Logistics provider (35 employees): planners spend 17 fewer hours a week on email

Example case · anonymized · figures are illustrative

The situation

A freight forwarder and logistics provider with its own cross-dock has 35 employees and five planners. It arranges transport for about 120 clients: its own trucks for the region, carriers for the rest.

Requests come in three ways. By email, in free text (“can you pick up 6 pallets in Barendrecht tomorrow, must be in Antwerp by Friday”). Through three different client portals the planner has to log in to. And now and then by phone. The planning itself runs in a TMS that works well, once the data is in it.

The process before Optimi

1. The planner reads the request, in the email or in the portal.
2. They look up the client in the TMS and create a new order: pickup address, delivery address, date, number of pallets, weight, special instructions.
3. If something is missing (often), they email back and wait.
4. Meanwhile the next question comes in: “where is my shipment?” The planner looks up the trip, checks the latest status, and types a reply.
5. That happens about 40 times a day, in between the planning.

Together, the five planners lost about 24 hours a week to email: 15 hours retyping requests and 9 hours on status questions.

Where it failed

Not just the hours. The scan showed that a planner was interrupted every ten minutes on average. Planning is thinking work: combining routes, estimating waiting times, comparing carriers. That thinking was constantly cut short by a question a system could have answered.

A request took 45 minutes on average to get from inbox to scheduled order. Not because the work takes 45 minutes, but because it sat in line behind the status questions.

What Optimi did

Scan. Two weeks shadowing the planners. We counted emails, categorized them, and measured the lead time per request. Two candidates stood out: the requests and the status questions. They share the same source (the inbox) and the same destination (the TMS), so we took them on in one project.

Workflow automation. First the requests:

• One shared planning inbox where all requests come together; the three portals are read through their email notifications or an export.
• An AI step that pulls the fields out of the free text: addresses, date, quantity, weight, reference.
• A prefilled order in the TMS, with a short list of what is missing.
• The planner opens the order, fills in what is missing or sends a standard question to the client in one click, and approves.

Business AI and agents. Then the status questions:

• A status agent in the business AI workspace, with read-only access to the TMS.
• On a status question, the agent looks up the shipment and drafts a reply in the company’s tone, with the latest status and expected delivery time.
• The planner sees the draft and approves or edits it. The reply only goes out after approval.
• A weekly report of the types of questions, so the company sees which clients deserve a portal integration.

The models run in an EU region, in our managed environment. Client data is not used for training and does not leave the EU.

A person stays in the loop for: every order before it is scheduled, and every reply to a client before it is sent.

The result

Indicative results for this scenario, after two months:

• Planners on email: from 24 to 7 hours a week.
• Request to scheduled order: from 45 to 8 minutes.
• 85% of status questions answered within 15 minutes with an approved draft. The rest are questions that really need a planner.
• 0 client data outside the EU region.

The effect the planning team lead mentioned most: the calm. The planners are back to doing what they are good at, in blocks of an hour instead of ten minutes.