AI and beyond: How frontier technologies are reshaping industrial operations

We stand at a pivotal moment in industrial operations, driven by the emergence of frontier technologies that will reshape the field. Artificial intelligence (AI) is at the heart of this transformation, powered by exponential advancements in computing power. AI is not only enhancing existing technologies but also paving the way for entirely new innovations.

A significant breakthrough in AI has been the development of multimodal foundation models, which have unlocked a vast array of new industrial applications. Multimodality extends AI's capabilities, enabling models to process text, sound, videos, and other data types that exist in manufacturing.

Among these, large language models (LLMs) such as the well-known ChatGPT-4, demonstrate AI's potential to transcend expert-only applications, making sophisticated AI accessible to everyone across industries.

AI agents are the next leap in AI. These agents amplify the impact of LLMs by granting them the ability to access tools and enhancing their ability to plan, prioritize, and execute actions. In combination with other technologies, AI agents unlock frontier technologies in operations: paired with robotics, they will make humanoids a reality; with digital tools, they enable autonomous systems.

These frontier technologies result in substantial benefits for productivity, flexibility, workforce empowerment and sustainability throughout the value chain. The potential is immense, but what will these applications look like in practice? How mature are these technologies, and what key challenges must be overcome for successful implementation?

AI agents are expected to transform operations in two ways. Knowledge agents will empower workers, while automation agents automate tasks in operations.

Knowledge agents leverage the entire data platform of the company to support workers and engineers in decision-making. These agents assist teams by providing insights and recommend required actions, tapping into relevant data sources and using any digital tool to achieve their goal.

The human–agent interaction is possible by speech or text, while multiple technologies can be leveraged for the visualization – for example, smart devices such as tablets and watches or immersive technologies such as mixed reality.

For example, a maintenance recommendation agent will combine existing capabilities from machine learning (ML) to predict failures, with specific repair instructions that it will get from various sources such as the machine handbook.

Additionally, it will be able to identify the root cause of failure, tell the worker on the shop floor the required actions and prepare the purchase order for the required part. Similar applications can also be developed for production scheduling.

Automation agents can execute automation tasks autonomously based on their reasoning and execution skills, enabling both virtual and embodied automation agents. According to a BCG report, autonomous agents can directly “tell” other enterprise systems what to do, fundamentally changing company operations and enabling more holistic automation, significantly reducing labour costs.

These agents automate software through their ability to independently access and use different tools. A widely known example is an automated travel booking agent. In operations, applications can include setpoint optimization, where a given machine parameter is automatically adapted by the agent to optimize performance.

It combines ML-based performance analytics and machine vision from the quality system with automatic update of machine parameters improving production in real-time. These agents can also be leveraged in production planning to automatically test scenarios for the production plan based on customer demand, inventory, and plant performance.

These agents transform physical automation by combining robots with situational understanding, reasoning, and execution skills. These new capabilities considerably expand the automation scope, overcoming existing challenges.

AI agents give robots the necessary flexibility to adapt to different environments and the dexterity to handle any object. Companies like Covariant have successfully implemented robots in distribution centers that can perform complex kitting tasks, picking and placing random items with high precision.

Beyond articulated robots, embodied AI agents enhance the capabilities of many robot types, including autonomous mobile robots (AMRs) for material transport and drones. Another breakthrough is the development of humanoid, bi-pedal robots that can perform human-like tasks.

Startups like Agility Robotics with Digit and Boston Dynamics with Atlas have launched these developments in recent years. While current applications are mainly demonstrations and pilots, further development is required for broader deployment.

AI agents only successfully work in combination with other technologies and underlying foundation layers:

Beyond these layers, the right computing power (both cloud and edge), network connectivity, and cybersecurity are essential enabling technologies.

Some challenges remain in deploying automation agents. The reliability and safety of these systems are concerns that tech companies need to address before large-scale deployment. In the meanwhile, companies still need to prepare themselves and consider these technologies in their mid-term strategies.

Understanding their potential impact for respective industries and how to enable their deployment at scale across supply chains, will be key to drive a successful and responsible transformation of the industrial sector.

To support manufacturers in understanding transformative impact of upcoming frontier technologies, the World Economic Forum in collaboration with Boston Consulting Group has launched the global initiative Frontier Technologies for Operations: AI and beyond, which builds on the learnings from the previous initiative on AI-powered Industrial Operations.

This new effort seeks to create an overview of frontier technologies that will reshape operations, deep-diving into AI, in collaboration with the AI Governance Alliance, as well as on other cutting edge technologies to assess their value potential and maturity by developing target scenarios for 2030.

The World Economic Forum invites manufacturing companies to join this initiative to collectively accelerate the transformation towards AI-powered industrial operations.