Production planning that is done before the team arrives
The company makes its own products and ships them from a 3PL warehouse. Its plan lives in a manufacturing ERP, purchasing in an inventory planning tool and Odoo, and stock in the warehouse system. We built the n8n workflows that keep those systems in step, so the plan is ready when the team starts.
The routine runs on its own. Anything unusual stops and goes to a person.
- Team
- Operations
- Built with
- n8n
- Manufacturing ERP
- 3PL warehouse
- Odoo
- OpenAI
- Slack
The problem
The plan was rebuilt by hand every morning.
Someone checked stock against reorder points, then created the purchase order in one tool and the manufacturing order in another. The production lead assigned the day's orders to operators before the shift started.
Finished orders sometimes missed their packaging bookings, and warehouse receipts sometimes arrived without a batch number. Both only surfaced when stock counts or traceability checks did not add up.
Suppliers who did not reply were chased by memory. When the person who remembered was busy, nobody chased.
What we built
Reorders that create their own orders
When an item drops below its reorder point, the workflow works out how much to make and when, then creates a purchase order in the inventory planning tool and a linked manufacturing order in the ERP. If the manufacturing order fails, Slack hears about it.
Operators assigned at 06:00
Every weekday at 06:00, the workflow freezes the day's plan and assigns each manufacturing order to the right operator. It never overwrites an assignment a person made by hand.
Packaging booked only when the numbers match
Every morning, finished orders that are missing packaging bookings get them added, but only when the warehouse receipts match what was made. Each booking is read back to confirm it. Anything that does not match goes to Slack.
Batch tracking checked every morning
At 08:00, the workflow checks goods received at the warehouse for a valid batch number and switches batch tracking back on where it was turned off. Traceability gaps show up the next morning instead of during an audit.
Suppliers chased until they answer
Every 15 minutes, an AI model reads supplier replies and stops the sequence when a real answer arrives. Otherwise the workflow sends a first and a second follow-up from Odoo, then asks the buyer in Slack to pick up the phone.
Guardrails
Where it stops and asks
Production data feeds stock, costs and traceability. These rules keep one bad run from spreading.
- Manual operator assignments are never overwritten
- A run that assigns nothing fails on purpose, so it cannot pass silently
- Packaging is booked only when warehouse receipts match
- Every booking is read back before it counts as done
- If more than 100 products have batch tracking switched off, nothing is changed and a person is asked to look
- Supplier chasing stops the moment a real reply arrives
Why we built it this way
A canary for the quiet failure
The worst failure is a run that succeeds and does nothing. A check that fails the run when no order was assigned turns that into an alert.
Self-healing with a ceiling
Switching batch tracking back on for a few products is helpful. Hundreds at once usually means something upstream broke, so above 100 products the workflow stops and asks.
People keep the last word
Planners can still move work by hand. The automation fills the gaps around their choices instead of replacing them.
Measured
What operations can see
Exceptions arrive in the channels the team already reads.
- Orders the reorder check created, and any that failed
- The morning assignment run and its canary result
- Packaging and batch exceptions, posted to the stock channel
- A Monday all-clear when the batch check finds nothing
- Suppliers who need a phone call
More case studies
Other systems our team built.
- Customer supportAn AI support desk where a person still decides what gets sentRead the case →
- FinanceA finance back office that reads its own post and supplier mailRead the case →
- E-commerce operationsShop, marketplace and warehouse kept in step without anyone watchingRead the case →
- Product and contentShopify product pages built from one approved recordRead the case →
- Leadership and engineeringReports that draft themselves and alarms that fire before things breakRead the case →
Related
Is your production plan still built by hand every morning?
Tell us which planning, warehouse and purchasing tools you run. We'll show you which steps can run before your team arrives.
Book a 15-min call