Invoice to purchase order matching
Month-end close drags because someone has to match every supplier invoice against its purchase order and chase the mismatches. Design an agent that does the matching — reading invoices from the inbox, pulling purchase orders from your accounting system, and flagging discrepancies in amount, quantity, or terms for a human to resolve. What does it get access to, what can it never approve on its own, and how would you verify its match accuracy? Finish by selling me the business case in a minute.
classify_invoice_match(po_number: str, invoice_total_cents: int, po_total_cents: int, invoice_quantity: int, po_quantity: int, is_duplicate: bool) → str["PO-1041",125000,125000,10,10,false]out"clean"["PO-1041",126500,125000,10,10,false]out"price-variance"["",50000,50000,4,4,false]out"missing-po"Turn the fuzzy goal into a bounded agent task. Say what the agent reads, what it may do on its own versus draft for a human, and what it must never touch. Then earn the trust: how you’d verify it before rollout, and the pitch (the problem, what you built, why it matters).
Agent build: design the agent in plain language (or narrate it). The coach grades the decomposition, the guardrails, the verification plan, and the pitch.