• Written by Sonali Subhadarsini, Product Manager

A maintenance manager's dashboard, once a promising tool for overseeing the upkeep of an organisation's assets, now gathers dust in a corner of the office, its potential untapped due to the limitations of traditional query-based systems. The shift supervisor, tasked with ensuring the smooth operation of the production line, still relies on manual checks and intuition to identify potential issues, rather than leveraging the power of data-driven insights. This is where agentic AI workflows for maintenance come into play, offering a more proactive and automated approach to managing maintenance tasks.

Defining Agentic Workflows

An agentic workflow is distinct from a chatbot in that it uses tools, maintains state, and works towards a specific goal, rather than simply providing answers to user queries. In the context of maintenance, this means that an agentic workflow can automatically generate work orders, assign tasks to technicians, and track progress, all while taking into account the current state of the assets and the maintenance schedule. The category manager, responsible for procuring spare parts, can also benefit from an agentic workflow that optimises inventory levels and minimises downtime.

The key to an effective agentic workflow is its ability to interact with various systems and tools, such as computerised maintenance management systems (CMMS) and enterprise resource planning (ERP) software. By integrating with these systems, an agentic workflow can access and utilise the data it needs to make informed decisions and take proactive measures to prevent equipment failures.

Human Approval and Oversight

While agentic workflows can automate many aspects of maintenance management, human approval and oversight are still essential in certain situations. Removing the human approval step prematurely can lead to unintended consequences, such as unnecessary repairs or misallocated resources. The maintenance manager, for instance, may need to review and approve work orders generated by the agentic workflow to ensure that they align with the organisation's maintenance strategy and priorities.

However, this does not mean that human approval is always required. In fact, an agentic workflow can be designed to operate autonomously in certain areas, such as routine maintenance tasks, while still allowing for human intervention and oversight when necessary. The production team can benefit from this approach, as it enables them to focus on high-priority tasks while the agentic workflow handles more mundane activities.

Maintenance Data and Dependencies

An agentic workflow for maintenance relies on access to various data sources, including work orders, asset history, spare parts inventory, and condition data where available. The accuracy and completeness of this data are crucial to the effectiveness of the workflow, as it uses this information to make decisions and take actions. The maintenance team, for example, may need to ensure that the CMMS is up-to-date and accurately reflects the current state of the assets.

The dependencies between these data sources are also important to consider. For instance, the availability of spare parts can impact the scheduling of maintenance tasks, while the condition data can influence the prioritisation of work orders. By taking these dependencies into account, an agentic workflow can optimise maintenance activities and minimise downtime.

Impact on Maintenance Operations

The introduction of an agentic workflow can significantly impact maintenance operations, particularly in terms of reactive versus planned work. By automating routine maintenance tasks and optimising resource allocation, an agentic workflow can help reduce the amount of reactive work, freeing up resources for more proactive and strategic activities. The maintenance manager can focus on developing a comprehensive maintenance strategy that aligns with the organisation's overall goals and objectives.

Another benefit of an agentic workflow is its ability to identify and address recurring issues with the same asset, even if they are described differently in the maintenance records. This can help reduce the likelihood of equipment failures and improve overall asset reliability.

Metrics and Evaluation

To evaluate the effectiveness of an agentic workflow, various metrics can be used, such as the planned versus reactive ratio, mean time between failures, and spares availability. These metrics provide insights into the workflow's impact on maintenance operations and help identify areas for improvement. The maintenance team can use these metrics to refine the workflow and optimise its performance over time.

By tracking these metrics, the organisation can also identify opportunities to improve its overall maintenance strategy and make data-driven decisions to support its goals. The agentic workflow can provide the necessary data and insights to inform these decisions and drive continuous improvement.

Limitations and Wrong Choices

While agentic workflows can be highly effective in maintenance management, they are not a panacea for all maintenance-related issues. In some cases, an agentic workflow may not be the best choice, such as when the maintenance data is incomplete or inaccurate, or when the organisation lacks the necessary resources and expertise to implement and maintain the workflow.

In these situations, alternative approaches, such as traditional query-based systems or manual maintenance management, may be more suitable. The organisation should carefully evaluate its needs and capabilities before deciding whether to implement an agentic workflow for maintenance.