
What happened
A 16-year hardware engineer says that after months using Claude Code, research, spec drafts, test sequencing and per-project handover notes are almost fully delegated, while judgment and physical checks stay human.
Why it matters
Tasks that used to eat half a day are handled by AI, freeing the engineer to focus on hands-on work and on spotting flaws in AI's plausible-sounding answers, which he says are often wrong.
What to watch
The split hinges on AI still having no hands, so plugging in a USB cable or repositioning a board remains human work; the engineer now documents this as reusable paid templates.
WHO IT HITSHardware engineers and small prototyping shops take note: routine documentation, datasheet lookups and test logging can shift to AI tools, but bench work and safety judgment stay with experienced staff. Contract clients asking for faster turnaround may see shorter explanation time, while engineers must still validate AI's technically plausible but incorrect diagnoses.
Summaries like this, in your inbox every morning.
The author introduces himself as someone who has spent 16 years designing circuits and boards, from arcade game machines to toys, and now also runs a children's robot class. Over recent months he worked with Claude Code on a measurement logger built on an M5Stack, an avatar on a Raspberry Pi that appears and talks when it detects a person, evaluation of a new Bluetooth ranging feature measured between 1 and 5 meters, and prototype firmware for a battery-powered toy. Different genres, but in each one he says the AI became another development team member.
His account rests on concrete friction. The distance-measurement evaluation failed intermittently at 2m; he and the AI read logs and rechecked settings, but the answer came the moment he unplugged the USB cable connecting to the PC, after which the measurements stabilized. In a client inquiry about output remaining ON after a power-off, the AI offered a USB-power explanation he judged inconsistent with the hardware logic, and only after he pushed back did the work move toward a fix that cuts output in every power-off case.
What makes this account notable is the line he draws rather than the tools themselves. Since the work spanned a signal logger, an interactive avatar, ranging evaluation and toy firmware — and since long gaps between client requests are normal in contract work — the memory of a project becomes a deliverable an AI can maintain, which is why he has handover documents written and updated as work proceeds. The likely test of this arrangement is whether such handover records stay good enough that both he and the AI can regain context in minutes after weeks away, and whether younger engineers without his bench instincts can catch the plausible-but-wrong answers he describes catching.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Chunghwa Telecom president Rong-shy Lin said the shift from conversational AI to autonomous AI agents is resha…

Apple and Meta said they would jointly release a new open-weight AI model designed to run on high-end Macs, ai…

A writer and entrepreneur opened his daily to-do list and found 156 items, and concluded that an AI assistant…

A Zenn contributor sorted AI use into three modes — AI as an assistant, AI running alongside people, and AI th…

A non-engineer operator says one person ships four smartphone apps on Google Play, three on the App Store, wit…

A developer released agent-guardrails, a one-script Python standard-library plugin that guards Bash, Write, Ed…
