Workflow Overview

Freigent's reactive work queue, how freight events become AI-handled tasks you approve

What Workflow Is

Workflow is your work queue. Freight events happen all day, a broker sends an offer, a driver goes quiet, a POD doesn't arrive, a permit comes due. Freigent turns each one into a task, an AI agent analyzes it, and you get either a ready-to-approve recommendation or (if you have opted into autonomy) it executes itself and tells you what it did.

You work the queue at Workflow in the app. That one screen is where the back-office day happens: tasks arrive prioritized, each with the AI's recommendation, its reasoning, and one-click actions.

Workflow is not the booking agent itself, when a task needs live negotiation or document processing, workflow hands it to the specialist agents and tracks the result. It is the queue and the judgment layer over everything else Freigent does.

How It Works

  1. Something happens. An email lands, a deadline passes, a status changes, a cron sweep notices a load that has gone quiet. Each maps to a task type.
  2. A task is created in your queue with a priority, a deadline, and everything the AI needs to understand it.
  3. The AI works it. Before any model runs, deterministic guards check the facts: entity grounding, duplicate suppression, hours-of-service math, floor rates, blocked brokers. If the answer is already determined, you get it instantly with no AI run. Otherwise a domain-specific agent analyzes and drafts.
  4. You decide. Approve, edit then send, reject, or override. In AI Auto mode, tasks the autonomy gate clears execute on their own, everything else waits for you.
  5. It follows up. Tasks you don't answer re-fire at their deadline and escalate. Nothing silently expires while you are driving.

What Makes Workflow Different

  • It comes to you. You are not refreshing dashboards; the queue is fed by events, deadlines, and escalation sweeps that run continuously.
  • Every recommendation is grounded. The AI works from your loads, drivers, brokers, and rate data, not guesswork, and says no (CANNOT_RECOMMEND) when the facts do not support an answer.
  • Your decisions teach it. Rejects and overrides calibrate autonomy per task type, so the system grows into what you trust it with.
  • Money and safety stay yours. Invoices, safety finalizations, and similar high-stakes actions are human-held by design. See Automation Modes.

Next

AI Assistant

Ask about Freigent docs

Welcome to AI Assistant

Ask me anything about Freigent documentation, Booking Assistant setup, API integration, or platform features.

Install the Booking Assistant extension

Press Enter to send, Shift+Enter for new line