Key Takeaways
- Retrofitting aging equipment with controllers to generate data
- Data first, then agents
- Bringing skilled workers in as validators
- Can you count production volume and defect rates by equipment right now?
Sizzle approaches AX for small and medium-sized manufacturers by installing its own custom-designed controllers on aging equipment such as presses and welders to generate production and quality data. In a September 24, 2026 report by Gukje News, Sizzle CEO Lee Ji-hyun emphasized that domestic manufacturing is facing growing difficulty amid competition with China and the United States, and that SMEs need practical solutions they can apply immediately.
Where should a factory with no existing data even begin?
CEO Lee Ji-hyun founded Sizzle in 2016. Gukje News reported that the impetus came from witnessing decades-old equipment in aging factories churning out defective products. Here is how Sizzle generates data on the factory floor:
- It installs proprietary controllers, designed in-house, on aging equipment such as presses and welders.
- The controllers measure the equipment's production output and quality, generating data.
- High-precision data on equipment movement and operational status accumulates at a scale of tens of billions of records.
Gukje News reported that adopting Sizzle's technology allows companies that previously had no process data whatsoever to secure key data, and that AI-driven analysis of that data enables process innovation without large-scale investment. CEO Lee stated, "Manufacturing is a foundational industry for our country, but it still often runs on guesswork," explaining that building a foundation for systematically assessing a company's situation through digital transformation is both Sizzle's goal and the starting point of AX.
How do you turn skilled workers' know-how into data?
CEO Lee stated, "The core task isn't simple automation, but converting the accumulated intuition and know-how of manufacturing into data." Know-how such as gauging lubricant levels or equipment condition just by listening to machine sounds used to be critical, but the aging workforce and labor shortages are reducing the pool of skilled workers. Sizzle's technology focuses on using AI to validate and reinforce the experience and intuition of skilled workers.
CEO Lee said, "AI isn't replacing skilled workers—it's turning their know-how into an even more powerful asset." There was some pushback from workers during the initial rollout on-site, and CEO Lee has continued to build trust by spending time with frontline workers.
What does Sizzle's AI agent actually do?
As of September 2026, Sizzle is focused on developing an 'AI agent.' This goes beyond simply providing data—it's a stage where AI analyzes factory conditions on its own and supports decision-making. Examples cited in the report include the following.
- When a plant operator asks about the cause of a rise in defect rates, the AI agent analyzes process data to pinpoint the source of the problem down to a specific piece of equipment or product line.
- It also suggests, in real time, the causes of declining production and directions for improvement.
Gukje News highlighted that this agent can access in-plant data directly, allowing it to make judgments specifically tailored to actual manufacturing floor conditions. CEO Lee explained, "We're building a structure where AI analyzes the situation and flags it proactively, so plant operators don't have to keep walking the floor checking everything manually." Sizzle plans to expand its technology's application beyond manufacturing into industries such as smart farms, smart cities, and carbon reduction infrastructure.
Implications for AX at Korean Companies
Sizzle's approach starts with generating data from equipment and leads into an AI agent that analyzes that data. If you're considering an AI agent before you even have process data in place, it's more sensible to first identify which equipment will supply the data the agent will rely on for its judgments.
If converting skilled workers' know-how into data is the core task, the people who hold that know-how need to be part of the project. If you're preparing to adopt this approach, it's better to build in, from the outset, a role for skilled workers to validate the AI's analysis results.
Check whether your own factory can count production volume and defect rates by equipment right now. If you can't, the first decision to make is which equipment to start generating data from.