Quick answer: Freight forwarding automation tools break into seven jobs, not seven brands: inbound email intake, rate procurement, quoting, shipment creation, documentation, tracking, and billing handoff. Point tools go deep on one stage — Sedna for email, Raft for documents, Wisor for quoting. AI-native TMS platforms span several — Zavin, FreightSuite, LogBot. Buy by the stage that costs you the most hours, not by the vendor with the loudest demo.
Most guides to freight forwarding automation are vendor lists. That format is easy to write and nearly useless to an operations director, because it answers a question you did not ask. You do not wake up wanting to evaluate eleven platforms. You wake up because quotes take four hours, a rate request to a Shanghai agent has gone unanswered for two days, and someone is retyping a booking confirmation into the TMS for the second time this morning.
So this guide is organized the way the work is organized: by what gets automated. Walk the lifecycle stage by stage. For each one, what the manual version actually costs, which tools exist, and what "good" looks like when it is working. The vendor table comes at the end, where it belongs — after you know which column matters to you.
(Disclosure: Zavin is our product, an AI-native TMS with a freight CRM and pricing automation. It covers intake through tracking on one data spine. It does not file customs entries and it does not run an accounting ledger, and we say so plainly at the two stages where that matters.)
Key takeaways
- Freight forwarding automation breaks into seven stages, not seven brands: point tools go deep on one — Sedna for email, Raft for documents, Wisor for quoting — while AI-native TMS platforms such as Zavin, FreightSuite, and LogBot span several stages on one data model.
- Automate inbound email intake first: roughly 80% of freight forwarding runs on email, and every downstream stage is starved by slow or lossy intake.
- Rate procurement is the stage almost no automation tool touches; Zavin adds an outbound loop that sends structured rate requests to agents and carriers, chases the ones that go quiet, and compares the replies side by side.
- Zavin handles roughly 50% of routine email end to end on a trained desk and returns a sent quote in under 30 seconds when rates are on file, with up to 85% less RFQ processing time.
- Raft reports processing more than 130,000 documents a month; CargoWise and Descartes carry native customs filing, which the AI-native cohort has not matched.
- Zavin publishes pricing at $125 per seat per month with every module included and goes live in under 14 days, syncing bi-directionally with CargoWise and Magaya.
The seven stages, at a glance
| Stage | The manual bottleneck | Tool category | Maturity |
|---|
| 1. Inbound email and intake | Triage latency; threads without an owner | Freight email AI; AI-native TMS | High |
| 2. Rate procurement | Chasing agents by hand; rates that expire in an inbox | Rate management; outbound rate loops | Low |
| 3. Quoting | Rebuilding the quote; margin applied from memory | AI quoting; TMS quote modules | High |
| 4. Shipment creation | Re-keying the won quote and the booking confirmation | AI-native TMS; carrier EDI/API | Medium |
| 5. Documentation | Reading a PDF and typing it into a form | Document AI; customs-native TMS | High |
| 6. Tracking and updates | "Where is my container?" answered one email at a time | Visibility platforms; carrier feeds | High |
| 7. Billing and AP/AR | Reconciling the carrier invoice against the quote | TMS accounting; AP automation | Medium |
Stage 1 — Inbound email and intake
The manual version. One shared address — quotes@ or ops@ — carries spot RFQs, booking confirmations, SI and VGM reminders, arrival notices, invoice disputes, and carrier marketing. A human reads all of it in arrival order. The cost is not the reading; it is the latency and the loss. An RFQ that lands behind eleven arrival notices reaches pricing after lunch, and the shipper has already booked. Around 80% of freight forwarding runs on email, which makes the inbox the widest and least instrumented part of the operation.
The tools. Sedna is the established email management platform built specifically for freight — shared-inbox discipline, tagging, and routing at high volume for maritime and forwarding teams. LogBot takes the same problem from a different angle with phone AI alongside email AI and an AI dispatcher; if a real share of your bookings still arrives by telephone, that is a gap most of this category leaves open. AI-native TMS platforms including Zavin and FreightSuite read the inbox as the entry point to the rest of the workflow rather than as a standalone product. Outlook and Gmail rules are the floor: they match keywords and forward mail, which is not the same as understanding a thread.
What good email intake automation looks like
What good looks like. Classification into freight categories, not generic labels — a spot RFQ is distinguishable from a rollover notice and a VGM reminder. Extraction of POL and POD, equipment, chargeable weight for air, Incoterms, and ready date, from the body and the attachments. A named owner and a clock on every classified thread. And critically, the extracted fields landing as a structured record the next stage can price, not a summary in a sidebar someone still has to retype.
Run it in draft-and-approve mode first and widen what sends automatically as accuracy proves out. Zavin handles roughly 50% of routine email end to end on a trained desk, with the operator keeping approval control on anything customer-facing. The mechanics are covered in depth in AI email automation and the deployment playbook in freight forwarder email automation.
Stage 2 — Rate procurement
The manual version. This is the stage almost no automation tool touches, and it is where the hours go. An RFQ arrives for Nhava Sheva to Rotterdam. Nobody holds a valid rate on that lane. So an operator opens a fresh email, BCCs six agents and two carriers, and waits. Three reply within the day. Two need chasing on Thursday. One never answers. The replies arrive in four different formats — a PDF tariff, a pasted table, a line of text, an Excel attachment — and get copied into a comparison spreadsheet by hand. The winning rate then lives in one person's sent folder and expires unread.
The tools. Most of what is sold as rate automation is rate management: storing, versioning, and applying rates you already hold. TMS platforms including CargoWise, Magaya, and GoFreight all have rate modules of this kind. Carrier APIs and digital marketplaces cover contracted and spot ocean rates where the carrier exposes them. Wisor works on AI quoting and rate automation with TMS integration and is the closest focused comparison in the market. The structural gap across the category is the same: these tools assume the rate exists somewhere already.
What good rate procurement looks like
What good looks like. Two things. First, a rate database that builds itself from traffic already flowing through the inbox — every agent quote and carrier tariff that arrives by email becomes a structured, searchable, validity-dated record owned by the company rather than by whoever received it. That is the job of rate intelligence. Second, an outbound loop for when nothing valid exists.
That second half is where Zavin differs from the rest of this category. Most AI quoting tools price from rates you already hold. Zavin does that — and when no valid rate exists for the lane, it goes and gets one: sending structured rate requests to your agents, partners, and carriers, chasing the ones that go quiet, parsing each reply as it lands, comparing them side by side on carrier, transit time, validity, and total cost, applying your margin rules, and returning a branded customer quote. The whole multi-party exchange runs as one automated conversation instead of a person working a mail-merge.
Stage 3 — Quoting
The manual version. The rate is found. Now someone rebuilds it as a quote: opens last month's template, retypes the lane, adds THC and BAF, applies a margin from memory or from a rate card nobody has updated since March, exports a PDF, writes a covering email, and sends. Then the follow-up gets scheduled in their head, which means it does not get scheduled.
The tools. Wisor is the focused AI quoting specialist. FreightSuite positions itself as an AI-native FMS covering quote to cash across ocean, air, road, and customs. Every full TMS has a quote module, but they are template-driven and human-entered: the system stores the quote, it does not compose it. Zavin's pricing and quote automation sits on the same spine as the intake and rate stages, which is what makes the handoff between them free.
What good quoting automation looks like
What good looks like. Margin applied as policy rather than memory, with floors that hold when the pricing manager is on a flight. A branded quote the customer can read without decoding. The follow-up scheduled at the moment of send, on a cadence you choose, so silence is the one outcome the workflow refuses to accept. And win/loss logged against the lane, so next quarter you know which trades you are actually competitive on.
The numbers Zavin measures here: an RFQ becomes a sent quote in under 30 seconds when rates are on file, with up to 85% less RFQ processing time. Pilot teams have seen a +6 percentage point lift in win rate once quote speed and follow-up discipline are both automated.
Stage 4 — Shipment creation and booking
The manual version. The quote is won. Someone opens the TMS and retypes what already exists in the quote — shipper, consignee, lane, equipment, Incoterm, charges. The carrier booking confirmation arrives as a PDF two days later and gets retyped again: booking number, vessel, voyage, ETD, ETA, container numbers. The same facts are entered twice or three times, and each pass is a chance to transpose a container number.
The tools. This is the traditional TMS's home ground — CargoWise, Magaya, and GoFreight all manage the job file well, and if operational depth across many countries is your requirement, that is where to look first (our full TMS comparison covers that decision). Carrier EDI and API connections automate the booking itself where the carrier supports it. Raft automates the lifecycle from booking through documentation and customs. AI-native platforms create the shipment from the document rather than from a form.
What good shipment creation looks like
What good looks like. The shipment object is generated from the artifact that already exists — the won quote, the booking confirmation — with zero manual data entry, and it lives in one place that both sales and operations read. If the ops team's shipment record and the sales team's quote record are two different objects, you have automated the typing but kept the reconciliation. Shipment management in Zavin creates the file from the booking confirmation and keeps the commercial record attached to it.
Stage 5 — Documentation
The manual version. HBL and MBL, HAWB and MAWB, commercial invoice, packing list, certificate of origin, arrival notice, pre-alert, delivery order. Most of these arrive as PDFs or scans and leave as forms someone filled in by reading the PDF. This is the purest form of the data entry problem, and at volume it is the largest single labour line in a forwarding back office.
The tools. Raft is the clearest specialist here: built for forwarders and customs brokers, spanning booking through documentation and customs clearance, with document processing as its centre of gravity at reported volumes above 130,000 documents a month. Freightmate concentrates specifically on document processing and data entry. CargoWise and Descartes carry native customs filing across major markets, which is a different capability from extraction and is not something the AI-native cohort has matched.
What good document automation looks like
What good looks like. Extraction with an explicit confidence threshold, and a human queue for everything below it — a system that guesses at a chargeable weight is worse than one that asks. Mismatch detection across documents: shipper name against the account, gross weight against chargeable weight, container count against B/L lines, quantity against the commercial invoice. And the extracted data landing in the shipment record rather than in a folder, because extraction that terminates in a PDF on someone's desktop has moved the work, not removed it.
Honest scope: Zavin issues documents against the shipment record, but it does not file customs entries. If your bottleneck is entry volume or brokerage compliance rather than commercial speed, Raft or a customs-native TMS is the sharper tool and we would rather you knew that before the demo.
Stage 6 — Tracking and customer updates
The manual version. "Where is my container?" A coordinator opens the carrier site, finds the booking, checks the milestone, switches back to email, and writes a reply. Multiply by every open file and every customer who checks twice a week. Worse is the inverse: a container rolls at transshipment and nobody notices until the consignee calls, which turns a schedule problem into a relationship problem.
The tools. Carrier EDI and API status feeds are the raw material. Descartes' Global Logistics Network and the dedicated visibility providers aggregate them. Every TMS has a milestone screen, though a screen only helps the person looking at it. Customer portals shift the work to the customer, which some shippers accept and most do not.
What good tracking automation looks like
What good looks like. Milestones pulled automatically by EDI and API rather than by a person refreshing a carrier page. Exceptions — a rolled container, a blank sailing, a schedule slip — pushed to the customer before they ask, in the account's own language and timezone. And the update written against the shipment record so the account history shows what was communicated and when. Schedules and tracking in Zavin answers "where is my container" continuously rather than on request.
Stage 7 — Billing and AP/AR
The manual version. The carrier invoice arrives and does not match the quote. Someone reconciles the difference line by line, decides whether to absorb it or rebill it, chases the carrier on the disputed accessorial, issues the customer invoice, and posts the file. Margin per file is discovered at month-end, which is too late to act on.
The tools. This is accounting territory. CargoWise and Magaya have integrated ledgers and are strong here; GoFreight includes accounting; Raft automates accounts payable alongside its document work; FreightSuite covers quote to cash as a stated scope. If AP reconciliation is your largest cost, this is the stage to shop first.
What good billing automation looks like
What good looks like. Charge codes carried unchanged from the quote through to the shipment and the invoice, so a file reconciles itself and variance is flagged the day it appears rather than at close. Margin visible per file, per lane, and per customer while the shipment is still moving.
Honest scope: Zavin does not run an accounting ledger. It holds the commercial record — quoted charges, actual charges, margin per file — and hands off to your accounting system. That is a deliberate boundary, not a roadmap item we are hiding.
A word on generic RPA and Zapier-style chains
Almost every forwarder tries this route first, and it is worth being straight about why it usually ends. A Zapier chain moves the attachment from the inbox to a folder and pings a channel. A UiPath bot clicks through the carrier portal at 6am and copies the milestone into a spreadsheet. Both work, for a while.
They break because they are deterministic and freight is not. The carrier redesigns the portal. The RFQ arrives as a forwarded thread with the actual request four replies down. The packing list is a phone photograph. The person who built the chain leaves and nobody documented the credentials. Deterministic automation is excellent at deterministic work, and a forwarding desk has less of that than it looks like from the outside — the exceptions are the job, not the edge case.
Where RPA still fits
The honest use for RPA here is one narrow, stable, high-volume handoff between two systems that will not change. Building your intake on it is how teams conclude that "AI did not work" when what did not work was a rules engine pointed at unstructured input. The distinction is drawn out further in AI in freight forwarding.
Vendor comparison
| Tool | Stages covered | Core strength | Public pricing |
|---|
| Sedna | 1 | Email management built for freight, at volume | No |
| LogBot | 1, partial 4 | Phone AI, email AI, AI dispatcher | No |
| Wisor | 2–3 | AI quoting and rate automation with TMS integration | No |
| Raft | 4–5, 7 (AP) | Document processing and customs; 130,000+ docs/month reported | No |
| Freightmate | 5 | Document processing and data entry | No |
| FreightSuite | 1–7 (stated) | AI-native FMS, agentic from day one, quote to cash | No |
| GoFreight | 3–7 | Cloud-native TMS with accounting; 4–8 week implementation | Partial |
| CargoWise / Magaya / Descartes | 4–7 | Operational depth, customs filing, integrated accounting | Partial |
| UiPath / Zapier-style RPA | Any single handoff | Cheap to start; brittle on unstructured input | Yes |
| Zavin | 1–6 | AI-native TMS + freight CRM + rate procurement loop | Yes — $125/seat/mo |
Capabilities and pricing models change; confirm current details with each vendor. Where pricing is marked "No," the vendor does not publish it.
Zavin's shape on that table is the point of it: twelve modules on one data spine covering intake through tracking, rather than a point tool bolted to five others. It works alongside CargoWise and Magaya with bi-directional sync, connects carriers via EDI and API, and reads Gmail and Outlook inboxes. Pricing is $125 per seat per month with every module included (see the pricing page), and deployments go live in under 14 days because nothing migrates. It is in pilot with a small number of forwarders running live RFQs and shipments daily.
How to sequence your automation
Order matters more than selection. Automating stage five while stage one is still manual gets you a faster back office feeding on stale inputs.
- Measure where the hours actually go. Two weeks, logged by stage. Most desks guess "documents" and discover it is intake and rate chasing. You cannot prove ROI on a baseline you never took.
- Automate intake first. Everything downstream is starved by slow, lossy intake. This is also the cheapest stage to prove, because the metric — time from arrival to first customer-visible reply — is unambiguous.
- Fix rates before quoting. A quoting engine with no valid rates produces fast requests for rates, which is not a quote. Build the rate database from inbound traffic first, then add the outbound procurement loop, then automate the quote on top.
- Then execution: shipment creation and tracking. These are the stages where re-keying compounds, and by now the upstream data is structured enough to flow into them without a human in the middle.
- Documents and billing last — unless they are your bottleneck. If you are a customs broker or your AP queue is the constraint, invert the whole order and start at stage five with a specialist. The sequence follows the pain, not the diagram.
Two rules at every step
Two rules that apply at every step: run draft-and-approve before auto-send, and pilot on one lane and one customer cluster rather than the whole book. The AI does the typing; your team keeps the judgment.
Bottom line
There is no single tool that automates freight forwarding, and any vendor list that implies otherwise is skipping the question that decides your outcome: which stage is costing you the most, and what order do you fix them in. Point tools win their stage — Raft on document and customs volume, LogBot when the phone still rings, Sedna on pure email management at scale. Full TMS platforms own execution and accounting. What none of the incumbent generation ever touched is the front of the funnel, where the inbox meets the rate desk.
That is the ground Zavin was built on: reading the mail, going out to get the rate when you do not hold one, and returning a quote in under 30 seconds when you do — as one system rather than five integrations. Start by measuring your stages. Then buy the one that hurts.
See it run on your own RFQs: get a 30-minute walkthrough, explore the platform, or take the free AI audit.
