Every tool we actually run a cross-border business on. What each one costs, what it replaced, and where it lets us down.
Disclosure. Some links on this page may be affiliate links, meaning we could earn a commission if you sign up through them at no additional cost to you. Nothing here is a paid placement. We only list tools we pay for and use in our own work, and we say plainly where each one fails. If a tool stops being useful, it comes off this page.
We run a company that works in two directions — overseas companies entering Japan, Japanese companies going abroad — from a small team in Tokyo. That means every tool has to work across time zones, in two languages, and be operable by people who are not engineers.
Most "tool stack" pages are written by people who signed up for a free trial. This one is a record of what we pay for every month, and it is updated when something changes rather than when it is convenient.
Nobody at this company writes code. Everything below is chosen on one criterion: will it still be running in three weeks without someone technical to fix it.
An AI that works inside your files and terminal rather than in a chat window. We use it to build internal tools, draft and send outbound email, produce documents, and render video. The internal sales tool we run was built with it in an evening.
Hosting for our own site and everything published from it, including this page. Push to a repository and it deploys. The free tier is genuinely sufficient for a company site.
Renders HTML and CSS animation to MP4. We use it to turn data we already have into short vertical video for social, without opening an editor.
Workflow automation you can host yourself. We use it to move data between systems and to run scheduled jobs that would otherwise be someone's Monday morning.
Cross-border work is mostly a language problem wearing a business costume. These are the tools that carry that weight.
Mail, calendar, documents, and video calls. Unremarkable, and that is the point — every counterparty we deal with, in every country, can already open what we send.
Where project material lives so that a member in Brazil and a member in Tokyo are looking at the same page.
A stack page that only lists what someone likes is an advertisement. These are the categories we have tried and left.
We tried several. Our pipeline currently lives in a plain text file that our own tooling reads and writes. Two hundred prospects is not a CRM problem — it is a discipline problem, and paying for software does not solve discipline.
We will need one eventually. Not yet.
The category that promises to send hundreds of personalised emails per day. We do not use them, and it is a deliberate choice rather than a budget one.
Our reply rate went from 3% to roughly 15% when we stopped selling and started asking a real question in each message — and a real question requires knowing something specific about the company you are writing to. That cannot be templated, which means it cannot be scaled the way those tools promise.
We send ten to fifteen a day, by hand, and each one names something true about that company's situation in Japan.
There is a popular genre of tutorial explaining how to redirect paid AI tools through free third-party providers. We hold client contracts, a signed exclusivity agreement, and a pipeline of named companies. None of that goes through infrastructure we cannot account for, at any saving.
The question is never what a tool can do. It is whether it will still be running in three weeks without someone technical to fix it.