AI agents for purchasing and logistics in SMEs

Today

Three suppliers reply with PDFs laid out differently: units, lead times, shipping costs and payment terms are never in the same place. The buyer copies everything into Excel and chases each supplier for missing information.

With AI

When a request for quotation closes, an off-the-shelf agent connected to your mailbox and SharePoint (Microsoft 365 Copilot, ChatGPT Business or Claude with connectors) gathers the quotes received, puts them into a comparison table on the same criteria and flags missing information or unusual terms. It drafts the questions to send to each supplier. The buyer checks the figures against the originals, sends the questions and makes the choice.

Today

Writing a clear request for quotation, in German for a supplier in German-speaking Switzerland or in English for one abroad, takes time. Vague requests come back as quotes that are hard to compare.

With AI

Starting from your notes and a shared template, the buyer asks the assistant to structure the request: quantities, specifications, deadlines, selection criteria and expected reply format, in the supplier’s language. The assistant does not act in any tool: the buyer completes it, has the technical requirements approved by the relevant department and sends it personally.

Today

At goods receipt, the storekeeper compares the paper or PDF delivery note with the order in the ERP, line by line. Partial deliveries and substitute items easily go unnoticed.

With AI

We build an agent triggered by each delivery note that is scanned or received as a PDF: it reads it, matches it to the open order in your ERP, prepares the goods receipt with the quantities delivered and lists the differences (quantities, references, missing or substitute items). For each difference, it drafts the message to the supplier. The storekeeper confirms the receipt in the ERP, and the buyer decides what to do with the supplier.

Today

Checking stock, open customer orders and each supplier’s minimum order quantities is done by hand in the ERP and Excel. When time runs short, items run out or orders go out as rush jobs.

With AI

We build an agent connected to your ERP: as soon as an item drops below the threshold you have set, it checks available stock, open customer and supplier orders, and the supplier’s lead time and minimum order quantity, then prepares a draft supplier order, grouped by supplier, with an explanation for each line. The buyer accepts, corrects or rejects it and releases the order according to your signing rules; the agent sends nothing on its own.

Today

The foreman orders by text, photo or voice message, without item numbers or cost centre. The office reconstructs the order, calls the site back and passes it to the supplier at the end of the day.

With AI

We configure an agent that receives the site message in Teams or by email, finds the site and cost centre in your construction software, matches the items to your usual suppliers’ catalogues and asks the foreman for anything missing. It prepares the order with the requested delivery date; the site manager or buyer approves it, then the agent sends it to the supplier and records it in the site file.

Today

At a carrier or warehousing firm, much of the phone and email traffic asks where a shipment is. The team looks it up in the transport system and partner portals, then replies case by case.

With AI

We configure an agent connected to your transport management system and carrier portals: for each tracking request, it identifies the shipment from the tracking number or customer reference, checks the status and replies with the next known step, signing as an AI agent. Delays, damage and disputes go to a staff member with the file already gathered; they decide on the reply and any compensation.

Today

For each import or export, someone has to go through the commercial invoice and packing list, find the tariff number and origin, then enter the declaration. Errors cost delays or corrections after the fact.

With AI

We configure an agent triggered when a shipment is created: it reads the commercial invoice and packing list, takes from your ERP the tariff number and origin already assigned to each item, suggests some for new items from your history and flags inconsistencies in weight, value or origin. It assembles the file ready to enter in Passar or to pass to the forwarding agent; the declarant checks, corrects and files the declaration personally.

Today

Each week the head chef draws up food orders from memory and a quick stock count. Quantities do not always follow bookings or planned menus, leading to shortages or waste.

With AI

Each week, the agent reads the planned menus, the bookings in your software and your recipe cards, deducts the recorded stock and prepares an order per supplier according to their delivery days and minimums. The chef adjusts it to the market and approves it; the agent then sends the orders by email or through the supplier’s portal and keeps a record of what was ordered.