Dealership operations software

Not a dashboard of averages. A worklist of specific, current conditions: this lead has had no contact attempt, this deal has been finance-pending past its normal window, this declined service recommendation was never followed up, this approval has been sitting unanswered.

Each condition is detected on a schedule, carries the record it came from, and closes when the underlying data changes — so the list reflects the store, not a snapshot someone exported.

Work falls through the gaps between departments and between shifts. A manager finds out about a stalled deal at month end. Service recommendations that a customer declined are never revisited. Accountability is discussed in a meeting rather than visible in the system.

The underlying issue is that the store's exceptions live in people's heads, and heads go home at six.

Detectors run tenant-scoped against real records and write durable operational signals with an explicit lifecycle. Signals are grouped for managers by department and severity, and each one links to the record that produced it.

Where a safe, specific action exists for a signal, the system proposes it: the proposal states the change, is validated against the acting user's role, and executes only on human approval. Executions are idempotent so an approved action cannot be duplicated by a retry.

Outcomes are then measured through department-specific attribution rules, with sales and finance results reported as assisted rather than caused. AI usage is tracked with model, latency, and cost telemetry so the economics of the assistance stay visible.

Signals that depend on outside data — messaging delivery status, DMS-sourced repair order history, lender status, or market valuation feeds — require the corresponding approved provider connection. Detection on data ShiftPilot AI already holds does not.

General managers work the daily exposure list. Sales and service managers own their department's signals. BDC leads see unworked follow-up. Dealer principals read approvals, overrides, and the audit trail.

Signals are only useful if the thresholds match how your store actually operates — what counts as aged, what counts as stalled, who owns which department's queue. Expect to tune those before expanding the detector set, and to assign ownership explicitly rather than by default.

Operations reads across the CRM, desking, and document workflows. For how the AI layer is governed, see dealership AI.