatorProven in a live 16-rooftop group first. Built alongside the staff who use it, pilot-first, so adoption is designed in.An owned asset

The build-it-alone graveyard is the best argument against building. It's an adoption problem, not an engineering one.

You don't have to take my word for it

Listen to how dealers themselves talk about it.

"We should be getting more efficient over the years, and I think we've gone backwards."

Liza Borches, CEO, Carter Myers Automotive - on what the tool stack did to the cost of selling a car

"They're not built for your organization - they're built for everyone's organization."

Michael Kelley, GM, Bourne's Auto Center - on why he built his own software instead

"We've got the tech expense and we've got the headcount."

David Hult, CEO, Asbury Automotive Group - on tools that promised efficiency and delivered a second bill

"There's nothing to hold them accountable once they have your signature."

Kyle Coleman, CEO, Coleman Automotive Group - on long-term vendor contracts

Real, on-the-record quotes from dealership operators, said publicly on the Car Dealership Guy podcast. Not one of them is a vendor.

How we work

We build you an asset, not another expense.

Not SaaS, not a monthly fee. A durable good the group owns - code, data, systems - transferable or sellable with the stores.

Start with one thing

One use case, the data it needs mapped, proven in 30 days. No 18-month data project.

Ship every month

Something goes live in month one, rather than after a six-month wait.

You own it

The fastest path is often retiring a subscription already being paid; the savings stay with the group.

The terms, in plain English

The terms most operators look for, stated up front. Pilot one store first - proven on the group's own numbers before wider rollout. Month-to-month - no five-year term, no auto-renew trap. If it stops earning its keep, kill it. Quarterly review of the numbers together - it has to pencil, and Vespoke says so first if it does not. On exit the group keeps everything - the code, the warehouse, the data. That is what ownership means.

What we won't do

No per-rooftop pricing that penalizes smaller stores. No positioning of software as a way to reduce headcount - the stack exists to make staff more effective. And no removal of a vendor that is earning its keep: where the right answer is "keep the tool and actually use it," that is the recommendation given. The enemy is unaccountable spend, not vendors.