{"id":795367,"date":"2025-07-16T09:16:18","date_gmt":"2025-07-16T09:16:18","guid":{"rendered":"http:\/\/test.daliworks.net\/index.php\/2025\/07\/16\/generative-ai-in-smart-factory-en\/"},"modified":"2026-07-28T01:22:02","modified_gmt":"2026-07-28T01:22:02","slug":"generative-ai-in-smart-factory-en","status":"publish","type":"post","link":"http:\/\/test.daliworks.net\/index.php\/2025\/07\/16\/generative-ai-in-smart-factory-en\/","title":{"rendered":"How Can Generative AI Be Used in Factory Operations?"},"content":{"rendered":"

<\/h1>\n

Beyond Data-Driven Operations, Toward AI-Driven Decision-Making<\/b><\/strong><\/h2>\n

This strategy series builds on \u300c<\/span>Top 5 Smart Factory Trends for the Second Half of 2025<\/span>\u300d.<\/span><\/p>\n

In this installment, we cover the next stage of factory operations: ‘generative AI-based operating strategy.’<\/strong><\/span><\/p>\n

Generative AI Has Arrived in Manufacturing in Earnest<\/h2>\n

Led by ChatGPT, <\/span>LLM (large language model)<\/b><\/span> technology is spreading rapidly beyond text-based industries into <\/span>data operations and decision-making on the manufacturing floor<\/b><\/span>. In the second half of 2025, we will see AI that goes beyond simple anomaly detection and alarms to <\/span>summarize production status, draft work orders, and explain equipment conditions<\/b><\/span> in earnest.<\/span><\/p>\n

From data-driven operations \u2192 to natural-language AI interface operations<\/b><\/span> \u2014 the factory floor is entering a new stage of evolution.<\/span><\/h4>\n

5 Roles Generative AI Can Play in Manufacturing<\/b><\/strong><\/h2>\n

\"\"<\/p>\n

1. Automated Equipment Status Summaries<\/h4>\n

\u2013 Previously, people had to interpret sensor and instrument data,
\n\u2013 Now AI can automatically analyze and summarize: \u201cThis unit’s vibration currently exceeds the threshold, and bearing wear is suspected\u201d<\/p>\n

2. Auto-Generated Explanations of Anomalies<\/h4>\n

\u2013 Explains the \u201cwhy\u201d when an alarm goes off
\n\u2013 Example: \u201cThe vibration reading on the Line B motor is up 30% from last week and its temperature has risen, so a failure is possible\u201d<\/p>\n

3. Auto-Generated Work Orders and Reports<\/h4>\n

\u2013 Automatically writes up inspection results and process status summaries in natural language
\n\u2013 Example: \u201cJuly 3, 9:00 AM: minor vibration detected on the Line A conveyor due to belt wear. Inspection required.\u201d<\/p>\n

4. Summarizing Unstructured Data<\/h4>\n

\u2013 Summarizes and classifies text-based anomaly reports, maintenance histories, and quality inspection notes
\n\u2013 Grasp the overall issue context even without a dedicated staffer<\/p>\n

5. An Operations Q&A Agent<\/h4>\n

\u2013 Generates data-backed answers to natural-language questions
\n\u2013 \u201cWhich equipment is drawing the most power right now?\u201d
\n\u2013 \u201cWhich equipment failed most often last month?\u201d<\/p>\n

A Real-World Example \u2013 Thingplus + Generative AI<\/b><\/strong><\/h2>\n

\"\"<\/p>\n

Thingplus is an <\/span>AIoT platform<\/b><\/span> that collects and visualizes data from a wide range of sensors and equipment. Adding generative AI capabilities on top enables innovations like the following.<\/span><\/p>\n

\"\"<\/p>\n

What Are the Prerequisites for Adopting Generative AI?<\/b><\/span><\/strong><\/h2>\n

\u2460 Clean Equipment Data<\/h4>\n

\u2013 Data from sensors, PLCs, MES, and other sources must be well connected<\/p>\n

\u2461 Historical Data with Context<\/h4>\n

\u2013 Descriptive text is needed for maintenance records, defect causes, quality issues, and more<\/p>\n

\u2462 Learning the Site Context<\/h4>\n

\u2013 The LLM must learn process characteristics and each organization’s terminology to produce high-quality answers<\/p>\n

That is why pairing generative AI with <\/span>a platform with a solid data foundation, like Thingplus<\/b><\/span>, is the most effective approach.<\/span><\/h4>\n

Ready to Start Generative AI-Based Factory Operations?<\/b><\/span><\/strong><\/h2>\n

\u2705 Is equipment and process data being collected in real time?<\/p>\n

\u2705 Have on-site event and history data been accumulated?<\/p>\n

\u2705 Do templates exist for alarms, reports, and anomaly logs?<\/p>\n

\u2705 Can AI responses be customized to your company’s situation?<\/p>\n

\u2705 Can the LLM be deployed in line with internal security and data policies?<\/p>\n

\u201cData Does Not Speak \u2014 You Have to Make AI Speak for It\u201d<\/h2>\n

\"\"<\/p>\n

Generative AI is not a passing technology trend.<\/span><\/p>\n

It is <\/span>a new way of operating<\/b><\/span> that <\/span>translates complex data into interpretable language<\/b><\/span> and <\/span>automates the day-to-day work of running a factory<\/b><\/span>.<\/span><\/p>\n

In the second half of 2025, the heart of smart factory strategy comes down to <\/span>how you make your data speak<\/b><\/span>.<\/span><\/p>\n

\n

\u2705 Recommended Reading<\/p>\n

\n

Preparing for the Era of Rising Electricity Rates with FEMS<\/a><\/p>\n<\/blockquote>\n