A layman discussion on how to improve efficiency

Efficiency is a key performance index in any organisations spanning from education, service, manufacturing, space to medical and others.

All organisations want to pursue the highest efficiency they can achieve. Efficiency is identical to low-cost, low waste and high profit margins.

In this layman discussion about efficiency, we will discuss efficiency from two perspectives: technical related (such as machinery, operation and logistics aspects) and people related.

Because, technology and people always exist hand-in-hand, and we cannot separate them in the discussion.

Let’s go into the discussions.

READ MORE: Research how to: A practical guide

Improving efficiency considering both technical and people (non-technical) perspectives

Improving efficiency means we need to streamline our processes. Processes here includes both technical aspects (such as equipment, logistics, procedure and others) and people (both as an individual and as a group).

We cannot separate the discussion for these two aspects, technical and people, as they are tightly coupled. Hence, we should improve or streamline both technical and people aspects.

To streamline our processes, here are things we can do:

Identified and eliminate non-value-added processes

To streamline process, we should start by observing our processes, such as production processes and also people process (how these people work and collaborate).

We need to find what process steps do not add values to the object of the process.

From technical point of view, such as production lines, we can analyse what process steps which do not, or very little, add values to materials they process.

For example, maybe at some stages, the materials only wait to be processed. This stage is a non-value-added process and should be eliminated or at least reduced by, for example, streamline the process so that no buffers are needed.

Other examples of technical process streamlining are:

• Part handling or transfers are too long. Solution: we need to optimise trajectory or routing for part handling between processes.
• Offline quality inspections require to take-off parts from production lines. Solution: establish an in-line quality inspection process.
• Underutilised machinery that works under their capacity. Solution: increase demand to increase production load, reduce maintenance by implementing proactive maintenance, and/or optimise machine scheduling to avoid machine waiting time and improve machine utilisation.
• Non-optimised space in warehouse. Solution: optimise warehouse layout and optimise packaging geometry.
• Cluttered administration or paperwork processes. Solutions: implementing document automation and database to ease document searching and filing or automatic data insertion by using barcode or other technologies.

From people perspective, non-value-added activities are for examples long processing time to finish a task, repair and adjustment activities and transporting parts or components from one process to another manually.

To solve these issues, we can do the following:

• Long processing time to finish as task. Solution: we must improve skills of our workers so that they can finish their tasks faster or within a given time. Another solution could be improving the ergonomics (and/or work conditions) of tasks or processes performed by people.
• Repair and adjustment activities. Solution: this problem can also be reduced by training and increase job experiences on our worker.
• Transporting parts between processes. Solution: use additional tools to speed up part transfers or using automated systems for part transfer.

After all, workers are human. Hence organisations should give the workers a proper treatments and appreciations.

By doing these our workers can feel to be appreciated and are willing to put their best in doing their assigned tasks or jobs.

Find and improve the bottleneck point

There is a famous quote: “The strength of a chain is determined by its weakest link”.

This quote tells us to streamline any processes, both technology- or people-related and we should find the bottleneck on the process flow.

From technology perspective, the bottleneck is the slowest process or machine. Meanwhile, from people perspective, the bottleneck is the least skilful or experienced workers.

We can straightforwardly spot bottlenecks due to slow machinery. For example, we can relatively easily check the processing time for each machine to find the slowest machine.

From here, we can either improve the machine, improve the machining strategy of the machine, improve the design of the processed part or simply change the machine with the faster one.

Similarly with logistics or warehouse systems, we can inspect data to find what parts or components are the slowest to be supplied and try to find the root cause of problems, for example maybe due to a part shortage or due to inefficiency part routing.

To solve these problems, we can optimise buffer stocks, improve demands and supply forecasting, improve and optimise part routing, optimise part placement and other strategies.

However, spotting bottlenecks related to people is difficult!

It is not that easy to find who are the worker that is the bottleneck in our process. Here, we need to understand our worker and connect with them more personally.

Open communication between the management and workers is key to understanding what problem each worker has and what they need to improve their performance.

Maybe they need more training and simply need more time to get more experiences. We need to understand these issues.

Open dialogs should be promoted across our organisations to find problems the people have and find solutions for their problems.

Reduce frictions

We all know that friction is the enemy of contacts between moving parts or flows. All machineries will require lubricant to reduce frictions between two moving components, such as gears, shaft, pistons and other moving parts.

Commonly, lubricants for machineries are all set up by default.

In this aspect, we want to focus on frictions among people. When people interact each other, like moving parts, there will be frictions.

These frictions may be small or large depending on factors such as culture differences, language differences and other aspects. The larger the differences, the more frictions it will create.

When there are frictions between people, jobs or works cannot be efficiently and effectively done. Hence, reduce overall efficiency of an organisation.

The lubricant to reduce frictions among people is “manner” [1].

We need to have manner to collaborate with different people at work. Manner can be in the form of respect, learn to understand each other and listen to other people carefully.

Good manner should be promoted in all organisations. Manner can be built by for example more collaborations, more interactions and more time to meet together in person.

Low-cost automation

Yes, it is very clear. Automation can significantly speed up repetitive and boring tasks.

Indeed, this aspect likely relevant to technological aspects. The main idea is what and how to convert a manual task into an automated task, reducing human involvement and increasing machine utilisation.

We focus on low-cost automation that can be implemented in all types and size of industry, from small to large.

Typical tasks that can be automated at low-cost are those with characteristics as follows: non safety critical task, supporting of production operations, maybe temporary or for trial, need for a simple and one-off solutions [2].

Nowadays, tools for low-cost automation are available. Many low-cost compute devices from different vendor can be bought easily with affordable prices which a normal person can purchase.

For example, the most famous compute tools for low-cost automation are Arduino and Raspberry Pi. These devices can be bought for not more than than 100£ or even less.

There are more expensive devices with much more powerful compute power for artificial intelligence (AI) on the edge the above two devices, for example, Nvidia Jetson with price of around 200£-300£.

We should utilise these devices to apply automation that require a one-off solution, quick development, flexible at low-cost.

In general, low-cost automation is suitable for data capture and visualisation, data analysis, actuation and support system.

Some applications suitable for low-cost automation are as follows:

• Automatic monitoring system: Automatic buffer management so that when the buffer is less than certain numbers, it will directly give notification so that shop floor manager can take a preventive action to not stop production lines.
• Data capture and visualisation: Automatic measurement logging for quality inspection. For example, when a worker inspects a part with a manual measuring instrument such as vernier calliper, the result of the measurement is automatically logged into a database.
• Actuation: automated part delivery, part feeder and tool changer.
• Support system: work/sale order automatic system and database system, process and tooling simulation.

Data leveraging

Currently, data is central for advanced machine learning and artificial intelligence (AI) system.

When our organisations have been operating for some time, most likely our organisations have been collecting a large amount of data, for examples sales data and have accumulated large internal company knowledge and procedures.

With these data, we should leverage them and use machine learning, such as forecasting, to predict demand and supply and/or use machine learning for classification such as part defect inspection. By using this prediction, we can prepare for unforeseen situations such as part shortage.

Machine learning for forecasting and/or for classification is becoming accessible as the framework such as PyTorch and TensorFlow, and models are available to be used freely.

Agentic AI and LLM-based foundation models [3] are nowadays the hottest AI systems which disrupt the world. This disruption is started by the launching of ChatGPT by OpenAI.

These foundation models can understand human language and can generate text, answering a question, summary text and other open-ended tasks that human can do.

Meanwhile, agentic AI is an augmented LLM-based foundation models so that this agentic can extract data, have a memory mechanism and perform actions (accessing tools) such as opening a calendar and accessing email systems.

Nowadays, many open-source foundation model can be used, for example Llama from meta and other foundation model provider.

One of the most common uses of agentic AI is to build an internal retrieval augmented generation (RAG) system. With this system, an organisation can insert their accumulated knowledge and procedures into a database.

The RAG system can use an open-source foundation model, such as Llama as their agentic engine.

From this database, an RAG system can access information following a user query (prompt) and insert this information into their context and then provide relevant information that satisfies the query (prompt).

For example, agreement document can be inserted into a database, either standard or vector databases. This database then can be access by a RAG system such that users can query about specific information related to a specific contract.

For example, a user can prompt the system to ask the details and/or summary about the requirements of a specific contact with a specific client and check the valuation of the contract.

This search automation can significantly reduce search time which are a non-adding-value activities.

Contained variations

Variations or uncertainty is one of the main enemies of any organisations.

With uncertainty, we will have difficulty to make strategies and make decisions. However, we cannot avoid this situation.

Uncertainty is everywhere in our daily life, starting from traffic uncertainty, weather uncertainty, demands uncertainty, supply uncertainty to global economic uncertainty.

We must face uncertainty and should make effort on how to “contain” uncertainty or variations such that we are ready to response any unexpected situations.

Some ways for containing uncertainty are as follows:

• Utilise data and modelling methods, such as ML and statistical tools, to perform forecasting of future events. If we can predict there will be delays in part supply, we can mitigate part shortages by building more part buffer to keep our production lines running.
• Build situational awareness. For example, always following the situation of our industry. For example, if our industry is food and beverage, we need to keep awareness of information related to this industry such as what is happening with farming industries, what is happening with general health situations, what new policies the government are making and other types of information that directly or indirectly related to our industries.
• Keep our organisation agile. One of the best ways to face uncertainty is by having flexibility to counter act the uncertainty. For example, if new product demands emerge, we have enough expertise to design and manufacture a new product that can satisfy those demands or our production layout is reconfigurable such that we can responds to part demand variations.
• Upskill our people. This is also very important. We need to always upskill our people so that they can develop their knowledge both technical and non-technical. At the end of the day, our organisations are a collection of people with different expertise and capacity who collaborate together to achieve a goal. If our people have good knowledge, they will be more agile and flexible and ready to take new challenges.

Decentralisation for flexibility

The agility of our organisations to responds to any changes depends on how flexible our organisations are.

This flexibility is very difficult if we are a centralised organisation.

Centralised organisations are usually slow. All urgent decisions need to be propagated up to few people. This situation will cause delay to respond to urgent situations.

We must make our organisations to be decentralised. This including our, for example, production systems and warehouse systems.

With the advanced of technology, this decentralisation becomes easier.

For example, at organisational level, we should give freedom to mid-level managers or branches to take decision to responds market demands in their regions without waiting a decision from the head office management.

By having a fast decision, the company can respond to demand variations promptly and will not lose opportunities.

Also from technological level, many edge devices have sufficient computational capacity such that the devices can store and run ML or AI models to independently react to their environments, such as to adjust machining parameters when in-situ variations, such as thermal and pressure, occur.

This in-situ response can prevent a machine or system damage at a large scale.

READ MORE: Research is like a puzzle: Start from anywhere to reach a goal

Conclusion

In this post, we briefly discuss strategies we can implement to improve efficiency in our organisations.

We discuss the strategy from both two perspectives: technology-related and people-related.

Technology and people are tightly coupled and hence we cannot separate these two aspects when talking about strategy for efficiency.

Finally, the advance on AI system nowadays can also play a central role to improve the efficiency in our organisations.

Reference

[1] Drucker, P.F., 2008. Managing oneself. Harvard Business Review Press.

[2] McFarlane, D., Hawkridge, G., Kaiser, J., Mukherjee, A. and Terrazas, G., 2024. Progress Towards Low-Cost Industrial Digitalisation for SMEs. IFAC-PapersOnLine, 58(19), pp.825-830.

[3] Acharya, D.B., Kuppan, K. and Divya, B., 2025. Agentic AI: Autonomous Intelligence for Complex Goals–A Comprehensive Survey. IEEE Access.

You may find some interesting items by shopping here.