Hybrid intelligence
Dealership-specific machine learning finds the relationships people cannot review at dealership scale.
MIQ uses a governed set of machine-learning models to connect patterns across departments, time, transactions, handoffs, and market context. It ranks patterns by their potential relevance to gross, cash, capital, or risk, then gives the dealership the source context and operating path to act.
- Deterministic bound · source, fields, units
- Rule conditions
- Dealer decides
- Training boundary
- Guides attention · establishes no fact
- 01Authorized signals
Begin with dealership-authorized exports, source context, required fields, units, known rule conditions, and explicit data boundaries.
- 02Governed model-assisted discovery
Governed machine-learning models examine candidate relationships across departments, time, transactions, handoffs, and market context, then rank the opportunities that warrant attention. They do not establish truth.
- 03Deterministic corroboration
Where an implemented rule path applies, deterministic checks test what the source records can support while the potential economic effect, exclusions, missing inputs, and uncertainty stay attached for review.
- 04Evidence and confidence boundary
Confidence, exclusions, missing inputs, and source authority remain visible before a candidate can be treated as supported.
- 05Dealer authority
An authorized dealership role reviews the opportunity, chooses the next action, and approves the owner, deadline, and process.
- 06Accountable work
A supported priority can be routed to an owner, an authorized next action, a deadline, and a reviewable evidence state.
- 07Outcome and recurrence boundary
The result is checked against outcome evidence and later patterns. Model output guides attention, while the dealership decision and supporting evidence establish what happened.
- 01
Authorized signals
Begin with dealership-authorized exports, source context, required fields, units, known rule conditions, and explicit data boundaries.
- 02
Governed model-assisted discovery
Governed machine-learning models examine candidate relationships across departments, time, transactions, handoffs, and market context, then rank the opportunities that warrant attention. They do not establish truth.
- 03
Deterministic corroboration
Where an implemented rule path applies, deterministic checks test what the source records can support while the potential economic effect, exclusions, missing inputs, and uncertainty stay attached for review.
- 04
Evidence and confidence boundary
Confidence, exclusions, missing inputs, and source authority remain visible before a candidate can be treated as supported.
- 05
Dealer authority
An authorized dealership role reviews the opportunity, chooses the next action, and approves the owner, deadline, and process.
- 06
Accountable work
A supported priority can be routed to an owner, an authorized next action, a deadline, and a reviewable evidence state.
- 07
Outcome and recurrence boundary
The result is checked against outcome evidence and later patterns. Model output guides attention, while the dealership decision and supporting evidence establish what happened.
Whole-dealership reach
Models built around how dealership money moves.
This is not one generic score. MIQ examines economically distinct patterns across six connected dealership domains, then carries useful findings into owned work and outcome review.
- Sales, desking, and front-end gross
- F&I revenue, disclosure, and chargeback patterns
- Cash flow, funding velocity, and lender routing
- Service, warranty, parts, and fixed-operations recovery
- Inventory, floorplan burden, and market-pricing drift
- Financial structure and OEM recovery
Next decision
See how analysis becomes action.
Follow a cross-department cash problem from machine-learning detection to a named owner, dealership-approved process, supported value, and recurrence check.