This is a composite, illustrative scenario, not a report of a real customer. We wrote it to show the shape of a first deployment: what a team sets up, in what order, and what changes. The organization, the people and the figures are invented for the example.
The situation
The team in this scenario buys industrial parts for a regional distributor. One category, standard fasteners, is high in volume and low in drama: the same few suppliers, the same terms, small price movements. Yet each quote still takes a chain of emails, a call, a spreadsheet update and a manager’s nod.
The cost is not the time on any single quote. It is that the evidence of each agreement is scattered. When an invoice arrives at a price nobody recognizes, the team searches four places to find out who agreed to what.
Before: the evidence is everywhere
Email for the request, a call for the negotiation, a PDF for the quote and a spreadsheet for the agreed price. Each side keeps its own version.
After: one record, both sides
Every message, offer and approval lands in one hash-chained record. Both organizations hold the same copy, signed by both.
How they set it up
The team resists the urge to automate the whole category at once. They pick the narrowest useful job and write it down.
Verify the organization and the agent
The distributor verifies itself once, then issues a credential to one procurement agent. Approved suppliers connect their own agents, or use a hosted one, and verify themselves the same way.
Write a narrow mandate
Standard fasteners only, approved suppliers only, a per-quote ceiling and fixed payment terms. The head of procurement signs version 1.
Name the approver
One category manager handles every escalation, with a target of answering within the working day. A deputy covers holidays.
Run in parallel for two weeks
For the pilot, people still see every agreed quote the next morning. Nothing changes in the ERP until a person has read the record.
The first four weeks
Week 1
Narrow start
The agent handles requests to two suppliers. Most quotes fall inside the mandate; the ones that do not are mostly payment-term changes, which the mandate treats as out of scope by design.
Week 2
Reading the record
The category manager reviews every escalation in the record. Two rules fire far more often than expected, both for the same reason: a supplier’s new minimum order quantity.
Week 3
Version 2
The team adds the minimum order quantity to the mandate as a known term and signs version 2. The record shows exactly which quotes ran under which version.
Week 4
Widening carefully
A third supplier joins. The mandate is not widened in any other way until the team has another two weeks of records to read.
The agent did not make our decisions for us. It made it obvious which decisions were actually decisions, and it kept the receipts for everything else.
What changed
The most visible change is where people spend their attention. Routine quotes inside the mandate no longer need anyone. Exceptions arrive with the full context attached: the offer, the rule it broke, the history with that supplier. The approver decides once and the decision is written into the record, not into an email that someone has to find later.
The less visible change is in disputes. When an invoice and a quote disagree, there is one place to look, and both sides are looking at the same thing.
Questions to ask before your own pilot
- Which category is routine enough that a mistake would be cheap and visible?
- Who signs the mandate, and who answers escalations within a working day?
- Which of your suppliers can connect an agent, and which need a hosted one?
- What will you read at the end of week two to decide whether to widen?
