Freight forwarders
AI automation for freight forwarders
Ops is tooled to the teeth. The sales desk still runs out of a shared inbox. For freight forwarders and small 3PLs, the enquiry-to-quote side is where the money leaks, and that is the side I build. I built the CRM and placement platform for a shipping and logistics training company, so I know this world.
Who this is for
Built for the middle of the market
Not the top 25. The forwarders between five and fifty people: forwarder on one file, NVOCC on the next, agent for a partner overseas on the third. If your licence numbers live in your footer and your rate desk lives in a shared inbox, this is for you.
The work you do
Transpacific, transatlantic, LATAM and cross-border lanes. Ocean FCL and LCL, air express and deferred, FTL and LTL over the road, drayage at the port. Consolidation, warehousing, cargo insurance.
The people you serve
Importers and exporters with real import programs: e-commerce brands, manufacturers, perishables, project cargo. Clients who call you when something sticks, because you are the one who answers.
The big platforms chase the top of the market, and the load boards own the commodity bottom. The automation gap lives in the middle: forwarders, small 3PLs and asset operators whose operations are systematised and whose sales side runs on memory. That middle is where I build.
The mess
What lands in your inbox
- A rate request with lane, cargo, weight and dates buried three replies deep in a forwarded thread.
- "Can you quote 2 x 40HC, Ningbo to Long Beach, cargo ready end of month?"
- A pre-alert from the origin agent, six attachments, and the same file data re-typed into the TMS for the fourth time.
- Status questions your team answers by copy-pasting out of the ops system, one email at a time.
- Quotes sent, follow-up forgotten, win rate by lane unknown.
None of this is a people problem. When Freightos surveyed the top 20 forwarders in 2018, manual quoting averaged 57 hours from enquiry to quote. The desk is short-staffed, not slow; the mess is structural.
The build
What I would automate first
Rate-request capture
Lane, cargo, weight, dates and terms extracted from emails and threads into one clean pipeline. Works whichever hat you are wearing on the file: forwarder or NVOCC.
Fast quote drafts
The quote email drafted in your voice from the structured request, ready for your rates and a human check, out while the shipper is still deciding.
Follow-up until it books
Every open quote chased until it books or dies honestly. No thread depends on somebody remembering, and the written rate trail exists before the cargo loads.
Status updates in your voice
Proactive shipment updates drafted from the data your ops system already holds, approved by a person, so clients stop asking and your team stops copy-pasting.
Quote-desk visibility
Quotes, wins and margin by lane and by customer, answered in plain language, so you see which lanes are carrying the year.
The second act: the docs desk
Pre-alert and arrival-notice processing
The file data lifted off the pre-alert and the arrival notice the moment they land, so free time is not discovered on a Friday afternoon.
Document pre-fill
Commercial-invoice and shipping-document fields pre-filled from what is already in the thread. Your team checks every field; nothing sends itself.
Carrier-invoice reconciliation
Debit notes and carrier invoices checked line by line against the quote and the file before they hit your P&L.
Email-to-TMS entry
The re-typing between the inbox and the ops system done by the machine, checked by a person.
You do not need all of these. I start where the impact is biggest for the least effort, intake before documents, and a person stays in the loop on everything that reaches a client.
Do the math
What is sitting in your unchased quotes?
Chasing each open quote until it books or dies. The win-rate lift is your assumption, not my promise.
Example: $25.22 an hour, the BLS wage for office and administrative support, June 2026
Adjust my assumptions
40-hour week, 2,080 paid hours a year. On-costs: x 1.46 total compensation per dollar of wages (BLS, office and administrative support).
Cycle time
Stock (optional)
Quotes won (freight follow-up)
10 hours back a month
3 people-weeks a year (0.06 of a full-time person)
Request to done:
$1,767 to $4,418 a year (expected about $2,209)
Extra booked value: about $9,000 a month at your lift. That is revenue, not profit.
At the cautious figure, a one-off fix costing up to $1,767 pays back within 12 months ($3,534 within 24), before running costs.
120 x 5 minutes saved / 60 = 10 hours a month.
This is an illustration with your numbers. A sit-in measures the real ones.
Link copied. It carries your numbers, never your salary.
Opened from a shared link. Salaries are never shared, so add your own to see money.
How this works, and where the defaults come from
- Hours a month = volume x (minutes today minus minutes after) / 60. People-weeks = hours a year / the standard week.
- Money is a band: the ceiling counts every freed hour at the loaded cost of the person; expected counts half of it, because not every freed minute becomes useful work; cautious takes a further 20% off.
- Stock: the yearly carrying saving is counted; the one-time cash released is shown on its own and never added in.
- Payback is shown as a budget: the most a one-off fix could cost and still pay back within 12 or 24 months. No prices.
- Every volume and minute figure is an example. Replace it with yours.
Sources, last checked 12 October 2026:
- Forrester Total Economic Impact method: 50% productivity recapture, 10 to 20% risk adjustment (this study was commissioned by Microsoft)
- APQC open standards benchmark: cycle time from requisition to PO, goods, median 2.0 days
- Carrying cost rule of thumb, 15 to 25% of stock value a year, citing ASCM guidance (MRPeasy, a software vendor, January 2025)
- Freightos (a freight marketplace) survey of the top 20 forwarders, 2018: manual quoting averaged 57 hours from enquiry to quote
- Vietnam employer social, health and unemployment insurance (21.5%) and union fee (2%), statutory rates
- Rodl and Partner, Vietnam statutory base salary and insurance cap from 1 July 2026
- US Bureau of Labor Statistics, employer costs for employee compensation, June 2026
- GOV.UK, rates and thresholds for employers 2026 to 2027
- GOV.UK, workplace pension contributions
- ONS, Annual Survey of Hours and Earnings 2025
Your stack
CargoWise, Magaya, or a very good spreadsheet
The ops system keeps doing operations. I build the sales layer that runs before and around it, wired to the data you already hold, and I build with what is actually on your desks: a TMS, Outlook, QuickBooks, a shared inbox. Nothing gets ripped out, and you do not move your business onto my platform, because there is no platform.
What I build, you own. Your accounts, your data, documented, and your team can run it without me. It does not cost more when you ship more: there is no meter on your growth.
Why me
I have worked in this world
I built the CRM and placement platform for a shipping and logistics training company, so I know this world. It now runs their live sales pipeline. I also structured messy enquiry email into clean, tracked records for a travel and car-rental group: a different industry, the same leak. The gap between heavyweight ops software and a sales desk running on memory is the gap I build in.
Ergosys AI is one senior builder, not a platform with a sales team. You deal with the person who does the work, the scope is written down, and when it is done your team runs it without me.
Fair questions
Asked before, answered honestly
We run CargoWise, Magaya or similar. Does this clash?
No. The ops system keeps doing operations. I work the sales side that happens before and around it: enquiries, quotes, follow-ups and client updates, wired to the data your system already holds.
Our rates change weekly. Won't the drafts be wrong?
The system drafts; your people price. Rates come from wherever they live today, your sheets and your contracts, and a person confirms every number before anything goes out. What stops being manual is the reading, the re-typing and the remembering, not the pricing.
We are eight people, not a top-25 forwarder. Is this overkill?
The opposite. The enterprise AI platforms in this space are priced and sized for the top of the market, with volume minimums to match. A small desk is exactly where one well-built automation pays for itself: you feel every lost quote, and the person checking the output is usually you. I build to the size of your desk, not to a seat count.
What will you not automate?
Anything with your name or your judgment on it. Pricing decisions, claims, compliance sign-off, anything with legal exposure: a person does those, every time. And if the honest answer is that a shared-inbox rule and a same-day reply habit would fix it, I will say that instead of selling you a system.
This business runs on relationships, not software.
Agreed, and relationships are kept by fast answers and promises that do not slip. The automation makes sure the enquiry gets answered today and the follow-up actually happens; your people stay the relationship.
Tell me where your desk leaks. If AI will not help, I will say so.
Get in touch