Your metasfresh server is up, you have root, and the ERP holds your orders, stock, bills of materials and invoices. But clicking through the WebUI for every small chore is slow. The Vibe-code agent in your Open WebUI chat can read the app’s configuration, query the PostgreSQL database behind metasfresh, run docker commands and do the work while you watch. Here are ten everyday jobs you can simply hand over.

1. Bulk-Load Business Partners from a Spreadsheet

Signing up a chain of new outlets, or onboarding a vendor list from an acquisition, means typing dozens of business partners into the CRM. The agent can take your CSV or Excel file, map the columns to metasfresh business partner fields, and dedupe against existing records by GLN, tax ID or email before anything is written. It assigns the right price list, payment terms and business partner group as it imports, and shows you a summary of what it created. Try: “Import the customer list from /root/new-customers.xlsx, skip rows that already exist by GLN, and put everyone on the wholesale B2B price list with 14-day payment terms.”

2. Roll Out a Seasonal Price Increase

metasfresh prices live in price lists with versions that have a valid-from date, which is exactly what you want for a clean price change. The agent creates a new version of your B2B list, applies the percentage increase across the products, and flags the ones where the new price would drop below your margin floor. Contract customers on fixed prices can be left untouched if you tell it which group they sit in. Try: “Create a new price list version for the B2B list, valid June 1, raising everything by 4 percent — keep the contract customers on their current prices and show me the 20 biggest increases before you save.”

3. Get a Morning Operations Brief

Owners ask the same questions every morning: what came in, what went out, what is stuck. The agent can write a small script against the database and schedule it to give you a daily digest — sales orders received, shipments completed, goods receipts posted, dunning candidates created overnight, and products that fell below their reorder point. Once it runs, you read one message instead of five windows. Try: “Set up a 6:30 AM report: sales orders received, shipments completed, receipts posted, new dunning candidates, and any product under its reorder point — email it to me and post a copy in chat.”

4. Work the Dunning List

Chasing overdue invoices is a daily reality in wholesale. The agent pulls today’s dunning candidates with amounts and days overdue, shows you who moved up a level, and prints or emails the notices you approve. For chronic late payers it can set a credit limit or zero-credit hold so new sales orders stop until they pay. Try: “Show me all dunning candidates over 1,000 EUR more than 21 days overdue, email level-2 notices to the ones I approve, and set a credit limit of zero for Acme GmbH until they pay.”

5. Clear the Purchase Disposition

Material planning in metasfresh produces purchase candidates — suggested order lines with quantity and needed date. Turning those into real purchase orders is clicking work the agent can do: it groups candidates by vendor, checks the last purchase price for each product, respects minimum order quantities, and creates the orders. Anything suspicious it leaves for you with a note. Try: “Turn all purchase candidates due this week into purchase orders, group the lines by vendor, and flag any line that is more than 10 percent above the last purchase price for that product.”

6. Post Goods Receipts Against Open POs

Every delivery needs a material receipt against the purchase order, often with handling units, lot numbers and tolerances for over- or under-delivery. The agent finds the POs due, posts the receipt with the quantities from the supplier’s delivery note, records batch numbers where needed, and reports shortfalls so you can chase them. Try: “Post a receipt for delivery note 45821 against purchase order 10234 from supplier Natura — then list every open PO line that is still short and by how much.”

7. Clean Up Product Master Data

Product data rots quietly: duplicates creep in, a UoM conversion between “each” and “box” is missing so orders cannot be packed, GTINs are blank, or ingredient and allergen attributes are empty while your food customers expect them on labels. The agent can audit the whole product master, produce a ranked problem list, and fix the boring ones in bulk once you approve. Try: “Audit the product master: list every product with no GTIN, every pair of possible duplicates, and every product missing a UoM conversion between each and box — don’t change anything yet, show me first.”

8. Hunt Dead Stock

Slow movers tie up cash and shelf space. Because metasfresh tracks every stock movement, the agent can rank products by months without a single outbound movement and by the purchase value sitting on the shelf, then propose a clearance price list version for the worst offenders. You review the numbers, adjust the cut-off, and let it create the version. Try: “Rank all stocked products by months without outbound movement and the value on the shelf, and draft a clearance price list for the worst 30 products at cost plus 5 percent.”

9. Answer “Can We Ship This Today?”

Every wholesaler gets the call: is order 10442 going out today? Answering honestly means checking real stock against reservations for every line, seeing whether manufacturing can cover a shortfall, and whether an alternative batch exists. The agent does that read across inventory, reserved handling units and open manufacturing orders, and gives you a straight answer with the gaps spelled out. Try: “Sales order 10442 is promised today — check real stock and reservations for every line and tell me honestly if we can ship it complete, and if not, what is missing.”

10. Keep an Eye on the Server Itself

metasfresh runs as several docker containers — app, webapi, webui, postgres, rabbitmq and the search service — and the ERP is only as healthy as the box under it. The agent can check that every container is up, disk and memory headroom, errors in the app logs from the last 24 hours, and whether last night’s backup is actually restorable. Try: “Check the metasfresh stack: are all containers healthy, how much disk is left, any errors in the app logs in the last 24 hours, and is last night’s backup restorable?”

Start with one job from this list and watch the agent work — ask it to explain each step as it goes. Once these feel routine, the automation and customization articles below will take you further.

See metasfresh on OpenSysLab for the server that runs all of this out of the box.