One queue, described so it can be measured.
Everyone can send a chaser. Nobody has time to work out which ones to send.
Orders go out and confirmations are meant to come back. Some do. Some come back with a different date or quantity, and some never come at all. Someone is meant to check the order book each week and chase the gaps.
Most orders are confirmed without a fuss. The trouble comes from the ones that are not, and they are found late, usually when a delivery date is missed and a customer asks why. By then the chaser is a phone call and an apology.
What it usually looks like
You are not short of people who can write to a supplier. You are short of the hours it takes to find out who needs writing to today.
One AI employee, hired on a job description the operations lead writes.
Nothing new is installed for the team. The AI employee is connected to the operations mailbox, the order system and the delivery schedule the team already uses, and only to those.
On day one it only reads and lists. Every chaser waits for a person. Before anything runs, the operations lead sets two things, and both are recorded as decisions with their name on them.
How many days after an order a confirmation is expected, by supplier and by kind of order, and when a silent supplier becomes an escalation.
Nothing leaves the organisation without a person. Chasers are drafted with the order and the dates attached, and sent by a named person.
The order system stays the system of record. Xalios reads it, compares confirmations to orders and drafts the chasers. It does not change an order.
“Watch every order and request placed with a supplier. Compare each confirmation to the order, list what has not been confirmed by the date expected, draft the follow-up, and escalate anything that stays silent. Report to the operations lead.”
Four steps. One of them a person’s, and it takes a minute.
One order, from the moment it is placed to the moment it is confirmed. Nobody spends time on it until there is something to decide, and what reaches the operations lead is a chaser to send, not a report to read.
Order 55120 for 400 units is placed with a supplier that usually confirms within two days.
No confirmation has arrived. Drafts a chaser quoting the order number, the quantity and the delivery date, and holds it for a person to send.
Attaches the order and the supplier’s recent confirmation times, and puts the chaser in the operations lead’s queue with a note: this supplier has been late twice this month.
The operations lead reads the chaser and sends it. The confirmation arrives the next morning with the right date, and the order is marked confirmed with the exchange attached.
What lands with the operations lead is a chaser to send. Not a list to build.
What the auditor would read, and what the operations lead reads on a Monday.
The same record, with no separate reporting layer. Both can be replayed. Six months later the order, the confirmation and the chasers open as they were when the order was closed, not as they are today.
For the operations team, twelve weeks in.
Not fewer people. Building the list stops getting in the way of managing suppliers. Negotiation, the difficult conversations and the decision to change a supplier stay with the people who are good at them, and missed confirmations stop turning into missed deliveries.
Chasing happens weekly, from a report someone has to run.
Every overdue confirmation is chased on the day it falls due.
Confirmations that change the date or quantity are filed unread.
Every confirmation is compared to the order, and every difference is flagged.
Two people chase the same supplier about different orders.
One chaser per supplier per day, with every open order listed.
Nobody can say which orders are unconfirmed right now.
The order book answers it in a click, with the chasers attached.
Trust in automation is all or nothing, so it stays at nothing.
Authority is earned for each kind of order, raised by the operations lead and withdrawn automatically when quality slips.
What stays with the operations team
The expected dates and the trust level
Both set and changed by the operations lead. Every change is itself a record with a name on it.
Judgement on the suppliers
Escalations, changing a supplier, cancelling or re-ordering and anything involving a relationship stay with the people who own it.
The order system
It stays the system of record. Xalios reads it and drafts from it, and does not change an order.
Open an order from week three and get week three.
Most questions about late deliveries are questions about follow-up: was the order confirmed, was it chased, and who knew. A useful test at week twelve is simple. Pick ten late orders at random and ask for the chasers.
The answer to “did we chase it” becomes a click, not a search through sent items.
One AI employee becomes an operating model.
The memory built on confirmations, which suppliers confirm on time, which change dates, which go quiet, is the foundation the next role starts from. Each addition is faster than the last, and each one starts again at level 1 for its own kind of work.
Follow-up desk
The pilot. Level 3 as the target, with preparing the escalations as the next step you can choose.
Order desk
Reads inbound orders, checks them against stock and confirms what can be delivered.
Request triage
Reads each request in the operations inbox, works out what it is about and routes it to the right queue.
