A single rate confirmation takes a dispatcher 3 to 5 minutes to read, type into the TMS, and double-check. Across 50 loads a day that is 2.5 to 4 hours of pure data entry, with mistyped ZIP codes, wrong commodity codes, and missed accessorials slipping through every week. TruckGPT closes that gap by reading the document and building the load in under 15 seconds at around 90 percent field accuracy.

How rate confirmations enter Datatruck

Rate confirmations arrive three ways: as a PDF attached to a broker email, as a phone-camera photo from a driver, or as a forwarded screenshot from a loadboard. TruckGPT handles all three formats from the same upload modal, with the rate confirmation preview rendering full-width so the dispatcher can sanity check the original document while the parsed fields populate to the right.

Inbox-connected accounts skip the upload entirely. AI Dispatcher watches the connected mailbox, surfaces inbound rate confirmations as a notification, and lets the dispatcher click once to create the load directly from the email thread.

What TruckGPT actually extracts

TruckGPT does not just OCR the document. It parses the structured trucking fields a dispatcher would have typed manually and populates the redesigned Load Creation V2 modal. The extracted fields include:

• Pickup and delivery stops with city, state, ZIP, and appointment times.

• Equipment type (FTL, LTL, multimodal, rail, air, maritime).

• Commodity description and weight.

• Load pay including line haul, accessorials, and fuel surcharge.

• Reference numbers (PRO, BOL, PO, broker load number).

• Special requirements: HAZMAT, TWIC, Team, Tanker, plus straps, tarps, pallet jack, blanket wrap, and dividers.

• Freight tags like protect-from-freeze and over-dimensional.

• Customer match with auto-complete on recent customers and MC Number auto-fill.

Why photo orientation and image quality no longer break it

Driver-captured rate confirmations used to break OCR pipelines the second the phone was held sideways. TruckGPT now reads work order images correctly regardless of photo orientation, so a rotated photo no longer produces partial or wrong extractions. CDL extraction is more reliable with first and last names mapped correctly, and the date parsing bug that occasionally swapped month and day has been fixed. Read the full breakdown in how TruckGPT turns a rate confirmation into a booked load.

What dispatchers see after the load builds itself

The build is not a black box. After TruckGPT runs, the dispatcher sees the full Load Creation V2 modal with every field populated and editable. Stops appear as collapsible cards the dispatcher can drag and drop to reorder. The rate confirmation preview auto-fits so the original document and the parsed load sit side by side, ready for the dispatcher to verify before hitting save.

If the broker sends a revised rate confirmation later, TruckGPT reads the new version and shows a diff against the existing load instead of forcing the dispatcher to edit fields one by one. The result is dispatcher in control, AI in the typing seat.

Edge cases TruckGPT handles cleanly

Three edge cases are where most OCR-only systems break and where TruckGPT differentiates:

The numbers, and where the time savings actually land

TruckGPT reads rate confirmations in under 15 seconds at around 90 percent field accuracy with 80 percent fewer rejected invoices reported by carriers using it daily. VIP Global cut rate-confirmation entry from 3 to 4 minutes down to 5 seconds, about 97 percent faster. The time saving compounds, because a 50-load day at 4 minutes per rate con is 200 minutes of dispatcher work that turns into 12 minutes with TruckGPT. A day in the life of a dispatcher using TruckGPT walks through that workflow end to end.

Where TruckGPT fits in the AI Dispatcher stack

TruckGPT is the entry point. AI Dispatcher then takes the built load and routes it across DAT, Truckstop, 123Loadboard, Parade, RXO, and Uber Freight in one search with RPM sorting. AI Updater handles the broker status emails after dispatch. The four AI products run on one Datatruck database, so the rate con TruckGPT reads is the same record AI Dispatcher books and AI Updater communicates against. See the full AI-native platform breakdown for how these pieces connect, and the integration footprint for the 100-plus partners on the same stack.

FAQs

How accurate is TruckGPT on rate confirmations?

TruckGPT extracts rate confirmation fields at around 90 percent accuracy on document quality typical of broker emails and driver photos. Carriers using it daily report 80 percent fewer rejected invoices because field-level errors that previously slipped into accounting are caught before save.

Does TruckGPT work on photos taken by drivers?

Yes. Recent updates fixed rotation handling and image-quality edge cases, so a driver-taken phone photo of a printed rate confirmation extracts correctly regardless of orientation. Partial or rotated photos that used to break the pipeline now parse cleanly.

What happens when a broker sends a revised rate confirmation?

TruckGPT reads the new document, compares it against the existing load record, and surfaces the field-by-field differences so the dispatcher can accept the changes in one step instead of editing fields manually.

Can TruckGPT handle multi-stop, HAZMAT, or oversize loads?

Yes. Multi-stop loads parse into collapsible stop cards in correct order. HAZMAT, TWIC, Team, Tanker, and oversize tags are extracted as chips on the load record so dispatch, safety, and compliance see them at a glance.

See TruckGPT read your own rate confirmation by booking a Datatruck demo.

Book a Datatruck demo