• Written by Muhammed Abdulla NC, CEO

This week, the operations director must decide where to cut costs without disrupting production, and the decision will be based on a detailed analysis of current spending and production workflows, rather than just a straightforward review of unit costs. The distinction between unit cost and total cost of an AI workflow is crucial in this context, as it is the total cost that reveals the actual financial impact of a particular process or system.

The unit cost trap

The unit cost of an AI workflow is often the primary focus of cost optimisation efforts, but this approach can be misleading, as it does not take into account the overall cost of the workflow, including factors such as the number of units being processed, the complexity of the process, and the resources required to support it. A judgement that prioritises unit cost above all else is wrong, because it ignores the potential for economies of scale and the impact of other factors on total cost.

A more effective approach is to consider the total cost of the AI workflow, including all the resources required to support it, such as personnel, equipment, and maintenance. This allows for a more accurate assessment of the financial impact of the workflow and identifies areas where costs can be optimised without compromising production.

Real levers for cost optimisation

There are three key levers that can be used to optimise the cost of AI workflows in operations: caching, batching, and model routing. Caching involves storing frequently-used data in a readily-accessible location, reducing the need for repeated calculations and improving processing speed. Batching involves processing multiple units together, reducing the overhead costs associated with individual processing. Model routing involves directing different types of data to the most appropriate processing model, reducing the risk of errors and improving efficiency.

These levers can be used in combination to achieve significant cost savings, but their effectiveness depends on a detailed understanding of the production workflow and the specific requirements of the operation. For example, a shift supervisor checks the maintenance log to determine the optimal batching schedule for a particular production run.

Dependent operations data

The effectiveness of AI cost optimisation in operations depends on access to high-quality operations data, including process and exception records, project status, budget consumption, and supplier performance. This data provides the insights needed to identify areas where costs can be optimised and to monitor the impact of changes to the production workflow. The category manager compares two suppliers based on their performance data to determine which one to use for a particular project.

Without this data, it is impossible to make informed decisions about where to cut costs and how to optimise the production workflow. The operations director must therefore ensure that the necessary data is being collected and made available to support cost optimisation efforts, and can use tools like operations management software to streamline this process.

What it changes

AI cost optimisation in operations can reveal anomalies that are only visible when considering the production workflow as a whole, rather than just focusing on individual components. For example, a particular process may appear to be operating within budget, but when considered in combination with other processes, it may be revealed to be consuming budget at a faster rate than progress. This allows for more targeted cost optimisation efforts and can help to prevent costly errors.

Projects that are consuming budget faster than progress can also be identified and addressed, reducing the risk of cost overruns and improving the overall efficiency of the operation. The operations director can use this information to adjust the production workflow and ensure that costs are being optimised without compromising production.

Measuring success

The success of AI cost optimisation efforts in operations can be measured using a range of key performance indicators (KPIs), including on-time delivery, exception rate, and expediting cost. These KPIs provide insights into the impact of cost optimisation efforts on the production workflow and allow for adjustments to be made as needed. For example, a reduction in exception rate may indicate that the cost optimisation efforts are having a positive impact on production efficiency.

By monitoring these KPIs and making adjustments to the production workflow as needed, the operations director can ensure that cost optimisation efforts are having the desired impact and that production is being optimised without compromising quality or delivery times. If further support is needed, the operations director can contact us for more information.

Limitations and wrong choices

AI cost optimisation in operations is not a silver bullet and there are situations where it may not be the best choice. For example, in situations where the production workflow is highly variable or unpredictable, AI cost optimisation may not be effective, as it relies on patterns and trends in the data to identify areas for cost optimisation. In these situations, other approaches to cost optimisation may be more effective.

In general, AI cost optimisation in operations is most effective when used in combination with other approaches to cost optimisation, such as lean manufacturing and total productive maintenance. By using a combination of these approaches, the operations director can ensure that costs are being optimised without compromising production, and that the production workflow is being continuously improved.