Employee AI Usage Capped at $2,000 Per Month. Atlassian Starts Visualizing ‘AI Costs’

Atlassian has begun full tracking of employee AI usage, setting a monthly cap of up to $2,000 per employee in a ‘wallet’ system that comprehensively monitors who is spending what and on which services.

This translates to a maximum annual cost of $24,000 per employee (approximately 3.6 million yen). For a company with 1,000 employees, this means an annual cap of 3.6 billion yen on AI usage costs. For large enterprises, AI has become a cost center that needs to be managed, rather than a ‘free tool’ that can be used without limits.

However, the question to consider is this:

‘What should small businesses do now that large corporations are rushing to track and manage AI usage?’

The answer is clear: it is not to simply mimic the same approach. Small businesses have their own ways of doing things.

The ‘Monitoring’ of Large Corporations vs. the ‘Systematization’ of Small Businesses

Atlassian’s approach is essentially ‘monitoring to prevent overuse.’ In organizations with thousands of employees, it becomes difficult to see who is doing what, hence the need for a tracking system. This is a reasonable decision.

But is there any point in doing the same in a small business with 10 to 50 employees? No.

The problem for small businesses is not ‘overuse’ but rather ‘inconsistent usage.’

• Employee A is using ChatGPT to summarize meeting minutes.
• Employee B is unaware of the existence of the same tool.
• Employee C is pasting confidential information directly into external AI.

This situation is the most dangerous. It is not a cost issue, but rather the quality and risks becoming personalized that is the problem.

Large corporations solve this through ‘tracking.’ Small businesses solve it through ‘systematization.’ If this distinction is missed, investing in expensive monitoring tools may lead to the worst-case scenario where nothing changes.

The Risks of ‘Leaving It All to AI’ Are Evident in the Numbers

The need for systematization is backed by numbers, not just intuition.

An analysis by GitClear of over 53,600 code editing data points revealed that much of the code generated by AI required human revisions. Cases where AI-generated code could be deployed directly in production are limited, leading to additional review and correction efforts.

In other words, using AI does not necessarily lower costs; rather, costs can increase depending on ‘how AI is used.’

Another point: research on AI-based resume screening has confirmed that tools claiming to be ‘fair AI’ still exhibit biases similar to those of humans. The belief that delegating judgment to AI will result in fairness is a fallacy.

These issues may pertain to large corporations, but they are not irrelevant to small businesses. In fact, small businesses, which often have weaker oversight, may suffer greater damage. A single email sent to a client based on AI output could lead to a loss of trust.

Three Steps to Start ‘Systematizing AI Usage’ at Zero Cost

There is no need for a tracking system like that of large corporations. What small businesses should do is systematization that can be initiated with just one sheet of paper and a 30-minute meeting.

Step 1: Summarize ‘What Information Can Be Shared with AI and What Cannot’ on One Page

The first step is to clearly define usage rules. However, there is no need to create a lengthy guideline. One A4 sheet is sufficient.

You only need to write three things:

1. Information that can be input into AI (public information, general internal documents, etc.)
2. Information that cannot be input into AI (customer personal information, unpublished financial data, contract details, etc.)
3. Tasks where AI output can be used directly and tasks that require human review.

With just this, you can prevent incidents like Employee C pasting confidential information into ChatGPT. The cost is zero, and it takes only 30 minutes.

Step 2: Decide on Three Effective Ways to Use AI in Your Company

Next, you need to ensure that everyone can use AI in the same way.

The key is not to say ‘let’s use AI for everything,’ but rather to narrow it down to ‘let’s use AI in these three tasks this way.’

For example, a local manufacturing company (20 employees) decided on the following:

• Drafting Quotes: Use ChatGPT to create a draft based on past quote data → The responsible person reviews and revises.
• Drafting Response Emails for Complaints: Incorporate standard patterns into prompts to generate draft replies → A supervisor reviews and sends.
• Summarizing Daily Reports: Use AI to summarize daily reports from the field, allowing the management to grasp the overall situation in just five minutes each morning.

With just these three tasks, the time to create a quote was reduced from 40 minutes to 10 minutes per case. The initial response time for complaints was cut from an average of 2 hours to 30 minutes. Daily report reviews, which previously took an hour, now take just five minutes.

Telling all employees ‘you can use AI freely’ is less effective than saying ‘let’s use AI this way for these three tasks.’ This approach prevents personalization, allowing for reproducibility even when personnel changes occur.

Step 3: Hold a Monthly 30-Minute ‘AI Reflection Meeting’

The final step is to establish a feedback mechanism.

Once a month, take just 30 minutes to ask the following:

• What went well with using AI?
• What were the failures when using AI?
• Are there any new AI usage methods you would like to try?

That’s all it takes. You can have AI summarize the minutes.

After three months of these reflections, ‘knowledge of how to use AI’ will naturally accumulate within the company. Without relying on external consultants, your organization will develop optimized AI usage know-how.

The cost is zero. All that is needed is 30 minutes a month.

The True Strength of Small Businesses Lies in ‘Seeing Everyone’s Face’

Atlassian needs to implement a tracking system because there are too many employees to see who is doing what.

Small businesses are different. The CEO knows everyone’s face. They can share insights like ‘using AI this way worked well’ in morning meetings. They can tell the person next to them, ‘that would be faster if done with AI.’

This closeness is something that large corporations cannot replicate.

Instead of paying tens of thousands of yen a month for tracking tools, a simple one-page rule and a 30-minute monthly meeting can yield the same or even greater effects. This is the structural advantage of small businesses.

Create ‘Forms’ Instead of ‘Management’

What we should learn from Atlassian’s news is not ‘we must also start tracking.’

The most dangerous state is when AI usage is left unchecked.

And the solutions differ between large and small businesses. Large corporations manage through systems, while small businesses create and share ‘forms.’

Here are three things you can start doing today:

1. Summarize what information can be shared with AI and what cannot on one sheet of paper.
2. Decide on three effective ways to use AI in your company and have everyone use them.
3. Hold a 30-minute reflection meeting once a month.

There is no need for a tracking system. No consultants are needed. All that is required is to ‘decide first, try it out, and reflect.’

AI’s evolution won’t wait for you. Today is the best time to start systematization.