How GPT-5 turns equipment procurement into a lever for US reindustrialization

# Leaders & Admins

# ChatGPT

# Awareness

# Work

Diagon uses AI to compress weeks of machine sourcing into explainable runs, helping manufacturers expand capacity and onshore production.

With GPT-5, Diagon founder and CEO Will Drewery is turning capital equipment procurement into a quiet engine of American reindustrialization. After nearly two decades buying machines for factories in places as different as Baghdad and Silicon Valley, Drewery kept seeing the same constraint: ambitious manufacturers could not add capacity without highly specialized equipment. But the process for finding that equipment was slow, opaque, and dependent on spreadsheets, trade-show brochures, and a few experts’ Rolodexes. The stakes became clearest during his nearly five years at Tesla. Drewery helped manage roughly $3.5 billion in capital-equipment spending for Model S, Model X, and Model 3 production in Fremont and at Gigafactory Nevada. Over that period, Tesla’s headcount grew from about 4,000 employees to around 120,000, with most of that growth in factory roles. The lesson for him was that filling plants with the right machines is the precondition for revenue, new programs and large numbers of jobs. Drewery’s later work at robotics company Bright Machines, a factory-built construction startup, as well as launch-vehicle maker Astra revealed similar patterns in procurement. Diagon’s mission is to help rebuild America’s domestic supply chain, and it focuses on industries like aerospace and defense, renewable energy, and AI data center infrastructure. Customers such as Relativity Space and Rivian come in with dense engineering requirements for tools like physical vapor deposition coaters, battery manufacturing lines, or CNC (computer numerical control) machining cells. Diagon captures those requirements, cleans up the specifications with GPT-5, and turns them into structured briefs that an AI system can reason over. From there, GPT-5 becomes a specialized procurement assistant. It scans the universe of machine builders, including trade-show exhibitor lists, niche industrial directories, and long-tail suppliers whose websites are hard to navigate. In addition, Diagon maintains a growing internal database of machine builders, product families, and detailed specs. Using retrieval-augmented queries, GPT-5 compares many parameters at once, such as part dimensions, materials, throughput, footprint, and process type, then returns a ranked list of candidate machines along with nearby options that become viable if a buyer adjusts one or two constraints. Work that once took engineers days of manual filtering collapses into a single, explainable run with agentic search. Diagon is also experimenting with AI agents that gather contact details, populate web forms, and help drive toward initial quotes. Increasingly, buyers ask Diagon to prioritize American-built equipment, so that every new manufacturing line strengthens domestic supply chains as well as future growth. Each machine is a piece of a revived domestic ecosystem, making it more feasible to build batteries, rockets, and compute hardware onshore, with more people working alongside the machines.