Sona
The platform

Sona is becoming the data layer for the U.S. service bay. The workflow is how we capture it.

Every voice dictation produces a structured record that did not exist before, captured at the one moment it can be: a technician inspecting a car against the OEM vocabulary that lives inside dealerships. Workflow is the wedge. The data layer is the company.

Structured at source · live today

One spoken sentence becomes a structured record. Stored, keyed, pushed to the DMS.

app.sona.work/orders/RO-4521
Listening...
Speak your inspection findings clearly
Captured recordWritten to Tekion
vinJN1BV7AR4FM...
opcodeBR39 · front brake service
inspection_itemfront brake pads
measurement3 mm · right front
recommended_actionreplace pads, resurface rotors
decline_reasoncustomer declined on cost
warranty_windowexpires in 90 days · 41,200 mi

DMSs store this and never refine it. Foundation-model vendors cannot capture it. It only exists at the moment of physical inspection, against opcode and OEM vocabulary the public web does not carry. Whoever builds the workflow surface that captures it at the source owns the data layer.7

Live across 28+ Tekion endpoints, bi-directional
The system, in three layers

Each layer is built on the capture beneath it.

Layer 1Workflow wedgeLive today

Voice clock-in/out, RO-scoped technician-to-parts messaging, voice 3Cs and story dictation, automated multi-point inspection. Everything writes back to Tekion bi-directionally.

Layer 2Per-VIN maintenance graphArchitected · ~6 mo

Tuples are already VIN-keyed. The cross-shop join is engineering, not data acquisition. One vehicle across visits and shops becomes a longitudinal record no DMS keeps.

Layer 3DMS displacement and data channelDestination

Sona becomes the operational surface. The DMS becomes a billing and GL backend behind us, and the captured corpus opens net-new buyers.

Why it compounds

Every turn of the loop runs in our favor. A 2027 entrant starts at zero.

  1. 1Workflow adoption captures structured data at the source
  2. 2Capture accumulates into a per-VIN maintenance graph
  3. 3The graph sharpens prompting, parse accuracy, and retention
  4. 4A better product wins more shops
  5. 5A denser graph unlocks buyers the DMS never had
What a new entrant has to beat

Three moats live today. Two more architected.

A

Structured-at-source data flywheel

Live

Every dictation improves opcode matching, parse accuracy, and MPI auto-fill. Not replicable from outside the bay.

B

Workflow lock-in

Live

Voice clock-in is the daily habit anchor. One daily habit anchors the rest of the workflow.

C

Bi-directional integration as moat

Live

Tekion's 28+ endpoints reverse-engineered. CDK middleware screen-scraping a 1990s AS/400 thick client. Months per integration. Plaid's playbook for dealerships.

D

Per-VIN cross-shop graph

Architected · ~6 mo

Unlocks the car-owner buyer. Local density of a few shops in a metro demonstrates it; 15 to 25 makes it defensible.

E

Data products

Architected · ~12 mo

Anonymized analytics for software providers and externals. Built once corpus density supports an underwriteable signal.

A 2027 entrant starts at zero on the corpus, zero on the 28+ Tekion endpoints, zero on the CDK middleware, zero on the in-bay research, and zero on the regulatory window that is open right now.

One corpus, four buyers

The same captured data sells to buyers the original spreadsheet never contemplated.

One captured corpus
1Today's buyer

Dealerships and shops

~14,000 addressable U.S. franchise dealers x ~$30K ARR is a ~$420M ARR ceiling at standard pricing.10

2Architected

Car owners

Pre-emptive maintenance, best-fit routing, recommendation approval from the phone, and a portable history that follows the vehicle.

3Architected

Software providers

Technician-workflow telemetry, repair-procedure data, parts and task frequencies, and anonymized maintenance history.

4Architected

Externals

DMV title enrichment, insurer risk pricing from observed wear, used-car VIN intelligence, and OEM recall and warranty analytics.

Only category 1 monetizes today. Categories 2 through 4 are architected destinations, precedented by Carfax, Verisk, and Tempus.

The playbook has precedent

Own the workflow, own the data, sell it back. This is a worn path, not a bet.

Scale AI
$14.3B into Meta
Physical Intelligence
$400M at $2.4B
Verisk
~$40B insurance-data analytics
Tempus AI
~$8B clinical genomics
Flatiron Health
$1.9B to Roche
Carfax
$5B+ on a shallower automotive substrate
Bloomberg LP
the terminal as the data business

Owning the structured data layer beneath a workflow is a precedented path, not a speculative one. Sona is at workflow and capture. The next buyer categories are precedented, not invented.12

How it sells

Two buyers. Two motions. One wedge.

The wedge is the same everywhere: voice clock-in captures the daily habit, and inspections and 3Cs follow. The motion that wins an independent shop is not the motion that wins a dealer group, so we run both in parallel.

Independents
1 to 3 shops
Buyer
Service Manager
Decision cycle
Days to weeks
Budget
~$5K/mo discretionary
Motion
Founder-led, walk into the bay
Franchise groups
15+ rooftops
Buyer
Corporate VP of Service plus Corporate IT
Decision cycle
6 to 12 months
Budget
Standardized pilot to rollout RFP
Motion
Enterprise. SOC 2 and MSA required.

Live in premium franchise bays today, including an Infiniti rooftop in Bellevue, with multi-brand exposure. In active conversations with one of the largest U.S. dealer groups on the enterprise motion.

The dealer-group funnel

One rooftop earns the group. We know which stage each is in.

  1. 01Single rooftop
  2. 02KPI proof
  3. 03Corporate sponsor
  4. 04Procurement and security review
  5. 05Group pilot · 5 to 10 shops
  6. 06Full rollout

The founder-led motion that wins independents fails at groups, so the enterprise path runs through a corporate sponsor and a security review before any group-wide pilot. Holding the corporate demo until the CDK middleware is live in production is intentional.

The buying decision

At group scale, it stops being a buying decision. It becomes an obvious-spend decision.

Today, per shop
Xtime~$3K
myKaarma~$5 to $6K
TruVideo~$1 to $2K
Podium~$500
Combined~$10K / shop / mo
With Sona, bundled
One platform$2,500 / shop / mo
Saved
~$7K / shop / mo
~$8.4M / yr
in vendor savings across a 100 shops group, per year13

That is before $8K to $20K/mo recovered productive capacity per shop. Pricing is $2,500/mo bundled = $30K ARR per 25-technician shop.

Want to see it in a real shop?