AI that works inside the order flow — not a chat box bolted on the side

The code comes from an indexed library, not from the model — you can check it against the carrier.

It flags hazardous goods and asks for a UN number and MSDS. This is the dividing line — an AI built to look capable would have just handed you a code.

NMFC classification

Type a commodity in plain language and get candidate codes, the resulting class, and the reasoning. Misclassify once and the carrier reprices the load — this is where seasoned staff spend their judgement.

Leash: it may only pick from codes that exist in our indexed library — never generate one. Low confidence is flagged for human review.

Hazmat and lithium interception

Customers routinely book lithium batteries and power stations as ordinary freight, not knowing they are hazardous. On a keyword match the system refuses to self-classify and routes it into the hazmat flow instead — check the hazardous box, add a UN number and MSDS, and staff confirm whether UN3480 or UN3481 applies based on the actual product and packaging.

Leash: here the correct behaviour is to refuse. No self-serve code, and a UN number plus MSDS required — an undeclared hazmat does not cost a rebill, it costs fines, refusal, or an incident.

Address parsing

Customers send a blob — street, contact, phone, all mixed. Paste it in and it comes back as structured fields, pickup and delivery separated in one pass.

Leash: parsing only. Residential vs commercial is decided by the carrier API, not the model — and a human confirms before the order is placed.

Delivery-order extraction

A drayage delivery order carries a lot of scattered fields — bill of lading, vessel, ports, ETD/ETA, pieces, weight, CBM, multiple containers. Paste the text and the form fills itself.

Leash: it fills the form, it does not submit. Ops confirms first — one wrong container digit sends a truck to the wrong terminal.

Customer support

"Where is my shipment?" "When does it deliver?" — these dominate the support queue, and the answer is already in the system.

Leash: read-only, scoped to that customer's own shipments, grounded in system facts. No invented promises, no delivery guarantees. Data is redacted to the customer view — cost and markup never reach the model. Anything uncertain goes to a human.

Customer support

Customers ask directly in the group — "where's my shipment", "what do you need from me" — and the bot answers from real system data. Anything it can't resolve goes to a human, with a quick "got it, following up" in the group.

Leash: one group is bound to one customer — it only answers that customer's own data. Anything uncertain goes to a human; no guessing, no promised delivery dates.

Quoting

Paste a shipment description into the group — commodity, dimensions, weight, origin and destination — and the bot parses it, calls the real carrier rating APIs, and replies with the price and the carrier's full name.

Leash: this is a real quote, not an AI guess — it calls the exact same rating engine the website uses. Only the sell price and the carrier's full name are shown; cost and markup never appear.

Tracking

Order status changes — approved, carrier assigned, picked up, in transit, delivered — push into the group automatically. A daily digest summarizes today's scheduled pickups, and the bot flags it when a pickup or delivery runs overdue.

Leash: pushes fire only on events that actually happened in the system — nothing is announced early. Overdue alerts are computed from the actual scheduled dates, not a guess.

Order confirmation, BOL & labels

Once an order is approved and a carrier is confirmed, the system generates the BOL (plus a per-pallet label for LTL) and pushes the PDFs straight into the group — with a pre-pickup checklist: palletized and wrapped, labels facing out, a reachable contact, dimensions and weight matching what was declared.

Leash: the BOL and label are generated and pushed only once the order actually reaches that status — never early. The content pulls real order fields (carrier, PRO#, pickup window), not a generic template.

Proof of delivery

The moment a signed proof of delivery is uploaded, the system pushes it straight into the customer's group as a PDF, with a reminder to flag any damage within 48 hours.

Leash: it only pushes once the POD file is actually in the system — never announces delivery ahead of the paperwork. One push per document, no repeat spam.

Invoicing

The moment an invoice is issued with a PDF, the bot sends the amount, due date and the PDF straight into the group with a reminder to review and pay — no one manually chasing each customer.

Leash: only pushes invoices genuinely in "issued" status — drafts or estimates never go out. The one-group-one-customer binding means it can't land in the wrong group.

A fabricated code costs more than no answer

A language model will happily complete a plausible-looking code. Carriers bill against real ones — the moment it invents, the rebill is already in motion. So we retrieve from the real library first, then let the model choose among what actually exists.

Escalate on purpose

Every one of them carries a confidence check and a human hand-off. The system would rather say "not sure, please confirm" than hand you something that sounds certain and is guessed.

The model never sees what it should not

Order data is redacted to the customer view before it reaches the model. Cost, markup and upstream source never enter the prompt. If you run an agent network, this one matters — adding AI does not leak your buy rate.