The problem no one calls a problem
In most mid-size companies with supply chain operations, the ERP exists. SAP Business One, Dynamics 365, Sage, Odoo — the system is there. The problem is that it is not enough.
The operations or logistics team maintains parallel spreadsheets because the ERP does not respond at the speed they need, because the planning module is rigid, because the vendor has no API, or simply because "that's how it's always been done." The result is a dual ecosystem: the ERP as the system of record and Excel as the actual working system.
That gap has a cost. And in most cases, that cost has never been calculated.
What it costs to run two systems in parallel
The direct cost is the easiest to see: hours of qualified staff spent copying, pasting, cross-referencing, and validating data between systems. In a company with a four-person supply chain team, that work can represent between 15 and 40 hours per month. At an average cost of 25–35 €/hour, that amounts to 375 to 1,400 € per month in work that generates no value.
But the indirect cost is greater.
When data is not synchronized in real time, decisions are made on outdated information. A purchasing manager working from a spreadsheet updated on Monday is making decisions on Thursday with data that no longer reflects reality. That produces incorrect purchase orders, unanticipated stockouts, excess inventory in the wrong SKUs, and in some cases, penalties for failing to meet customer commitments.
In food or industrial manufacturing companies with tight margins, a 5% inventory planning error can represent tens of thousands of euros in tied-up capital or lost sales.
Why the standard response does not work
The typical reaction to this problem is one of two: either replace the ERP, or hire additional staff to manage the spreadsheets.
Replacing the ERP is an 18-to-36-month decision, with an implementation cost that in mid-size companies ranges between 150,000 and 500,000 €, plus the operational disruption during migration. For most companies, it is not a viable option in the near term.
Hiring more people addresses the symptom, not the cause. The volume of manual work grows with the business, and human error does not disappear by adding more people to the same process.
There is a third path that few companies consider: connecting what already exists.
How to close the gap without replacing anything
An AI agent designed for supply chain acts as an integration layer between the ERP and external data sources — spreadsheets, supplier emails, order portals, EDI files. It does not replace the ERP. It connects to it.
The agent can execute specific tasks:
Automatic inventory reconciliation. It cross-references stock recorded in the ERP with data from the warehouse or suppliers, identifies discrepancies, and generates an exceptions report. What a person does today in three hours, the agent executes in minutes and delivers in the format the team already uses.
Purchase order tracking. The agent monitors the status of open orders, cross-references against supplier confirmations received by email or portal, and alerts when there are deviations in date or quantity. The team stops chasing information and starts managing exceptions.
Stockout projection. With access to historical consumption data from the ERP and current stock levels, the agent calculates which SKUs are at risk over the next 15 or 30 days and generates a replenishment proposal. It does not make the decision — it prepares it so the responsible manager can validate it in minutes, not hours.
A concrete example
An industrial manufacturing company operating in Spain — 80 employees, Dynamics 365 ERP, three-person supply chain team — was spending approximately 20 hours per month reconciling inventory between the ERP and warehouse tracking spreadsheets. In addition, the purchasing manager was investing between 6 and 8 hours per week manually following up on open orders with suppliers.
With an agent connected to the ERP and corporate email, inventory reconciliation moved to running automatically three times per week. Order tracking was automated with exception-based alerts. The team reduced the time spent on those tasks by approximately 25 hours per month.
In direct cost terms, the estimated saving is around 700–900 € per month. The real impact was different: the purchasing manager recovered time to negotiate terms with suppliers — something he had not been able to do in months. In the first quarter, those negotiations produced improved conditions on two contracts, representing a significant additional saving.
The agent was in production in six weeks. The ERP was not touched.
What is required to implement it
The minimum condition is that the ERP has some form of data access — an API, a queryable database, or even scheduled exports. In most modern ERPs, that capability exists even if it has not been activated.
The implementation process follows three steps: first, map the workflows where manual work is greatest; second, define what information the agent needs and where it obtains it; third, build and validate the agent with the team that will use it.
The team does not need to know how to code. They need to know what decisions they make today and with what data. That is sufficient to design an agent that works.
Conclusion
The gap between the ERP and spreadsheets is not a technology problem. It is a process problem with a calculable monthly cost. And it has a solution that requires replacing nothing.
If your team spends hours each week reconciling data across sources, that time has a cost. The relevant question is not whether automation is possible — it is how much it is costing you not to have implemented it yet.
If you would like to run that calculation against your own operation, we can do it in a short call.
→ Request a free diagnostic
