SYSTEM NOTICE

Auto translation by AI. Be sure, accuracy, nuances and authorial intent may not be fully reflected.

Show original

For companies that installed ChatGPT but haven't changed: How to make AI your 'Business OS'

吹上由樹@株式会社Lean Stack

2026年1月19日 12:20

Hello, I am
Yoshiki Fukiage, CEO of Lean Stack Inc..

[Author Profile]

Yoshiki Fukiage
Representative Director, Lean Stack Inc.

April 2021: Joined the Ministry of Economy, Trade and Industry (METI)
July 2023: Resigned from METI
September 2023: Founded Lean Stack Inc.

Company Website: Lean Stack Official Site

YouTube Channel: DX Dekiru-kun
Click here for the channel

Introduction

"We installed ChatGPT, but nothing much has changed on the front lines."

Is your company experiencing this kind of 'implementation fatigue' as well?

In reality, the battle in 2026 won't just be about the intelligence of the model.

The way you win will be determined by who controls the 'gateway' to AI.

This is exactly the point that AI researcher Shota Imai mentioned on a tech program.

If the prospect of Apple advancing its next-generation foundation model based on Gemini becomes a reality, the iPhone's gateway will suddenly become 'Googlized'.

Even with the same product quality, the number of people passing through the gateway will differ.

So, what should you do?

I will outline the mindset for small and medium-sized enterprises to ensure they aren't 'left behind' in 2026, and how to take action starting today.

1. Why the 'AI Hegemony Battle' matters to small and medium-sized enterprises

You might think, 'Isn't that a topic for large corporations?'

But here is the key point.

The gateway becomes the standard

AI only provides value once it is used.

Therefore, in the end,
'familiarity' wins over performance.

If the iPhone becomes the standard gateway for AI, it is like having a new receptionist placed at the entrance you pass through every morning.

Employees' behavior will naturally gravitate toward it.

Ecosystems swallow up work

What I want to emphasize here is that

Google's strength lies not in the model itself, but in its ecosystem.

Gmail, Calendar, Drive, Photos, and YouTube.

When AI is deeply integrated into these, it becomes not just a 'search box,' but the flow of work itself.

When data is connected, the scope of AI's effectiveness expands dramatically.

'Can it be operated safely?' determines the winner

In 2026, it will not just be about the 'performance battle' of AI, but also about corporate guardrails (operational rules).

What is okay to input?
Which data is okay to link?

If you proceed while leaving these points ambiguous, you will be in a state where you don't know if your brakes will work.

2. Concrete usage: In 2026, it will become a 'Business OS' rather than 'AI alone'

From here on, I will break this down into 'how to use it' that works in the field.

The keyword is to place AI at the entrance of your operations, rather than using it as a tool.

1. Reduce email replies and summaries to '1 minute'

Open a Gmail thread and follow the past correspondence.

This task is as tedious as searching for documents in the back of a drawer every single time.

Simply having AI output the 'main points,' 'the other party's arguments,' and 'the conclusion to return' first will improve the quality and speed of your replies.

The goal is to reduce reply creation time from 5 minutes to 1-2 minutes.

Well, this is a case of 'every little bit helps'.

2. Organize 'points of discussion' and 'decision-making' before meetings

Companies with long meetings don't have bad meetings.

They just have scattered points of discussion.

Feed the AI the 'previous meeting minutes,' 'related emails,' and 'materials' to create a 'discussion point memo' before the meeting.

You can also aim to reduce meeting minute creation from 30 minutes to 5 minutes.

This is also a case of 'every little bit helps'.

3. Make internal manuals 'accessible by asking'

What troubles the front lines is not a lack of information, but the inability to find it.

Make regulations, procedure manuals, and quote templates in your drive accessible to the AI.

When a new employee asks, 'Which procedure applies here?', the AI will be able to return the relevant section.

In construction and manufacturing, there is a real labor shortage, and with foreign workers flowing into the service industry, it is truly serious.

This is a very good measure to help new employees become immediately productive and to reduce training costs as much as possible.

4. Use the healthcare field for 'organizing before asking a doctor'

Medicine is not a field to leave entirely to AI.

However, using it to organize 'what to ask the doctor' is practical.

Symptoms, history, medication, and what you want to ask.

Having AI organize this will make your medical visit significantly more 'condensed'.

5. Open models (e.g., DeepSeek series) are becoming an option for 'internal, closed use'

One of the reasons why China-based DeepSeek is attracting attention is that it publishes its research results and models, accelerating evolution.

The lesson for small and medium-sized enterprises here is simple.

If there is information you do not want to put into the cloud, the option of 'running it closed within the company' will become a reality in the future.

If you are incorporating it into your operations, be sure to check 'where it is running (cloud or in-house)'.

Thank you for reading this far.

This time, I have written for presidents and managers of small and medium-sized enterprises who are eager to make AI their business OS from now on.

I would be very happy if even one point is helpful to you.

Well then, see you in the next article!!