AI Notes Record Meetings, Memory Retains Context, and Agents Operate Autonomously—In an Era Where a ‘Zero Personalization’ System Can Be Set Up for Just ¥12,000 a Month, Are You Still Writing Meeting Minutes by Hand?

If the Person Who Writes Meeting Minutes Leaves, Your Company’s Meetings Become “Nonexistent”

Have you ever witnessed a situation like this in a small or medium-sized enterprise?

The person who best remembers the details of the meeting has been transferred. They have resigned. They fell ill. Suddenly, questions like “What happened with that issue?” start flying, and the same discussions have to be restarted from scratch.

This is the essence of personalization. Information exists only in the minds of individuals. It is not shared. It is not handed over. As a result, the company’s decision-making becomes dependent on “specific individuals.”

The problem has been that the cost of breaking this structure has been too high until now. Installing a dedicated meeting minutes system could cost several million yen annually. Implementing a knowledge management tool could exceed 1 million yen just for initial costs. This was not realistic for small and medium-sized enterprises.

However, starting in late 2024, the situation fundamentally changed. AI note-takers, persistent memory, and autonomous agents—by combining these three, a ‘zero personalization’ system can be set up for about ¥12,000 a month. Moreover, the time required for setup is less than half a day.

For those who are skeptical, let’s break down the calculations.

—

Step 1: Eliminate “Recording Costs” with AI Note-Takers

First, let’s talk about recording meetings.

Currently, there are plenty of options for AI note-takers. Otter.ai, Fireflies.ai, tl;dv, Notta—each can connect to Zoom, Google Meet, or Teams, transcribing spoken words in real-time and even automatically generating summaries.

Specific Costs:

• Otter.ai Pro: $16.99/month (about ¥2,500) per user
• Fireflies.ai Pro: $18/month (about ¥2,700) per user
• tl;dv Pro: $18/month (about ¥2,700) per user
• Notta Business: about ¥1,300/month per user (specialized in Japanese)

The key point is that licenses for everyone are not necessary. Since the AI note-taker simply “attends” the meeting to record, one account can cover all meetings within the company. Whether it’s a company of five or fifty, the cost remains around ¥2,500 to ¥2,700 per month.

Let’s consider the costs that have been incurred in creating meeting minutes until now. Suppose there are meetings three times a week, and each time it takes 30 minutes to write the minutes. That’s 12 meetings a month × 30 minutes = 6 hours. At an hourly wage of ¥2,000, that’s ¥12,000 in labor costs per month. Moreover, if that person misses something, the information is lost.

With an AI note-taker, for ¥2,500 a month, you get full transcripts of spoken words + summaries + action item extraction automatically. Labor costs of ¥12,000 are reduced to ¥2,500. And there are no omissions.

What’s important here is not the “accuracy of the recording.” The accuracy is around 95%, which is not perfect. However, meetings that previously had no records at all will now have full transcripts at 95% accuracy. The impact of increasing from zero to 95 is incomparable to increasing from 95 to 100.

—

Step 2: Accumulate “Context” with Persistent Memory

Even if records are kept, that alone does not eliminate personalization. Why?

“What was decided about Company A in the meeting three months ago?”—searching through piles of minutes to find this becomes a human task again. Even if you search, you won’t know which file or which part to look at. The context is fragmented.

This is where “persistent memory” comes into play.

Using tools like ChatGPT’s memory feature, Claude’s project feature, or NotionAI, the AI can “remember” past meeting contents and connect the context.

Specifically, it would work like this:

> “What progress was made on Company A’s project last time?”
> → The AI refers to the past three meeting records and responds, “In the meeting on September 12, it was decided to submit a quote. On September 20, a revision request was received from the other party. On October 3, it was resubmitted, and we are waiting for a response.”

This strikes at the heart of personalization. The shift from “I have to ask that person to know” to “I can ask AI to know” is transformative.

Costs:

• ChatGPT Plus: $20/month (about ¥3,000)
• Claude Pro: $20/month (about ¥3,000)
• Notion AI: $10/month (about ¥1,500) per user

Again, one account is sufficient. Feed the meeting records to the AI and use it as the company’s “memory device.” That’s ¥3,000 a month.

Combined with Step 1, the total monthly cost so far is about ¥5,500.

—

Step 3: Create “Autonomous Completion” with Autonomous Agents

Records are kept, and context is connected. The final piece is the “action” part.

Someone manually transcribing tasks decided in meetings into a project management tool, reminding the responsible parties, and managing deadlines—this task is the most prone to personalization, the most cumbersome, and the most easily forgotten.

Autonomous agents automate this process.

Example Configuration:

1. The AI note-taker extracts action items from the meeting.
2. Zapier/Make (formerly Integromat) automatically registers tasks in a task management tool (Asana, Trello, Notion, etc.).
3. Reminders are sent to the responsible parties via Slack/email before deadlines.
4. If progress is not updated, escalation notifications are sent.

This entire flow operates automatically without any human intervention.

Costs:

• Zapier Starter: $19.99/month (about ¥3,000)—up to 750 tasks per month
• Make Free to Core: Free to $10.59/month (about ¥1,600)
• Task management tools (Trello, Notion, etc.): Free plans are sufficient

Even adding Zapier’s monthly cost of ¥3,000, the total monthly cost for the three steps is about ¥8,500.

If you want a more feature-rich setup, combining the GPT-4o API usage fee (a few hundred to a few thousand yen per month) or n8n (an open-source automation tool that is free if self-hosted) will still allow you to create a setup for well under ¥12,000 a month.

—

Cost Comparison: Manual vs. AI Automation

Now, let’s compare the traditional “manual” costs.

*Estimated based on hourly wage of ¥2,000 × monthly work hours

Monthly costs drop from ¥35,000 to ¥8,500. That’s a cost reduction of about 76%. Moreover, the AI does not “forget,” “rest,” or “quit.”

However, the real impact is not just cost reduction.

—

The Essence is Not “Cost Reduction” but “Structural Transformation”

What this system truly changes is the shift from a structure where information is tied to individuals to a structure where information is tied to systems.

The biggest risk for small and medium-sized enterprises is that operations come to a halt when key persons leave. If a top salesperson quits, customer information disappears. If the accounting person is absent, billing stops. If the meeting facilitator is gone, the history of decision-making vanishes.

The combination of AI notes, memory, and agents can eliminate this structural risk for under ¥10,000 a month.

Moreover, this is not about implementing large-scale systems like those in big corporations. It’s simply about contracting three SaaS tools and connecting them. No IT expertise is required. Setup takes half a day. Because they are small and medium-sized enterprises, decision-making can be swift, and implementation can start tomorrow.

—

Realistic Three Steps for Implementation

Trying to do everything at once can lead to frustration. It’s more practical to try one step at a time in the following order.

Week 1: Introduce an AI Note-Taker to One Regular Meeting

• A free trial is sufficient. Both Otter.ai and Fireflies.ai have free tiers.
• First, share the experience of “automatically generating meeting minutes” with the team.
• At this stage, most will feel that “there’s no going back to handwritten minutes.”

Weeks 2-3: Feed Records to ChatGPT or NotionAI to Enable “Inquiry”

• Simply copy and paste the text of the minutes. Complex integrations can wait.
• Experience asking, “What was decided about Company A in last month’s meeting?” and getting a response.
• This will create a tangible feeling of “reduced personalization.”

Week 4: Automate One Task with Zapier or Make

• For example, just automating “posting action items from AI notes to Slack” is sufficient.
• A small success in automation will generate motivation for further automation.

By the end of one month, you will have started a cycle of “recording meetings → accumulating context → automating tasks” for under ¥10,000 a month.

—

So, What Should You Do?

The answer is simple.

Introduce one AI note-taker to tomorrow’s meeting.

It can be free. If you’re concerned about accuracy, you can try it once and then decide. While it may sound like a grand idea to “eliminate personalization with AI,” the entry point is simply “inviting one bot to the meeting.”

In the past, when trying to implement a meeting minutes system, the estimate was ¥3 million. Now it’s ¥2,500 a month. In this era of price disruption, saying “I’m still considering it” is effectively synonymous with “I’m not going to do it.”

Personalization becomes a liability for the organization the longer it is left unchecked. And that liability will suddenly become apparent the moment a key person leaves.

For under ¥10,000 a month, you can defuse that bomb. If you have a reason not to do it, I’d like to hear it.