AI Models Transform Supply Chain Operations

Automation within supply chains seldom captures public attention, yet it determines whether products reach customers on schedule, whether invoices are processed accurately, and whether seasonal items appear on store shelves. Behind the scenes, a century-old enterprise and a fledgling startup may appear worlds apart. Dr. Oliver Dlugosch, co-founder and chief commercial officer of GeneralMind, described why many logistics functions still rely heavily on manual effort and how large language models could transform the picture. The conversation was recorded for the latest episode of Behind the Click, hosted by Janine Heinrich.

Building an empire on an air mattress

Dlugosch’s professional path did not begin in online retail. After earning a degree in physics, he moved into management consulting and later spent a brief stint in operations at Fortuna Sulov. That position revealed that his analytical abilities were valuable not only for high-level strategy but also for day-to-day tasks. In 2020, a former classmate, Tushar, approached him with a proposal for a new venture called Razor Group. Dlugosch examined the plan for any fatal flaw and found none that would make it unworkable.

From Physics to Entrepreneurship

He admits that launching a company proved the toughest of his three career arenas. While physics felt intuitive and consulting was a natural extension, entrepreneurship demanded immediate action with minimal room for error. The founder’s anecdote fits that narrative: on Razor’s inaugural day of operations, he slept on an air mattress in the basement office because he had no accommodation in Berlin.

Read Also: Challenges of driving from Geneva Airport to Val Thorens

Razor Group proceeded to purchase more than three hundred small e-commerce outfits, many of which operated through Amazon’s Fulfillment by Amazon (FBA) service and moved a few hundred items daily. Dlugosch notes that this common platform made the aggregation model feasible, as Amazon already managed a large share of distribution and customer outreach. After the acquisitions, Razor streamlined the supply chain by consolidating suppliers, negotiating with larger carriers for better freight rates, and centralizing warehousing, content creation, and advertising functions.

The original founders exited after a transition period of four to six weeks, leaving Razor to run the businesses as a single integrated entity rather than a collection of subsidiaries. The first ten to twenty purchases represented uncharted territory. Roughly half a year later, the onboarding process evolved into a repeatable, assembly-line-like system with clearly defined steps and checkpoints that no longer required founder involvement.

The Central Nervous System of a Firm

A key early decision enabled this rapid scaling. Only about four months after the company’s inception and before generating revenue, Razor implemented a full enterprise-resource-planning (ERP) system. Dlugosch describes an ERP as the central nervous system of a firm, linking product movement with invoicing and payment processes. Critics called the timing reckless, but he maintains that without the system the business would have been “dead by 2021”, overwhelmed by complexity that spreadsheets could not manage.

Even with the costly ERP, Razor’s logistics network still employed 150 to 200 staff members. The majority were not executives or strategists; they handled procurement, import logistics, warehouse oversight, order processing, and accounting. Their duties involved responding to supplier inquiries, following up with freight forwarders and customs, and preparing necessary paperwork. The team believed this workload was unique to Razor, yet it proved to be a common industry challenge.

Read Also: Shield marks half-century with new UK campaign

Manual workflows also introduce significant risk. Dlugosch illustrates the point with a seasonal product such as Halloween decorations: a single missed email can cause a container to miss its sailing slot from Shanghai to Hamburg, forcing the cargo to sit idle for a year and potentially costing millions. When large language models became commercially available, the team recognized an opportunity to automate these repetitive tasks. GeneralMind now specializes in helping sizable enterprises embed AI precisely where it can alleviate operational bottlenecks.

Embedding AI Into Supply Chains

According to Dlugosch, GeneralMind targets any activity tied to the flow of goods or financial transactions. A typical scenario involves transmitting a purchase order to a supplier. Although the act sounds simple, someone must verify receipt, capture the supplier’s response, and log the outcome. Modern LLMs can read inbound emails, pull relevant data from PDFs and Excel sheets, validate the information, and feed it directly into the ERP.

The ERP remains the authoritative source of truth for supply-chain status. Dlugosch cites the extraction of data from a document and its cross-check against another system as a task that no human should perform manually. Matching a PDF file to ERP records is now a capability of automated tools. He anticipated strong pushback from workers but observes that resistance is uncommon.

Read Also: Wenzel’s launches autumn and Halloween menu

He offers several explanations. Many firms struggle to locate personnel willing to perform routine comparisons between PDFs and ERP entries. Some clients rely on teams nearing retirement, such as accountants who possess deep knowledge of how to post complex invoices correctly. The loss of that expertise would be detrimental, and AI can help retain it. Operators also experience relief: rather than sifting through hundreds of Outlook messages, they can concentrate on exceptions that truly matter.

Dlugosch does not claim that automation will preserve every job. He argues that technology inevitably reshapes employment, and a company that cannot reduce costs may face an even harsher fate. Dlugosch’s core message is that AI is far from a set-and-forget solution; it handles straightforward cases and frees staff to tackle the more complex ones. Human oversight, review, and critical questioning remain essential, as decades of experience are better spent on strategic decisions than on extracting a purchase-order number from a PDF.

A Natural Next Step for the Founders

The core team behind Razor Group is largely the same one now building GeneralMind, which is unusual given how often founder breakups sink companies. Dlugosch credits three things: complementary skills, mutual trust, and shared pain. GeneralMind is a natural next step because the founders are solving the problems they lived through at Razor. Two or three additional founders who were not at Razor have since joined and brought new capabilities.