TL;DR: The Quick Summary for Busy Managers
Most of the value is not a clever prompt. It is asking the question you already have, in the words you already use, and getting an answer from live warehouse data. Lookups come back as a list, a chart, or a file. Anything that changes the warehouse, or pages the floor, stops and waits for approval.
The questions below are the untidy ones. "What sold best last month?" rather than a report name. "Do we have stock of this SKU anywhere?" rather than a location enquiry screen. That is how people actually ask, and it is what they search for later.
Dave is the agent inside eveXso, not the chat on this website. It reads your live warehouse data and, where a person approves it, can act. The definition of that difference is in what agentic AI means in a warehouse. How it sits in the product is on the AI platform page.
01
"What sold best last month?"
What Dave does. It reads last month's shipped lines from live data and ranks products. Ask how that compares with the month before and it uses the same thread.
What comes back. A short ranked list. Ask for a chart and the bars appear in the conversation.
02
"Do we have stock of this SKU anywhere?"
What Dave does. It looks up on-hand quantity for that product across the warehouses you can see, rather than one bin you happen to have open.
What comes back. Where it is, and how much. If the answer is none, that is the answer.
03
"Which orders are at risk of missing their cut-off today?"
What Dave does. It checks orders due today against how far they have got, and what is still in the way.
What comes back. A ranked list. The useful line is the first order and what is blocking it, so you can decide what to run.
04
"Why has this order not released?"
What Dave does. It follows the order: status, hold, stock, and whether an integration error is sitting on it. Each step depends on what the last one found.
What comes back. A finding you can act on, not a ticket number. Status, what is holding it, and what to do next.
05
"What is sitting in the pick queue right now?"
What Dave does. It reads open picks as they stand, not a report saved yesterday.
What comes back. What is waiting, in enough detail to see volume and what is stuck. This is a snapshot of now.
06
"Show me picks by status this week as a chart."
What Dave does. It adds up this week's picks by status and draws the chart in the thread.
What comes back. A chart in the conversation. It is not a separate dashboard that updates itself after you close the chat.
07
"Export that as a CSV."
What Dave does. It takes the result already in the thread and builds a file from it. This only makes sense as a follow-up.
What comes back. A file you can download and open. If the rows are not there when you open it, it was not an export.
08
"Which supplier is late most often?"
What Dave does. It compares receipts with the dates on the purchase orders and counts who misses.
What comes back. A ranking, not a single anecdote from last Tuesday. Thin receipt history produces a thin answer, and it should say so.
09
"What changed on this order and who changed it?"
What Dave does. It reads the history already stored against that order: status, holds, and the user on each change.
What comes back. A short list. What moved, when, and who. Useful when someone says the order "just changed".
10
"Are any integrations failing today?"
What Dave does. It checks today's integration errors and reads the failure in plain language, using the logs rather than a guess.
What comes back. Which connection failed, and the error. Not every external system is in scope. If Dave cannot see that integration, it should say it cannot see it.
11
"Tell the picking team this wave is blocked."
What Dave does. It drafts a short alert to your own picking team. This is a write, so it stops and shows who will get it and the text. An authorised person approves it. Someone without that permission cannot send it.
What comes back. An in-app and push notification to that team, after approval. Dave does not email your customers, and it does not post to Slack or Teams.
12
"From now on, only show me the main warehouse unless I ask for another site."
What Dave does. This one is not a question. It is a rule you set once. Dave saves it as a standing instruction after you approve it, and you can read, change or turn it off later.
What comes back. Tomorrow, "what is in the pick queue?" answers for the main warehouse without you saying so. This is not memory and it does not learn your habits. It follows the sentence you wrote, which is why the answer is predictable.
The two that are easy to oversell
Eleven and twelve are the ones that get described as magic. They are not.
Telling the floor is a real send, limited to people allowed to notify that team, and it waits for a person to approve the message. The last one is not a question at all: it is a standing rule you write once. That is predictable in a way "the AI remembers you" is not. If you did not write the rule, it is not in force.
Bring your own Monday. Book a discovery call and ask one of these against your data, including one that has to wait for approval.
Prefer to read first? Start with what agentic AI means in a warehouse or how Dave works in the platform.
