The doctrine behind every agent on this page is three phrases long: deterministic signals, AI delivery, human approval. Engines that can be trusted to run alone, run alone — on schedules, overnight. Agents that draft and suggest do so with evidence attached. Anything that moves money or commits the dealership waits for a human click. That division of labour is what makes automation something a dealer principal can actually sign.
The engines that run the night shift
- Service leads generate themselves — campaigns process the contract book and maintenance schedules and create reminder leads for every vehicle falling due, queue-ready for the morning's calls.
- Claims assemble from evidence — the claims engine sweeps completed jobs and builds claims from parts, labour and photos on the configured schedule.
- Stock suggests its own orders — replenishment projects parts demand from consumption and lead time into suggested purchase lines a manager verifies, not retypes.
- Nudges publish overnight — scheduled sweeps surface the dormant customer, the quote with no follow-up, the receivable past terms, the vehicle awaiting invoice — each nudge carrying its evidence, waiting in the queue at open of business.
The agents that draft, and ask
Sense runs as the person using it — a service advisor sees their vehicles, a sales manager their pipeline — and answers in plain language: "which declined red items are due back this month?", "which claims will reject on current evidence?", "compare drive-to-booking across consultants." It drafts the follow-up calls, the dealer nudges, the transfer proposals — and every write waits for approval. The daily brief opens the day with each person's numbers and the nudges ranked, computed deterministically, phrased by the model.
Build your own dealership agents
AI Studio lets the dealership build agents on its own model — grounded in your knowledge and records, bound to an execution account whose permissions set the ceiling, metered by budgets enforced before the model call. Scheduled agent workflows can watch the business and publish suggestions autonomously; acting on a suggestion is still a click, and every run is audited. The governance contract is public: how Sense is governed.
Common questions
What can run fully automatically in a dealership?
The deterministic engines: service-lead generation from contracts and schedules, claim assembly from completed jobs, replenishment suggestions, scheduled nudge sweeps and the daily brief. They run on schedules and put their output in a human's queue — the trustworthy kind of autonomy.
Can an AI agent take actions on its own?
Agents read broadly under the user's own permissions, and publish suggestions autonomously on schedules — but writes that change records or move money are approval-gated. The honest phrasing: an agent can fill your queue overnight; acting on it is still your click.
What questions can Sense answer for a dealer principal?
Anything the records can answer within their permissions: claims stuck past age, declined health-check items due back, scheme budget burn, consultant conversion, workshop WIP — grounded in the live data, with the working shown.
Can we build our own agents?
Yes — AI Studio supports custom agents with your own models and keys, grounded knowledge collections, sandboxed tools, budgets and full audit. Dealership-specific agents are configuration, not a vendor project.