01
Start from real problems
Don't rush into a big system. Begin with the daily friction — screening inquiries, tidying rates, checking B/L fields.
For now we keep a few entry points that are easiest to land and best show the method.
Identify inquiry value, lane priority, quoting opportunities and to-do reminders.
Break down inquiries, extract fields, draft replies and list items to confirm.
Turn messy rate info from chats and emails into standardized tables.
Let AI compare fields and flag differences first; people confirm the final result.
Full automation in one leap is not the goal. Separate the two kinds of work first, and automation holds steady — and earns trust.
AI takes over the repetitive, easy-to-miss work; the calls that actually matter stay with people.
「Order first, then automate」Put the work in order first — fields, rules, risks and human-review points — then let AI carry it out within clear boundaries.
Don't rush into a big system. Begin with the daily friction — screening inquiries, tidying rates, checking B/L fields.
Which clients reply first? Which lanes get priority? Which fields must a human re-check? Turn that experience into clear rules so AI can hold steady.
High-risk actions — quoting, cargo release, documents, compliance — are never AI's call. AI preprocesses; people confirm and commit.
No need to define a giant AI project first. Bring an email thread, a rate sheet, a quoting rule or a B/L checking process — and see if AI can lay the groundwork for you.