AI in Manufacturing: Taming Intelligent Assistants to Stay Ahead of the Curve

Written by Arkadiusz Rataj | Aug 26, 2025 3:00:00 AM

In the fast-paced world of manufacturing, where innovation is paramount, Artificial Intelligence (AI) has moved beyond a futuristic concept to become a crucial tool supporting Human Intelligence (HI). In a recent episode of the "No Lids, No Limits" podcast, Arek Rataj and Kin Leong took us on a fascinating journey through the world of AI tools, dispelling the myth of "AI replacing humans" and highlighting its role as an exceptionally capable assistant.

Kin Leong, a seasoned professional with over 30 years of experience in manufacturing operations, emphasizes that instead of fearing a Skynet or Terminator scenario, we should focus on how AI can help every employee – from engineers to technicians and managers – to excel at what they do. He believes AI should be viewed as a "very smart intern".

ChatGPT and Gemini: Versatile Everyday Assistants

ChatGPT from OpenAI and Gemini from Google are powerful, versatile tools that are excellent for daily tasks. Arek and Kin agree that these Large Language Models (LLMs) are invaluable for:

  • Content Creation: They assist in writing persuasive sales emails, overcoming writer's block, or drafting blog articles. Kin notes that AI can draft a compelling email that would take him half a day to write.
  • Rapid Data Analysis: AI efficiently processes vast amounts of information, identifying relationships and anomalies that might be invisible to humans. This includes detecting a machine falling out of calibration or patterns in shipment delays from specific regions.
  • Acting as a "Brain": They can remember and build upon information, serving as a knowledge base.

Just as spreadsheets revolutionized number crunching, AI tools are transforming data optimization.

The Art of Crafting Effective Prompts

The key to unlocking the full potential of AI tools is the ability to formulate effective prompts (instructions). Kin Leong suggests including phrases like "think as if" or "act as if" at the beginning of a prompt to assign the AI a specific role, which significantly improves the quality of its responses. Arek adds that it's beneficial to ask the AI to ask clarifying questions until it is 95% sure it can execute the task.

An interesting, albeit accidentally discovered, trick by Kin is to use one LLM to create a better prompt for another. This "recursive magic" can turn frustration into "very beautiful" results.

Data Privacy: Secure Your Information with NotebookLM

While versatile tools like ChatGPT and Gemini are powerful, they come with serious risks related to data security and privacy. Uploading confidential data to these models means it could be sent to external servers and potentially used for future AI training, risking breaches of Non-Disclosure Agreements (NDAs). Kin describes this as a "scary thing".

This is where NotebookLM – Kin Leong's favorite AI tool – comes in. Designed as a research tool, it allows users to upload their own reference sources – documents, PDFs, or clipped websites – and ask the AI questions solely based on that provided information. This ensures data remains in a secure, controlled environment, acting like an intern working only with company-approved materials.

NotebookLM is particularly useful for:

  • Converting data into actionable insights.
  • Creating audio and video summaries from lengthy PDF documents.
  • Combining multiple data sources (e.g., machine maintenance schedules, employee rosters) to identify correlations that would be difficult to spot traditionally.

This tool empowers an entire team to act as one and eliminates the need to rely on specialized IT or engineering teams for specific analyses. However, it's crucial to remember the "garbage in, garbage out" rule: clean, high-quality data is essential.

Practical Steps for AI Implementation in Manufacturing

Before a company dives into AI, several key considerations should be highlighted:

  • Prioritize Value and ROI: Any AI tool must bring real value, generate business, acquire new customers, or better engage existing ones. Avoid tools that only create more work or unnecessary data.
  • Process Improvement First: AI is not a magic fix for broken processes. It's crucial to identify and improve existing processes before implementing AI. For instance, an inventory problem is typically a supply chain issue, not solely an AI problem.
  • Target "Low-Hanging Fruit": Start by applying AI to easy, time-consuming, and repetitive tasks. This aligns with Pareto's Law: 20% effort on these tasks can yield 80% return.
  • Clean Data is Crucial: The principle of "garbage in, garbage out" strongly applies to AI. High-quality, clean data from all sources – including customers (e.g., Bills of Material) and vendors (e.g., lead times, pricing) – is essential for successful AI implementation. AI can also help reveal and utilize previously unused data locked within legacy systems like ERPs, IoT devices, and manufacturing machines, aligning with Industry 4.0.
  • Integration and Compatibility: Consider how AI will integrate and maintain compatibility with existing legacy systems.
  • Involve the Team Early: This builds ownership and accountability.

The Future is Collaborative

AI is considered the most disruptive technology since the internet. Organizations that do not explore its deployment risk falling behind. The adaptability of AI, combined with human critical thinking and strategic capabilities, will accelerate progress exponentially. The goal is to leverage AI for good, streamlining processes and ultimately achieving better, faster, and cheaper operations.

OpenJar Tech, the company behind the "No Lids, No Limits" podcast, emphasizes solving problems and delivering real value without fluff. They assist manufacturing leadership teams with strategic planning, process (re)design, IT implementation, and data analysis, transforming data into actionable insights. The podcast serves as an open forum for sharing ideas and solutions on critical manufacturing topics, inviting audience engagement to guide future discussions.

Want to learn more? Listen to the "No Lids, No Limits" podcast and ask your questions to help shape future discussions!.