GEA Group started its automation journey four years ago, initially targeting what many may deem the “boring stuff”—antiquated processes that were ripe low-hanging fruit for optimization via automation. We started with finance, automating journal entries, invoicing, and goods received processes. While simple, these initial solutions helped prove the tangible benefits of automation by translating our vision into actual value, laying the groundwork for further business investment and future automation growth.

To deliver on that promise of growth and return on investment, we needed to cast a wider net and onboard new functional areas with automation opportunities that would significantly shift the organization’s DNA and evoke a wave of innovation excitement. One such opportunity presented itself within the Operational Buying department, and after three months of careful process orchestration, the solution we designed went on to generate incredible impact beyond just hard cost savings.

THE PROBLEM

Operational Buyers work with multiple vendors to confirm critical aspects of purchase orders like the price, quantity, delivery date, and country of origin. These four fields define the buyer’s course of action. If there is a discrepancy with any field, the buyer must take different and specific manual actions to remedy the issue before proceeding. When talking about POs with 50 – 100 lines each, requiring manual review and remediation, that’s an extremely time-consuming task. Ultimately, errors in these commitments have implications on other departments downstream, on the final delivery to the customer, and on GEA Group’s profitability margins.

To recap, these were the major pain points:

· 7 – 10 minutes of manual review per line item, 50 – 100 lines per PO

· Price mismatches causing delay in payment processing

· Delay in receiving shipments due to customs discrepancies

· Delay in shipments to customer

THE APPROACH

We needed to unravel this complex process before we could determine the optimal approach to a designed solution. Starting from the beginning, we needed a solution that could liaise between multiple systems, catch and resolve an issue at the first instance, escalate a later issue to obtain a swift resolution, and finally update the PO.

THE SOLUTION

First, we identified our top 20 vendors who could change their PO confirmation format to meet a new template we laid out. Those 20 vendors represented 22,000 line items that needed verification. This was more than enough volume to get an impactful solution off the ground.

Our final automated solution follows this process:

1. Monitor the purchase orders in draft status

2. Look for the confirmation email

3. Verify details against the price and delivery date using our own LLM

4. If no discrepancies, update the PO as confirmed and no further action is needed

5. If there is a discrepancy, trigger an exception workflow

a. Appropriate stakeholder verifies information and approves with a button click

b. Discrepancy is updated in the PO

6. Monitor Fedex tracking status and update Customer Service team to update end customer

The solution is unattended unless the exception workflow is triggered. Next, we are working to integrate solutions for the customs process and paperwork in which each country of origin will trigger a different action and different paperwork.

I also mentioned we are using our own LLM, but we are eager to incorporate more generative AI into the end-to-end process. We believe once we do that, we will be able to apply this solution to up to 100 vendors and add even more capabilities to the solution. Our scope is to scale this solution 4x in our North America region.

IMPACT

As I already alluded, this use case was transformative in more ways than one. The pilot launch with 20 vendors and 22,000 line items saved 1,820 manual hours with a 100% accuracy rate—an impressive productivity gain for any business.

Beyond hard metrics, the process has also improved customer satisfaction by reducing delays and integrating proactive communication regarding shipment statuses. It also allows us to gather data on vendor performance. With challenges documented, GEA Group now has more power when it comes to renegotiating supplier contracts.

Finally, as the organization has realized the potential of end-to-end automated solutions that leverage the capabilities of generative AI, everyone wants more! Processes like this have been the key to altering GEA Group’s internal DNA. Now, we are witnessing colleagues eager to propel innovation in all areas and we’re reaping the benefits of a full pipeline.

My ultimate advice to you? Take the time to craft automated solutions that will make a difference, share your victories with the business, and the business will prop up your efforts. When stakeholders take a lead in your automation program, then new opportunities will keep coming.