The agent reads from your live data — fleet, bookings, customers. Not a copy and not a snapshot. When a customer pays an invoice, the agent sees it the second your dashboard does.
The agent never executes destructive actions on its own. Bookings, refunds, calendar holds, and customer messaging — all queue with a one-click preview. Refunds show the original payment, remaining refundable, and Stripe payment ID. Calendar holds surface any rentals that conflict with the proposed block before you sign off. You approve the draft, the agent ships it. Read-only queries run instantly; anything touching money, the calendar, or a customer waits.
Tuesday morning I asked which trailers were dragging utilization, got the answer in three seconds. The week before, that same question would've meant texting my partner and waiting until lunch. I don't open the dashboard for routine questions anymore — I just ask.
Anything you'd ask a senior employee about your business — utilization, revenue, individual customer history, payment status, maintenance schedules, fleet performance, booking detail. If the data lives in TrailerBase, you can ask the agent for it in plain English. Operators most often ask about utilization trends, who's overdue on payment, which trailers are sitting, and what last week's revenue looked like.
Yes — the agent answers questions about your customers because it has to. For the AI Agent specifically, customer PII (last names, emails, phone numbers, addresses) is stripped from prompts before they leave your account; first names are replaced with stable internal tokens that we rehydrate to real names before they reach you. The agent works with customer IDs, not contact information. Other AI features that touch customer data — review-response drafts, renter chat replies, license OCR — are being progressively migrated to the same redaction layer; we'll tighten this language as each one ships. Queries route through our infrastructure rather than directly to a model vendor, and we do not use your data to train models — yours, or anyone else's.
Both, but every outbound action shows you a preview first. The agent can draft and send reminders, create rentals, generate quotes and invoices, issue refunds, and block calendar slots for vacation, weather, or fleet downtime — when an action affects money, the calendar, or a customer, the agent shows you the exact details and waits for one click of approval before anything fires. Refunds show the original payment, refund amount, remaining refundable, and Stripe payment ID before you sign off. Calendar blocks show the trailer, date range, reason, and any rentals that would conflict — operator decides whether to proceed. Read-only queries (utilization, revenue, customer lookups) run instantly with no approval needed.
The agent reads from the same database your dashboard reads — numbers match what you'd compute by hand, no statistical inference layer between data and answer. If the agent isn't confident in an interpretation, it tells you and shows the raw rows. Every answer includes a one-tap link to the underlying records, so you can verify in a second.
Yes. The agent runs everywhere TrailerBase runs — desktop dashboard, tablet at the counter, phone in your pocket. Same agent, same context, same answers. Most operators ask their busiest questions from their phone between calls.
A chatbot is trained on FAQs and gives canned responses. This agent reads your live operational data and answers from it. Ask it about TR-1207 and it pulls that trailer's actual booking history, last maintenance check, revenue this quarter. It is not pattern-matching against a script — it is querying the system that runs your business.