Conclusion

Let’s get straight to the point: A time will come when being ‘digitally lagging’ becomes an asset.

OpenAI is seriously gearing up to introduce a ‘phone agent’ that will replace apps.

This is a bigger boon for small and medium enterprises (SMEs) that still rely on FAX and phone for orders than for large corporations. Why? Large companies have already invested tens of millions to hundreds of millions of yen into existing SaaS and ERP systems, making the switching costs hefty. In contrast, SMEs can leap directly to the latest interfaces because they have ‘not implemented anything yet.’

This is akin to how mobile payments proliferated in Africa by skipping fixed-line phones. The absence of legacy systems becomes an advantage.

—

OpenAI’s Phone Agent Plan—What’s Happening?

According to industry analyst Ming-Chi Kuo, OpenAI is collaborating with MediaTek, Qualcomm, and Luxshare to develop new devices equipped with AI. The key point is that they aim to eliminate the very act of ‘opening an app.’

Currently, using a smartphone involves searching for an app, opening it, and navigating through it every time you want to do something. For instance, if a restaurant wants to order ingredients, they log into the ordering system, select items, enter quantities, and hit the send button.

In the world of phone agents, it would look like this:

‘Please order 10 kg of pork belly and 5 cases of cabbage from our usual supplier for tomorrow.’

That’s it. The AI identifies the supplier, checks unit prices from past transaction data, and completes the order processing. If confirmation is needed, it might ask, ‘The unit price has increased by 8% compared to last time; shall I proceed?’

In other words, the UI shifts from ‘screens’ to ‘conversations.’ This is not an improvement of apps; it marks the end of the app concept.

—

The Case of Choco—The Transition from ‘FAX to AI’ in Food Distribution

You might think, ‘Isn’t that just a future scenario?’ However, there are already examples in motion.

Berlin-based food distribution startup Choco has utilized OpenAI’s API to automate the order and procurement process with AI. Choco addresses the inefficiencies of the food distribution system that operates on phone and FAX.

According to Choco’s publicly available data, about 80% of order processing in the food distribution industry is done via phone, FAX, or SMS. The average time taken to process a single order, which used to take 6 to 8 minutes, has become almost zero-touch after the introduction of the AI agent.

Here’s how it works:

1. Restaurants place orders as usual via phone or SMS (no change in behavior required).
2. AI analyzes the voice or text, automatically structuring the items, quantities, and delivery dates.
3. Order data flows automatically into the supplier’s system.
4. If there are anomalies (like an order quantity three times the norm), a human is alerted.

What’s noteworthy is that the ordering restaurants have not changed anything. They simply say, ‘Please order 20 kg of chicken thighs for tomorrow’ over the phone. What has changed is the processing on the receiving end. This is the intrinsic value of AI agents, as they do not require users to change their behavior.

Choco currently serves tens of thousands of restaurants and suppliers, primarily in Europe and the U.S., processing a cumulative billions of dollars in food transactions. It also contributes to reducing food waste, with reports indicating that waste rates have improved by over 30% on average due to enhanced order accuracy.

—

Why SMEs Are the ‘Biggest Beneficiaries’—The Reversal of Cost Structure

Now, let’s get to the main point.

Traditionally, when SMEs attempted to digitize their operations, they faced the following costs:

• Implementing an order and procurement system: Initial costs ranging from 1 to 5 million yen, monthly fees from tens of thousands to hundreds of thousands of yen.
• Employee training: Several weeks to months.
• Coordination with existing partners: ‘We can only handle FAX’ often leads to deadlock.

The issue of ‘partners not being digitally equipped’ has been the biggest bottleneck hindering SMEs’ digitization. Implementing a system on one’s own is meaningless if the counterpart still relies on FAX.

AI agents fundamentally change this structure.

Whether the counterpart uses FAX, phone, or SMS, AI acts as an intermediary to translate.

There’s no need to ask partners to ‘please implement a system.’ The AI reads FAX orders using OCR and converts them into structured data. It transcribes phone orders into text through voice recognition and processes them automatically.

In other words, digitization can be completed on one side only. This is something that traditional system implementations could never achieve.

The cost structure also changes dramatically. Considering the cost of OpenAI’s API, the AI usage fee for processing a single order is just a few yen to tens of yen. For an SME with 500 orders a month, this translates to monthly automation costs of just a few thousand yen. The cost of implementing a 5 million yen system drops to just a few thousand yen a month. We are looking at a world where costs are less than 1% of what they used to be.

—

Multi-Agent Systems Will Absorb ‘Single-Function AI’—The Next Structural Change

Another important change to note is the rise of multi-agent systems.

Current AI agents generally operate on a structure where ‘one AI handles one task.’ There are AI agents for processing orders, managing accounting, and customer service, each operating independently.

With multi-agent systems, this will change.

• An order agent receives orders.
• An inventory management agent checks stock levels and suggests purchases if needed.
• An accounting agent automatically issues invoices.
• A delivery agent organizes the optimal delivery routes.

These agents collaborate autonomously to run operations without human intervention. Humans only handle exception processing and final approvals.

Microsoft announced orchestration features for multi-agent systems at the Build conference in May 2025. Google has also released a multi-agent framework based on Gemini. OpenAI, Microsoft, and Google are all investing in this area.

What this trend implies is that ‘single-function SaaS’ may become unnecessary as an intermediary layer. The world where you contract separate order management, inventory management, and accounting systems and log in to each one will come to an end. If AI agents can connect everything behind the scenes, there will be no reason to pay several tens of thousands of yen monthly for individual SaaS solutions.

For SMEs, this represents a structural change that liberates them from ‘SaaS poverty.’

—

So, What Should We Do?

I can hear the voices saying, ‘This is an interesting story, but what should we do?’

The answer is simple.

1. First, take stock of your company’s ‘phone and FAX operations.’

List all the tasks handled via phone and FAX, such as order processing, inquiry responses, reservation acceptance, and quote requests. Determine how many of each occur monthly and how long each takes.

2. Calculate ‘how much money you would save by replacing it with AI.’

For example, if you have 500 orders a month and each takes 8 minutes, that’s 67 hours a month. At an hourly wage of 1,500 yen, that’s 100,000 yen in labor costs. If automated with AI, the API usage fee would be just a few thousand yen a month. This results in a cost reduction of over 90,000 yen per month.

3. Experiment on a small scale.

There’s no need to change all operations at once. Start by assigning just one task, such as ‘phone order processing,’ to the AI agent. By combining the ChatGPT API with Whisper (voice recognition), you can create a prototype for just a few tens of thousands of yen.

4. Design it so that you don’t require anything from your partners.

This is crucial. Asking partners to ‘implement a system’ is a red flag for failure. They will remain on FAX or phone. Design the system so that AI can absorb everything on your side.

—

Summary—A Turning Point Where ‘Lagging’ Becomes ‘Advantageous’

OpenAI’s phone agent concept is not just about devices. It signifies a fundamental paradigm shift where the main interface moves from ‘screen operations’ to ‘conversations.’

And this shift will work to the advantage of SMEs that have not yet digitized.

• There are no investments in existing systems, so switching costs are zero.
• The ‘conversation-based’ workflow of phone and FAX can connect directly with AI agents.
• There’s no need to ask partners for behavioral changes.
• Implementation costs are less than 1% of traditional systems.

Companies that have said, ‘We still use FAX…’ will find themselves with cutting-edge AI workflows before they know it.

Such reversals have already begun. There’s no reason to wait. Choose one of your company’s phone and FAX operations and start experimenting next week.