Overview

New Smart Agent is an intelligent edge assistant that connects seamlessly to equipment. Embedded at the machine side, it reads production and equipment sensor data directly and brings together user manuals, anomaly-response SOPs, and frontline expertise. Inference runs locally on a Small Language Model (SLM), so data never has to leave the plant. Teams can ask questions in natural language, retrieve data and troubleshooting guidance, and receive scheduled reports or proactive anomaly alerts—spending more time on decisions and action rather than searching for information.

Core Capabilities

• Equipment Troubleshooting – Instant SOP Retrieval: When an anomaly occurs, frontline personnel can quickly retrieve the correct troubleshooting SOP from an AI knowledge base built from user manuals, anomaly-response procedures, and expert notes. Enter an alarm code or symptom keyword to locate the matching steps and reduce downtime, while senior engineers focus on maintaining the knowledge base instead of answering routine questions.
• Natural-Language Data Queries – No Pre-Built Screen for Every Question: Ask about yield, OEE, production status, or a specific time range in plain language. Unlike a dashboard, Smart Agent handles open-ended questions—simply define the meaning of each data field during deployment, and AI will interpret the question and retrieve the relevant data without predefined requirements or custom development.
• Proactive Alerts and Scheduled Reporting – Threshold and Rule Monitoring: Define anomaly thresholds and rules based on production data to monitor operations. When an anomaly is detected, the agent presents an incident summary, the matching response SOP, and recommended actions. Key information such as daily output, yield, and equipment status can also be summarized and delivered on a schedule.
• Centralized Multi-Site Management – Distributed Agents, Shared Compute: Multiple New Smart Agent units can be deployed on the same network and managed from a Central Management Console for unified deployment, updates, and monitoring. Edge nodes can run without an on-board SLM by drawing inference capacity from a shared LLM workstation, balancing management efficiency and hardware cost as deployments scale.
• Equipment Management Console – Smart Modules per Machine Role: From the Equipment Management page, administrators can attach the appropriate smart modules to each Smart Agent—such as equipment data query, SOP search, and anomaly alert notification—so every agent delivers AI assistance and alerting tailored to its machine’s role.
• User Management and Governance – Accounts, Roles, and Access Control: The User Management interface lets administrators create accounts, assign roles, and enable or disable users, ensuring that operators, engineers, and managers use New Smart Agent according to their permissions—supporting both security and governance.
• Human-in-the-Loop Review – Oversight on Critical Decisions: AI-generated recommendations, responses, and actions can be reviewed and approved by designated personnel before use, helping ensure that critical decisions retain operational oversight and frontline expertise.
• Product Highlights – Lightweight, Easy to Adopt, Customizable: A lightweight device that deploys directly at the machine side, with support for custom applications and knowledge management. Designed for easy deployment and maintenance to lower both the adoption barrier and long-term operating cost.