A customer wants to know whether a product is in stock and when it will be delivered. This is one of the most frequently asked questions to a customer service or sales support department, and it is also one of the questions that is easiest to automate. Not because the question is unimportant, but because the answer is already stored somewhere in a system. The task does not consist of coming up with an answer, but of retrieving it from ERP, WMS or CRM and presenting it correctly to the customer.
Three axes are decisive here: structuredness, volume and customer contact.
Structuredness scores high. Stock quantities and delivery times are fields in a system, not matters of interpretation. There is no document to read, no context to weigh: the answer is a number or a date that has already been calculated somewhere.
Volume is likewise high. This is not a task that occurs a few times a week, but a question that comes in dozens or hundreds of times a day, via phone, email, chat or a contact form. A task that occurs frequently and shows little variation is exactly the kind of work where automation has the fastest effect.
Customer contact scores moderate: there is a customer on the other end, but it concerns a factual question with a factual answer, not a conversation in which a customer needs to be reassured or convinced. That makes it suitable for automated handling, as long as the tone and presentation are right.
The remaining axes confirm this picture. Physical work plays no role, no creativity is needed, and compliance risks are limited: no binding commitment is made beyond what the system already indicates. The cost of errors falls in the middle segment. An incorrect delivery time leads to a disappointed customer, not a legal problem, but it is annoying enough to take seriously.
Room for judgment scores low, and that is precisely where the conditions come into play. A system may say that twenty units are still in stock, but whether that number is reliable depends on how up to date the link is. Has the stock just been sold to another customer but not yet been deducted? Is the delivery time a fixed value from the ERP system, or a prediction that takes the supplier, the season and the transport time into account? As soon as the question shifts from "what does the system say" to "what is likely to happen", an assessment is needed that a customer service employee today often still makes based on experience.
This is why the takeover is not unconditional: it stands or falls on two things. There must be a real-time stock link, so that the answer is not based on yesterday's export. And there must be a reliable delivery time prediction, not just a static number from the system. If either is missing, AI gives answers that match the system but not reality, and that is worse than no answer at all.
At a company with a well-organized WMS and an ERP that tracks stock and delivery time at order-line level, an AI agent can already handle most of these questions independently today, with an escalation to a human in case of doubt or in an exceptional situation such as a backorder or a supplier delay. At a company where stock data is maintained manually in a spreadsheet alongside the system, that takeover is not feasible without first getting the data source in order. So the difference does not lie in the task itself, which is virtually the same at almost every company, but in the systems behind it.
This task also does not stand on its own. It is connected to checking and communicating order status, which is very similar in structure: there too the answer is largely a system query with a presentation layer in front of it. Also relevant is entering orders into the system, because correct order entry forms the basis for reliable stock and delivery time information further along in the process. And where customer data plays a role in personalizing an answer, this touches on updating customer data in CRM.
This page describes capacity, not personnel policy. Whether and how a company does something with freed-up hours in customer service is a decision that lies with the employer and is subject to its own legal requirements. We do not provide justification for that choice, only facts about which part of the work a system can take over today. Anyone wanting broader information on where automation conflicts with responsibility can find an overview at high-risk use of AI in the workplace, and anyone wondering how long information about stock and communication must be retained can read about that at archiving documents and checking retention periods.
This task is one of the many components of the work in a customer service or sales support role. To see which part of the hours in such a profile can already be taken over by AI today, without this saying anything about your specific company, you can fill in the free quickscan: twelve questions, no account needed, with an immediate indication. The full work scan, which breaks down the work of your own company task by task and translates it into FTE capacity, is still under construction.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.