What Is the Hardest Part of AI Adoption for Freight Forwarders? A Shipping Authority's Answer

# What Is the Hardest Part of AI Adoption for Freight Forwarders? A Shipping Authority's Answer

The bottom line first: the hardest part of AI adoption for freight forwarders is not technology selection, but getting business knowledge out of people's heads, turned into rules, and taught to the AI. Martin Stopford, a shipping-industry authority, once said that in the AI era the scarcest resource in shipping is not technology but people who understand the business. Freight-forwarding knowledge lives almost entirely in the minds of veteran staff (rate structures, route connections, customs rules, exception handling); when a veteran leaves, the knowledge goes with them. AI amplifies the problem: no matter how smart the model is, without business rules to learn, it is an empty shell. Interviewing veterans to extract their judgment, writing it into rules, and feeding those rules to the AI is the real first step of AI adoption.

Why Forwarder Business Knowledge Is Passed Down "Person to Person"

Freight forwarding has no textbook for its core skills. A veteran with ten years of experience carries a complete judgment system in their head: for a dangerous-goods shipment, ask about the product name and MSDS first, then decide which carrier, whether to expedite, and what special rules apply at the destination port; when a client says the cargo will miss the cut-off, they immediately calculate whether to take the next sailing or transship, and what the cost difference is. These skills are honed entirely through practice; companies cannot carry them away, and newcomers learn slowly. AI sharpens this contradiction: AI depends even more on being taught. Feed it templated scripts and it becomes a parrot; feed it business rules and it can actually work.

Judgment Interviews: Extracting Rules from Veterans

When we implement AI for forwarders, the first step is never tools, it is interviews. For one forwarding company with fifteen years of history, we spent two weeks interviewing the owner and two veterans, extracting their decision processes one by one into 47 rules, from inquiry classification to quoting strategy, from exception handling to client tiering. Once the rules were written, the AI knew what to ask clients, which variables to consider when quoting, and when to escalate to a human. Those 47 rules became the first "business brain" of the company's AI employee.

An AI Without Business Knowledge Fails on Day One

One company bought an AI customer-service system. On the first day, a client asked: "My cargo arrived in Hamburg and the consignee information on the bill of lading needs to be changed. What should I do?" The AI replied: "Please provide your order number." The client was furious: they had already given the vessel name and voyage. The reason is simple: the vendor taught the AI how to talk, but not what freight forwarding actually is. The AI did not know the process for changing a bill of lading or what documents were needed, so it fell back on the dumbest templated reply. Without business rules, the AI is an empty shell.

Measurable Results After Business Rules Are in Place

Once judgment is extracted, rules are written, and the AI is connected, results can be quantified. A Shenzhen cross-border forwarder had five customer-service staff handling over 200 inquiries a day, with two-hour response times and a conversion rate stuck at 3%. We turned their quoting logic and common questions into a knowledge base and equipped them with an AI quoting assistant: quoting dropped from 30 minutes to 5 seconds, responses were nearly instant, response time fell to two minutes, conversion rose to 8%, customer-service costs fell 70%, and it went live in three days. Another company processes over 7,800 inquiries a year; previously ten people tracked them, now three people plus an AI daily report do the job, with people handling judgment and AI handling volume.

AI Adoption Is Essentially Business-Knowledge Organization

Models will keep getting stronger, but a model does not know your forwarder's quoting logic, your clients' special requirements, or the pitfalls you have stepped in. Whoever first turns business knowledge into rules will get AI to work for them. Judgment interviews, knowledge bases, and rule formalization sound unglamorous, but they are the most solid step in AI adoption.

FAQ

Q1: Is a judgment interview really necessary before AI adoption?

Not strictly, but strongly recommended. Tools can be deployed anytime; without business rules, AI is just a chatbot. Spend one to two weeks turning core business judgments into rules first, and the quality of AI output will be completely different.

Q2: Small forwarders with limited business knowledge, should they do this too?

The smaller, the more necessary. Large companies pass knowledge through systems and institutions; small companies rely entirely on a few core people, and when they leave, the knowledge is gone. Turning judgment into rules helps small companies build barriers more easily.

Q3: What exactly should the judgment interview ask?

Start with high-frequency scenarios: how to classify clients, how to calculate quotes, how to handle exceptions, when to escalate. Ask veterans to walk through their decision process for each scenario, write it into rules, and let the AI execute and ask questions according to those rules.

Q4: Will AI replace forwarder staff?

It will not replace people who understand the business, but it will widen the gap. Hand repetitive work (quoting, tracking, copywriting) to AI, and keep people focused on judgment and client relationships; the gap can widen within a year.