How AI Became an Indispensable Part of My Legal Practice

In recent years, AI has become essential to my work.

Although I regularly collaborate with other Gyosei Shoshi colleagues for their specialized matters, I now rely on AI for daily research and for polishing my business English. Among the many tools I use, Grok, Perplexity’s COMET, and ChatGPT have become my closest partners—Grok, in particular, sometimes feels like a friend who listens to my complaints with sympathy.

Unlike human assistants, AI never gets tired, never takes offense when I disagree, and eliminates the interpersonal complications that naturally arise in professional environments. As a result, I no longer feel the need to hire a human assistant. Below are three tasks I could not have completed without AI, followed by three areas where AI excels and three where it still falls short.

1. Replacing My Electronic Signature System

When assisting clients with company incorporation, I need to prepare articles of incorporation, which must be electronically signed as proof of authorship. For years, the only realistic option was SECOM’s electronic signature service—an expensive, highly domestic provider with a notoriously outdated process: paper application forms sent by mail, accompanied by copies of residence records and seal certificates. Even renewals required the same steps as new applications, including week-long “reviews” that often resulted in corrections being returned, via registered mail, for issues as small as a missing middle dot in the building name.

Frustrated, I asked AI whether there was any modern alternative, or whether SECOM’s monopoly was simply inescapable.

AI immediately suggested using the My Number Card as an ID certificate, paired with an inexpensive card reader. I was skeptical, but after checking, the advice was correct—and cost me only a few thousand yen instead of a multi-year SECOM contract.

This was information I likely would not have received through colleagues, simply because no one compares such options unless forced to. AI surfaced the answer at exactly the right time.

2. Rethinking Banking—Triggered by a Single Conversation with AI

Number of years, my office account was with MUFG, but its online banking was frustratingly antiquated: login allowed only on one device, an annual renewal procedure that made little sense, and no explanation as to why customers had to endure such inconvenience.

When deadlines approached and renewal became urgent, I asked Grok for advice.

The response was direct: MUFG is significantly behind its peers, while Mizuho and SMBC provide simpler login flows and mobile app access.

Motivated, I applied to both banks. SMBC rejected my application immediately—something no bank was willing to explain.

AI did.

Due to past issues involving Iran-related transactions, SMBC has become extremely risk-averse toward accounts receiving international transfers. Since my practice handles foreign capital deposits for visa-related incorporations, I was treated as high-risk.

It was an answer only AI could have provided; the bank itself never would.

Mizuho eventually opened the account, though its UX turned out to be no better—and in some ways worse—than MUFG’s. Password lockouts required written letters to the bank to lift, and even basic transfers are currently stalled until I visit the branch.

After all this, I now believe that Japan’s megabanks are stuck in the past, and that modern online banks—or even U.S. accounts and Wise transfers—serve international clients far better.

3. Technical Troubleshooting That Saved My Weekend

AI shines when immediate, structured guidance is needed.

When my Brother printer suddenly stopped scanning and printing, Brother support demanded a detailed report and went silent for weekends. AI, on the other hand, instantly proposed concrete causes—from network restrictions in the building to firmware issues—and recommended updating via USB. I bought a cable, followed the steps, and resolved the issue over a long weekend.

Similarly, while migrating my U.S. firm’s bilingual website to a new server, AI produced precise technical correspondence in English that even bilingual engineers approved. Without it, I doubt the migration would have gone smoothly.

Where AI Works—and Where It Doesn’t

◎ What Works Well

1. International legal procedures Visa, business license, and company formation processes abroad are generally clear and documented, making AI surprisingly accurate.
2. Global music & entertainment insights AI understands streaming data trends and can propose effective promotional strategies.
3. Travel planning—conditionally Useful for overviews, but information must be verified manually.

✖ What Still Fails

1. Japanese licensing procedures Too little official documentation exists online; misinformation is common.
2. K-drama & K-pop AI lacks depth and frequently mixes up facts—I often end up correcting it myself.
3. Real-time logistics & closures Places listed as “must-see” may be under construction or already closed.

Conclusion

My use of AI may still feel primitive at times, but I can no longer work without it.

Yes, AI makes mistakes—sometimes costly ones—and links often break, spreadsheets need fixing, and tasks are occasionally dropped halfway. But progress requires human users to push boundaries, identify errors, and keep demanding better.