Beyond RAG

In energy and manufacturing, vast volumes of unstructured data (think OEM manuals, maintenance logs, shift notes, correspondence, procedures), sit largely untapped. For decades, experienced technicians have compensated by carrying critical knowledge in their heads. But with retirements accelerating and fewer seasoned workers on the front line, this model is breaking down. 

New large language learning models that underpin technologies such as Grok and ChatGPT are being trained on this unstructured content to create context-relevant, queryable databases for industry. This technology, referred to as retrieval-augmented generation (RAG), could help unlock hidden knowledge across sprawling document sets.

Early attempts at RAG have certainly improved search, a task that consumes hours of scarce engineering time. However, companies quickly learned that speed and accuracy fall apart at scale, context matters, and lack of trust in the output leaves users frustrated and skeptical. 

The real opportunity lies in pairing RAG with agent-based AI systems designed for complex, asset-intensive environments. By reducing mean time to repair (MTTR), cutting rework, and extending the interval between failures, these solutions directly recover lost production capacity, which is an eight or nine-figure problem in many enterprises. For younger, less experienced workforces, AI tools are a critical equalizer, levelling the field against looming labor shortages.

In this episode, I speak with Mark Fosdike, CEO and co-founder of Datch, about how his company is pioneering AI-driven diagnostic agents for manufacturing and industrial clients. We explore the realities of implementing RAG in high-stakes industries, the economic drivers behind adoption, and why obsessing about customers’ problems is the key to success.

---

👤 About the Guest

Mark Fosdike is the CEO and co-founder of Datch, an AI company building agent-based diagnostic solutions for asset-intensive industries. With a background in aerospace engineering, shipbuilding, and manufacturing systems, Mark has spent his career navigating the complexities of the industrial value chain. His vision for Datch was born from frustration with traditional maintenance practices and a desire to put AI to work solving real-world challenges on the plant floor.

• LinkedIn Profile