The system says you have it. The shelf says otherwise.
We encode your match rules against the stock data you already keep, so reconciliation runs itself and every unresolved mismatch lands with its evidence.
Official services partner of the platforms defining AI
NVIDIA
Anthropic
The ERP says one number. The 3PL portal says another.
From operators we have sat with: mid-market manufacturers, distributors, and multi-location retailers between $100M and $500M — one ERP plus 3PLs, terminals, or branches, reconciled by hand. Not shopping for software; mid-audit or mid-month-end.
The weekly VLOOKUP
“Weekly stock comparison between the 3PL warehouse system and SAP, done by hand in Excel with VLOOKUPs” — a reconciliation required for audit compliance, currently manual.
Two formats, married by hand
One 3PL sends stock as a weekly emailed Excel report; the other has an online portal — two formats to marry before any comparison can run.
Blank location, quantity on hand
“Blank location with quantity on hand — someone has to go look.” The system says the item exists; a person hunts the aisle before any rule can run.
The stock position is a spreadsheet
“Product on hand spreadsheet.” No live view across terminals and suppliers, so every circulated copy is already wrong.
374 warehouses, nothing ties out
“374 active warehouses including technician vehicles.” Parts scattered across branches and tech cars — duplicate shipments because no record ties out.
Parts used but never billed
“Parts leakage — parts used but never billed; no real-time visibility.” Blind-received items surface as discrepancies weeks later, past the freight-claim window.
The weekly hunt costs more than the inventory adjustment.
What the manual way looks like at an operator marrying the ERP, the 3PL portal, and the shelf by hand. Your numbers will differ — the discrepancy audit puts figures on yours before anything gets built.
From anonymized engagements — mid-market manufacturers, distributors, and multi-location retailers, $100–500M in revenue
We encode your exceptions.
No new system for your team to learn. The match rules your best person carries in their head become the rules every export runs against.
- 01
Trace one month of mismatches
One warehouse/3PL pair, one month: every mismatch traced to its cause — receiving cutoffs, UOM conversions, emailed reports, transfer lag, blank locations.
- 02
Encode the match rules
Tolerances, identifiers, unit-of-measure conversions, lot traceability — which system wins when two disagree, and what needs human approval before an inventory adjustment posts.
- 03
Reconcile by exception
The rules marry the exports every cycle; your people see only the discrepancies the rules can’t resolve, each with its evidence attached.
A guide can list the causes. It can’t find yours.
A guide can list the causes. It can’t find yours. The software holds the stock ledger — it doesn’t hold your match rules, your tolerances, or which system wins when two disagree. We encode those.
Our AI does the reading — every 3PL export, emailed stock report, and ERP ledger line — against your encoded rules. Your rules do the judging; your people see only what the rules can’t reconcile.
The weekly VLOOKUP is gone. The system marries the 3PL export to SAP, and my team only touches the mismatches the rules can’t explain.
Led by Peter Enestrom.
Asked by inventory managers and controllers.
The straight answers, before you book anything.
An inventory discrepancy is a mismatch between the quantity a system says you have and the quantity that is actually there — ERP versus shelf, or ERP versus 3PL warehouse system. A stock discrepancy is the same mismatch at the single item and location. The record is wrong, the count is wrong, or the two disagree about when stock moved.
Almost never theft first. The usual causes are process: unit-of-measure conversions keyed inconsistently, receiving cutoffs that differ by system, transfer lag between locations, emailed stock reports keyed by hand, and blank locations that leave quantity on hand with nowhere to live.
No. A discrepancy is a mismatch between the records and reality; shrinkage is stock that is genuinely gone — theft, damage, loss. A discrepancy is investigated and usually traced to a process cause; what remains unexplained after that investigation is shrinkage, and it posts as a loss.
Most teams run the 3PL reconciliation by hand: export both sides, marry them in Excel with VLOOKUPs, and hunt the mismatches. The durable fix is encoding the match rules — tolerances, identifiers, unit-of-measure conversions, receiving cutoffs, which system wins — so the exports reconcile themselves and only unresolved mismatches reach a person.
After the inventory variance is traced to its cause and the approval rule has run — never as a quiet plug to make the numbers match. The rule proposes the adjustment, a person approves it, and it posts with its evidence and approver attached.
Start with one workflow.
Tell us where your team loses hours. We will come back with a straight answer on whether AI can help, what it would take, and what it would pay.