A professional engineer reads and reviews the business tools, scripts, and macros your employees built with generative AI such as ChatGPT, Claude, or Copilot. We find what is risky before you rely on it in production and work out the fixes with you. The first month is free.
Does This Sound Familiar?
Generative AI now lets people who never learned to program build tools that work. But working is not the same as being safe to keep using. Code written by AI looks plausible, and the person who built it cannot tell when something is wrong.
Common Problems in AI-Written Code
Failures go unnoticed
The code swallows errors, so when a step fails nothing is shown and the work moves on with data missing.
It calls commands that do not exist
The AI wrote functions or settings that do not actually exist, so the tool fails only under certain conditions.
Passwords or API keys are written in the code
The moment the file is handed to someone, the credentials go with it.
No check before overwriting or deleting
If the target is specified wrongly, the original data is deleted or overwritten.
It stops on conditions nobody tried
Blank cells, full-width characters, date formats, large volumes — input that was not tried when the tool was built makes it stop or produce wrong results.
Japanese text gets garbled
Differences in Windows character encodings garble Japanese file names and file contents.
These problems are often in use without the builder noticing, which is why it pays to have a third party read the code before it goes into production.
About the Founder Who Does the Work
Go Komura, our founder, has worked as a software engineer on the development and troubleshooting of device control, communications, and long-running systems. We also use AI coding assistants in our own day-to-day development, and we see first-hand where AI goes plausibly wrong.
Reviews are not subcontracted. The founder does them directly.
What We Offer
Review of AI-Written Code
We read the tools, scripts, and macros your employees built with generative AI and explain, in plain language, what is risky and why before you rely on them in production.
Fix Consultation and AI-Ready Fix Instructions
We work out how to fix each finding with you and also hand it back as instructions you can pass straight to your AI. We re-check the fixed code as well.
Consultation Before Building or Rolling Out
Talk through what to have the AI build and how, and how to distribute and keep using the finished tool inside your company.
Languages We Cover
- PowerShell
- VBA (Excel and other macros)
- C#
- Python
- Please ask about other languages too
Pricing
First month
FreeSend us your actual internal code and try the review and consultation free of charge. There is no fine-grained cap on the number of reviews during the free month.
From the second month
Monthly, excl. tax100kJPYThe standard fee is 100k JPY per month (excl. tax).
The service does not switch to paid automatically when the free month ends. Decide whether to continue after seeing what the first month delivered.
How It Works
Get in Touch
Try the first month free
"We want to know whether it is safe to keep using a tool an employee built with AI" is reason enough to get in touch. Tell us what tools you have and how you use them.
Code is received under an NDA