SUPPORT · FAQ
Frequently asked questions
Answers to the questions customers ask most about smart factory implementation, integration and operation, semiconductor and LED MES, and the manufacturing AI agent MESIA. Select a question to open the answer.
Implementation, integration & operation
What are the steps of an MES implementation?
The project runs in five phases: current-state assessment → target model → solution design → implementation and integration → stabilization and handover to operation. One team, the same team that develops our solutions in-house, carries the work from assessment to stabilization.
The duration depends on plant size, the number of operations and the scope of ERP and equipment integration. After checking your site in a free one-day on-site consultation, we propose the scope and schedule.
See smart factory implementation consultingDoes it integrate with our existing ERP and equipment?
fastMES comes with adapters for the SAP MM, PP, QM and SD modules, and we build interfaces to other ERP systems as well. fastWMS posts goods receipts and goods issues to ERP documents through real-time and batch interfaces, which removes double entry.
Equipment is connected through SECS/GEM for semiconductor tools, and through fastHMI's Modbus TCP and OPC UA data acquisition for PLCs and measuring instruments.
Learn more about fastMESCan you implement and support plants outside Korea?
We have delivered 27 projects in seven countries: Mexico, the United States, China, India, Vietnam, Malaysia and Serbia. Instead of building a separate system for each plant, we can define a standard business and data model and roll it out across multiple overseas plants.
After go-live, engineers at our headquarters provide remote support through the same process as in Korea. When a shop-floor supervisor makes a request in the local language, we respond in that language.
See overseas remote supportCan the project be combined with the Korean government's smart factory support program?
Yes. Haesol Information Systems is a registered supplier under the Korean government's smart factory support program. We define the implementation scope to fit the program type, implement and pass acceptance against its requirements, and continue with later enhancements. In 2017 we were selected as an excellent smart factory supplier case by the Ministry of Trade, Industry and Energy.
See government support program servicesHow do we get maintenance and support after go-live?
Submit a request through our maintenance management system (as.haesol.net) or call +82-31-296-2270. The assigned engineer is notified at once by email and text message; we resolve issues remotely first and visit the site when needed.
The engineers who developed and implemented the product support you directly. We provide scheduled preventive inspections, patches and upgrades, and operator training, and every request and action is kept as a service history.
Submit a maintenance requestSemiconductor & LED MES
How is a semiconductor or LED MES different from a general MES?
Where a general MES manages results by work order and lot, a semiconductor MES has to trace down to the wafer, carrier, lot and chip, and manage lot splits and merges, routes and recipes by operation, and equipment process parameters together with metrology data.
fastMES for Semiconductor is an MES built for semiconductor processes. On top of fastMES, it adds front-end and back-end (FAB and packaging) management, route and recipe management, SECS/GEM equipment integration, probe test and sorting data collection, and customer-specific labels and test reports.
Learn more about semiconductor MESHow are wafers and chips traced?
Process data from wafer receipt through front-end processing, probe testing, back-end processing and packaging is recorded at wafer level and down to the smallest unit, the chip.
You can trace backward from a chip to its wafer, carrier and lot, and forward in the opposite direction. When a yield excursion occurs, you can cross-check process, equipment and lot history to locate the abnormal window.
Learn more about semiconductor MESDo you support SECS/GEM equipment integration?
Yes. Status, process parameters and metrology data from equipment connected over SECS/GEM are collected automatically and recorded in the lot history, and measured values are dispositioned automatically against the specification.
Equipment utilization, maintenance history and usage counts of durables such as probe cards and reticles are managed as well.
Learn more about semiconductor MESCan it be applied to LED and optoelectronic processes?
Haesol Information Systems started in 2010 with an LED semiconductor MES implementation. We support PSS, EPI and CHIP process management, probing and sorting data collection, and yield analysis based on chip characteristics, and we have implemented a semiconductor MES for a laser diode plant in China.
See our historyManufacturing AI · MESIA
How is MESIA different from a general generative AI chatbot?
Haesol's AI Agent is not a separate chat window. It is a manufacturing AI agent that works alongside fastMES screens, looking at the same data. Questions are interpreted by MESIA (MES Insight AI), a structure developed in-house by Haesol, and answered from the plant's verified records.
Beyond answering questions, it detects process exceptions and delivery delays first and briefs managers, and when operators report by QR code, voice or photo, the reports are confirmed and then posted to production results.
Learn more about AI AgentHow does it keep the AI from making up numbers (hallucination)?
When generative AI writes SQL directly, the query may run yet return figures that differ from your business definitions. MESIA first structures each question into an intent representation (IR) of metric, dimension and time period, then executes it only through a knowledge catalog and verified SQL templates (Golden SQL).
Figures come from the database and explanations from the LLM, every sentence in an answer is linked to its source records, and when a question is ambiguous it asks back instead of guessing. We have filed a patent application for this manufacturing data question-answering AI (December 2025).
See the MESIA processing architectureDoes plant data go out to external AI services?
It is not sent to external general-purpose AI services. The default configuration is on-premises: LLM inference runs on GPU servers inside the customer's plant, and the vector database for document and history search and the data catalog also stay inside the customer's network. In this default configuration, production, quality and equipment data do not leave the plant.
If it is difficult to install GPU servers on site, the system can also be configured to use Haesol Information Systems' cloud servers, if you choose.
See the AI Agent operating environment