Investing in equipment but putting off investment in administration: A story about why I thought AI has a role to play there

Companies invest in equipment, but investment in administration is often put on the back burner.

This might be happening in quite a lot of companies.

Investing in equipment that directly leads to sales. Increasing production volume. Improving quality. Making it easier to run the shop floor.

I think it is natural to spend money there first.

On the other hand, the back office does not directly generate sales.

Attendance management. Accounting. Reconciliation. Daily reports. Verification tasks.

These tasks are necessary, but they tend to have a lower priority for investment.

I myself, when I wrote a proposal for introducing a system at my company,

received a reaction like,

"Is that really necessary?"

I understand the sentiment. That is because improvements in the back office are not as clearly visible as an increase in sales.

However, that does not mean they should be left alone.

Remaining on paper. Remaining as manual calculations. Remaining as people transcribing into Excel. Remaining as verification tasks that are dependent on specific individuals.

When these states pile up, time is lost behind the scenes of the workplace.

I believe that is exactly when it is worth asking AI.

Instead of suddenly installing an expensive system,

"Can't this task be made a little easier?" "Can't we make it smartphone-input?" "Can't we reduce manual calculations?"

Consulting with AI and trying to make it into a small form.

Recently, I received exactly that kind of consultation through an acquaintance.

A consultation where it is not yet clear if it will become work

The consultation is about the back-office operations of a company that produces and sells food products.

It is still at the stage of an initial inquiry. Honestly, I don't know if it will work out.

However, I believe that turning such vague consultations into proposals is an important part of business development.

From what I heard, the company invests heavily in equipment related to production.

Improving quality. Increasing production volume. Keeping the site running. They are putting proper effort into those areas.

On the other hand, it seemed that paper and manual work still remain in their back-office operations.

They hire more people during busy seasons. Attendance management is done with paper time cards. From there, they calculate by hand. They also want to make accounting reconciliation easier to understand.

Hearing this, I felt that "there might be room for improvement here."

Don't just suddenly say, "Let's introduce AI"

When I receive a consultation like this, I think it's better not to suddenly say, "Let's introduce AI."

What the other party is struggling with is not the fact that they aren't using AI.

What they are struggling with is paper-based attendance management. Manual calculations. Accounting verification tasks. Administrative work that increases during busy seasons.

Therefore, I don't think the first thing to convey is "Let's use AI."

What should be conveyed is, "We might be able to make your current manual work a little easier."

AI and apps are just tools. The goal is to reduce the administrative burden during busy seasons. It is to make it so that both the field staff and the administrative staff can work a little more easily.

First, I tried making a simple attendance app

Even if I explain it only in words, it's hard to get across.

"You can enter it on your smartphone," "You can check it in a list," "You can aggregate it."

Even if I say that, it's hard for the other party to visualize.

So, as a draft for the proposal, I tried using AI to create a simple attendance app.

Instead of paper time cards, use a smartphone or tablet to clock in and out.

Clock in. Clock out. Break start. Break end.

I believe that keeping it this simple makes it easier to use on-site.

This is not a finished product. It is merely a demo for proposal purposes.

However, to help the other party think, "We might be able to use this at our company too," it is easier to convey if there is a screen.

It is also important for it to be visible to the administrator

It is not enough for staff to just be able to input data.

What is important is how administrators and office staff can check it afterward.

In the case of paper time cards, you look at the stamped cards to calculate working hours. Check break times. Make corrections as necessary. And finally, link it to payroll calculation.

If this manual work happens every month, it becomes quite a burden.

Therefore, I also created a screen where administrators can check the entered attendance in a list.

Working hours. Daily wage. Remarks. Confirmation status. I have made this information visible.

This is also still a demo. But by looking at the screen, it becomes easier to imagine "to what extent the parts where you are manually calculating while looking at paper can be replaced."

Making it gather in a spreadsheet

In this demo, I have set it up so that the input content also gathers in a spreadsheet.

This is quite important.

This is because it is easier to start by making it visible in a Google Spreadsheet rather than suddenly introducing a large attendance management system.

Entered date. Work category. Clock-in/clock-out time. Break/working hours. Hourly wage. Daily wage. Confirmation status.

If this information is compiled into a list, it becomes easier to check later.

Staff enter data on their smartphones. Managers check it on their screens. The data remains in a spreadsheet.

When this flow is visible, it becomes much easier to convey as a proposal.

Have them look at existing case studies as well

I plan to have them look not only at demos but also at cases that are already being implemented.

This is about a daily work report app. It is a case of switching from paper daily reports to smartphone input.

There is also a story about a breathalyzer check app. This is a case where input that used to be done on office PCs at the start and end of the shift can now be done using a smartphone and a QR code. In this case, a preliminary estimate based on on-site interviews showed a projected reduction of approximately 362 hours per year.

By having them look at such cases, there is a possibility that they will feel that it might be applicable to their own work as well.

Being able to say, 'We are actually testing this with these types of tasks,' makes the proposal much stronger.

Wait for a reaction

Just because you made a proposal does not mean it will immediately turn into work.

The other party also has their own circumstances. They have a budget. They have timing. They might also have the feeling that things are working fine with the current method.

Therefore, first, I will have them look at demos and case studies, and then I will wait for their reaction.

If they show interest, the next step is to go and see the site. If the reaction is lukewarm, it might just mean that now is not the right time.

But even then, there is still something to be learned.

Where was the anxiety? Was it the cost? The operation? Getting it established on-site? If you understand that, you can use it for your next proposal.

If it goes well, go to see the site

If the reaction is good, the next step is on-site interviews. I believe this step should not be skipped.

How many time cards are there? How many people are added during busy seasons? How much time does the calculation take? Who handles the closing process? Who makes corrections when there are errors? Where do mistakes and verification tasks occur?

Unless you look at these things, you won't be able to make a proper proposal.

Understanding the workflow is more important than building an app with AI.

Demonstrating cost-effectiveness

To get hired for a job, you need to communicate the cost-effectiveness.

Just saying "it looks convenient" won't get things moving very easily.

For example, if time tracking takes 10 hours a month, at an hourly rate of 2,000 yen, that's 20,000 yen per month. That's equivalent to 240,000 yen per year. If it's 30 hours a month, it's equivalent to 720,000 yen per year.

It's not just about time saved; there's also the reduction of errors, the reduction of verification work, and the easing of the burden during busy seasons.

When you can show this with numbers, your proposal becomes much easier to convey.

And to demonstrate cost-effectiveness, you need to interview the people on the ground. How much time is it taking now? How many people are involved? Where are the mistakes happening? You can't turn this into numbers without asking these kinds of questions.

If it doesn't work out, I'll write an article about it anyway

I don't know yet if this current project will go well. There's a possibility it won't turn into a job.

But I think that's okay.

Even if it doesn't work out, "why it didn't move forward" is a learning experience.

Did it not resonate with the other party? Was the way the proposal was presented weak? Was the cost-effectiveness not communicated? Was the timing off?

Including those aspects in my records will be a learning experience for building my own business.

Rather than just writing about success stories, I want to leave a record of this kind of progress as well.

Perhaps this is what it means to create work

This consultation hasn't taken shape yet.

But perhaps creating work starts from places like this.

Listen to stories about what people are struggling with. Formulate my own hypothesis. Create a small demo. Show them examples. Observe their reactions. Go to the workplace. Organize tasks together. Calculate cost-effectiveness. Make a proposal. Deliver the solution.

I want to gradually build this process.

What I want to do is not to sell AI or IT.

It is to create a state where people in the field and in the back office can work a little more easily.

I don't know yet how this consultation will turn out. Whether it goes well or not, I will record the process.

There are still many tasks that remain paper-based or manual.

What I realized again after hearing about this consultation is that there are still many tasks that remain paper-based or manual.

Attendance management. Accounting reconciliation. Daily reports. Verification work. Repeating the same explanations.

These tasks have the potential to be reviewed on a small scale without having to introduce a large system all at once.

I am still feeling my way through this myself. But by listening to the stories from the field, consulting with AI, and shaping things on a small scale, I can check if they fit the actual work.

I believe that with this approach, back-office improvements can also be advanced little by little.

I will be sharing the practical content and the behind-the-scenes of creating teaching materials even before they are completed.

If you are interested in in-house AI education or operational improvements in the field, please be sure to follow me.

About the author of this note

Living in Onomichi, Hiroshima Prefecture. Over 10 years of experience in general affairs and human resources at a food wholesaler (annual sales of approx. 800 million to 1.2 billion yen, approx. 15 employees) and a food manufacturer (annual sales of approx. 6 billion to 9 billion yen, approx. 100 employees). Currently sharing AI utilization methods that can be used in the field of recruitment, labor, and operational improvement. Working with the theme of creating 'breathing room' for busy workplaces.

👉 Click here for my self-introduction