The Shift Report · Issue 14 · 4 October 2026
How to turn the night shift handover, SAP report and WhatsApp group into your morning briefing with Copilot or ChatGPT
Your 8 o’clock briefing comes from four sources that rarely agree — the handover note, the SAP report, the WhatsApp group and the late-night emails. AI can sort them into one briefing in under a minute — deciding what the site manager actually needs to hear is still yours.
You’re expected to brief the site manager at the 8 o’clock stand-up on what happened overnight. So at about quarter past seven, over a coffee, you put it together from the same places as yesterday.
The night supervisor’s handover note, written in the last ten minutes of a ten-hour shift. The production report out of SAP. The nights WhatsApp group: forty-odd messages, a third of them thumbs, and a photo of a leaking IBC that never made it into any log. And the dispatch inbox, where the haulier’s booking office emailed at quarter to six last night, an hour after your own office went home.
To give a good report you have to make sense of all this scattered information. Some of it will be conflicting, some slightly misleading, and some of it was overtaken later in the night by something else. Miss the thing that changed it, and the report is wrong.
I once reported a whole batch as ready for dispatch and overlooked that part of it had been put into quarantine for discolouration and suspected contamination.
ChatGPT, Copilot, Gemini and Claude have been able to help with this kind of task for a while now. Give one everything from last night and ask it to sort it out for you: the main points in order of urgency, anything that conflicts between the sources, anything that looks missing, and what to check before the meeting. It does the sorting in about a minute. You still decide what goes to the site manager.
How to do it tomorrow morning
- Open the tool. Most people on Microsoft products will have Copilot available for this. I find ChatGPT Plus so much better at it that it is worth trying, if your company policies allow.
- Give it the instructions first. You can start with this one:
“This is the information from last night’s shift, from four sources: the handover note, the SAP production report, the night shift WhatsApp group and the dispatch emails that came in after the office went home. Help me use it to brief the site manager at the 8 o’clock meeting. Give me a short summary of the main points, most urgent first. Highlight anything that conflicts between the sources or looks missing, and suggest what I should check before the meeting. Where the WhatsApp chat says a photo was sent, tell me which ones I should look at.”
What works for me is experimenting with the instructions. In my experience, the more specific you are and the more context you give the tool, the better the outcome. If you use Copilot regularly, it gradually learns more about your work and gets better.
- Add the handover note. If it is typed, copy and paste it. If it is written in a book, take a photo of the page. Pro tip: you can have the same chat open on the laptop and on your phone at once, so you can take the picture on the phone straight into the chat.
- Add the SAP report. Run it as usual and export it to Excel. Upload the file to the chat, or copy the rows and paste them in, or take screenshots and add those.
- Add the WhatsApp group. Most of you will not have WhatsApp connected to Copilot or ChatGPT, so you need to get the messages over to them. On your phone, open the group, tap the group name, then Export chat, Without media. That makes one text file with every message, which you can share straight into the Copilot or ChatGPT app on your phone.
The file leaves the photos out, but it shows where each one was sent, and the last line of the instructions asks the tool which ones to look at. If one of them matters, like the leaking IBC, open it in WhatsApp and add it to the chat from your phone, the same way as the handover photo.
- Add the emails. Copy the ones that came in after the office went home.
- Send it, then ask. Check every number, time and batch against where it came from before any of it goes into your briefing. If something is not clear, ask a follow-up question, including whether it is actually in the information you gave it. The nice part is you can have the output in whatever form you want: the numbers in a table, a short paragraph you can read out, or something ready to copy onto a PowerPoint slide.
The first time, about 10 minutes on top of your usual routine, most of it getting the WhatsApp file across. After that, closer to five.
Two things worth knowing
1. A wrong detail in an AI summary sticks
In a recent research test, AI summaries left out 51.6% of the central details on average. In one part of the test, people watched an animated video of an accident between a car and a pedestrian. Of those who then read an accurate summary, 83.6% answered questions about it correctly. Of those who read a misleading one, 44.8% did. The summaries were written by ChatGPT and Gemini. (Georgetown University and the University of Washington, reported by The National.)
My take: once your summary is written, people act on it. The site manager, the planner, the customer service desk. If it has the wrong collection time or the wrong batch status, that is what they remember and plan around. So whatever helped you write it, check everything in it before it leaves you.
2. Paper shop floor forms replaced by phone scans straight into SAP
Harney & Sons, a family-owned tea blender in New York, used to record goods-in and materials issued to production on paper, across more than 20 departments. The rules for what was acceptable lived in people’s heads, and mistakes like a receipt booked twice only showed up once stock and product costs were already wrong.
Now operators scan the barcode on the label for item, batch and location, either at a shared scanning station on the floor or with the camera on their own phone, and see what SAP thinks is there before they submit. Anything outside the department’s tolerance is held until a supervisor has checked it. Everything else goes straight into SAP, with a record of who entered it and when. They did not build this themselves. It is an add-on for SAP Business One, the version of SAP many mid-size firms run. Their SAP partner sells it and set it up around Harney’s own forms. (Full write-up.)
My take: the clever part is the hold. A wrong quantity gets stopped and looked at by a supervisor the moment it is entered, rather than found at month end when stock does not add up. If you run SAP Business One, ask your SAP partner what they offer for shop floor entry. Add-ons like this exist, and this one’s supplier says it takes weeks to set up, not a year. If you don’t run SAP Business One, the rule still works on paper: agree with your supervisors which numbers are checked before they are booked, and who checks them.
Final thought
The sources you use for the morning report will sometimes disagree, because different people saw the night from different angles, and the SAP report only shows what somebody booked. A tool can sort through all of it in a minute and point out where it doesn’t add up. Deciding what is true, and what the site manager needs to hear, is still the job.
I spend a lot of my time working out where operations lose time, especially in training, and how the new tools can help. Direct message me if you want to find out more or set up a call.
SOPwise turns a Standard Operating Procedure into a training package: task-based steps, critical mistakes, and scenario questions. £35 per SOP, no subscription. Built for UK manufacturing and chemical distribution.