Industrial DevOps: The Foundation for the AIFactory of the Future

The industrial automation world has slowly standardized the usage of IP-based networking for industrial devices and there’s been a rapid increase in the usage of cloud-based services on the factory floor. This enables automation engineers to have unprecedented access to tools that would’ve been unimaginable a decade ago. This is moving industrial automation towards the AI Factory of the Future, which will be enabled by Industrial DevOps.

Industrial DevOps is a suite of tools that multiply the capabilities of engineers allowing them to work faster and more efficiently, while reducing mistakes and downtime.

Industrial DevOps includes capabilities such as:

  • File management
  • Code review
  • Backup/deployment
  • Verification and validation

Over the last few years, there’s been rapid innovation in the usage of language processing for Large Language Models (LLMs), which at scale have developed emergent abilities allowing them to do tasks that previously weren’t thought to be feasibly performable by computers. 

Just as Industrial DevOps accelerates the work of automation engineers through enhanced collaboration and visibility, it's also uniquely positioned to harness the power of recent AI advancements, revolutionizing manufacturing and automation and making the AI Factory of the Future possible.

Challenges Facing AI Advancements in Industrial Automation

Industrial automation has characteristics that make it more challenging to directly leverage modern AI tools. For example, the most popular industrial programming languages are graphical and the file formats are binary, both of which make them unable to be directly processed by LLMs because they are designed to process text. 

In order to take advantage of LLMs, there needs to be a mechanism to translate binary files into a text-based representation that is processable by these large language models. Similarly, for LLMs to converse with humans about graphical programming languages, there needs to be a rendering mechanism that allows the output of the LLM to be understandable by humans. Incidentally, these two mechanisms of text-conversion and rendering are the bedrock of Industrial DevOps and provide tracking and visibility of code changes throughout the life cycle of industrial equipment.

There’s also been recent advances in LLMs such as Retrieval Augmented Generation (RAG), which allows LLMs to efficiently search and process large amounts of data. However, most industrial automation sites do not properly manage their files, so the data isn’t accessible by humans or by LLMs. 

Even when file management or version control systems are in place, changes are frequently deployed manually to equipment without an approval or code review process resulting in untracked changes. The file management system is then out of sync with the code on the device, which necessitates a backup/deployment mechanism to synchronize the file management system with the running production hardware. Maintaining a well organized, up-to-date, repository of the files that control the production equipment is a prerequisite for effective usage of AI tooling to search and process this data.

Agentic AI in the Factory: Accelerating Automation with Industrial DevOps

The AI Factory of the Future vision is built on agentic AI. Agentic AIs, using protocols such as MCP (Model Context Protocol), can initiate actions without human intervention and could also be an extremely powerful capability in updating code on the factory floor. Instead of constraining AI tools to the boundaries of a chatbox, AI agents could autonomously collect information about a system and take actions to update code files that can change the behavior of the system. With Industrial DevOps in place, those changes would then be submitted for human review. 

Unfortunately most industrial automation sites do not have code review or automated verification and validation in place. Without this essential part of Industrial DevOps, there is no way to ensure that code changes are high quality and do not endanger human safety; this is absolutely critical for machinery that operates in close proximity to humans. 

Only when such measures are in place can the true potential of agentic AI be unlocked to safely and responsibly accelerate the ability of engineers to rapidly update production systems.

The industrial automation world is in danger of being left behind as adjacent industries are catapulted into the future by the meteoric rise of AI technology, unless it embraces Industrial DevOps. File management, code review, backup/deployment, and verification and validation are all essential tooling for engineers empowered by cutting edge AI technology advancements. To thrive in the age of AI, industrial automation must embrace the transformative power of Industrial DevOps.

‍