AI Is Coming for Your CAD Skills. Let It.
AI will deskill hardware engineers the way calculators deskilled long division.
Good!
The job was never clicking in CAD. The job was deciding what to build, and making sure it works when it is built.
“Calculators deskilled long division”
This week, a post argued that AI is not replacing site reliability engineers. It is deskilling them. When the AI handles most of the incidents, the engineers get less practice on the hard ones. The Hacker News thread ran to hundreds of comments.
One reply said it best:
“Calculators deskilled long division! ... who gives a damn about long division? The real task is your taxes.”
I think this is exactly right, and it is even more true in hardware. Most hardware engineers I talk to spend most of their week on long division. Very little of it is the real task.
The long division of hardware
Look at a normal week on a device team. A lot of the time goes into work like this:
- Route 400 traces by hand, then rip up half of them after a review comment.
- Find a replacement for a part that went out of stock, and update the BOM.
- Chase design rule violations, one marker at a time.
- Move a connector 5 mm, then fix the opening in the enclosure to match.
- Model the same screw bosses, ribs, and fillets again for the next housing.
- Check by hand that the board clears the enclosure walls.
- Export Gerbers, drawings, STEP files, and pick-and-place files, and send them out for quotes.
This is skilled work. It took years to learn. But it is not why anyone became a hardware engineer. Nobody will miss it, the same way nobody misses long division.
In Flux, you tell the AI engineer what to do in one sentence. “Route this board.” “U3 is out of stock, find a replacement with the same footprint.” “Change the mounting hole pitch to 50 mm on the board and in the enclosure.” Flux does the work on the electronics and the mechanical parts together, in one project.
The real task is deciding what to build
When the long division goes away, what is left is the work that was always the job. And it gets a lot more of your time:
- What should this device do, and for whom?
- Which architecture? One board or three? Aluminum or molded plastic?
- What can you give up on cost, size, battery life, or heat?
- What can your manufacturer build, and at what volume?
- Does it work? How do you know?
These questions decide if a product succeeds. They used to get the last few hours of the week, after the CAD work was done. Now they get the whole week.
The skill does not disappear. It moves up. From operating the tool to deciding what to build and judging the result. That is a better job!
“But you need long division to catch the calculator's mistakes”
That was the strongest reply in the thread, and I agree with it. You need to know when an answer is wrong. An AI that only generates designs, with no one able to judge them, is dangerous.
But look at what that skill actually is. It is not routing traces by hand. It is knowing what a good result looks like, and asking the right questions about it. That is engineering judgment, and engineers with AI get to use it on every design instead of only at the end.
And a hardware AI engineer must check its own work before you do. In Flux, the agent does exactly that:
- It reads the datasheet of every part and checks the design against its pinouts, ratings, and limits.
- Electrical and manufacturing rule checks run on every schematic and every board.
- It checks every design against your layout rules and your manufacturer's capabilities.
- It runs circuit, electromagnetic, and thermal simulation, reads the results, and changes the design.
- It checks fit and clearance on the assembled board, enclosure, and brackets.
You review a device that has already been checked, and you ask why. “Why this regulator?” “What is the thermal margin at 60 °C?” That is how you judge an AI-designed board: by whether it works when it is manufactured, not by how it was drawn. It is also why choosing what to simulate is a skill that matters more now, not less.
Make the CAD disappear
I have been building AI for hardware for seven years. The goal was never AI-assisted CAD. The goal is to make the CAD disappear.
Software went through this first. Developers stopped writing every line by hand. They describe what they want, review the result, and ship more than ever. Nobody calls the best of them deskilled. Hardware is next!
The engineers who win the next ten years will not be the fastest at clicking in CAD. They will be the ones who know what to build and who can tell if it works. Flux gives them an AI engineer that does the rest.
FAQs
Yes, in the same way that calculators deskilled long division. AI takes over the repetitive CAD work, such as routing traces, fixing rule violations, redrawing enclosure features, and exporting manufacturing files. The skills that matter more are deciding what to build, choosing the architecture and the tradeoffs, and judging if the device works.
No. AI changes what hardware engineers spend their time on. In Flux, the AI engineer does the design work in the project, and the engineer decides what to build, reviews the result, and makes the decisions.
Product and system thinking, architecture, tradeoffs between cost, size, power, and heat, knowledge of manufacturing, and the judgment to tell a good result from a bad one. Writing a clear request in plain words is a skill too.
Check it the same way you check any design: electrical validation, manufacturability, fit, and cost. In Flux, the agent runs electrical and manufacturing rule checks, checks parts against their datasheets, runs circuit, electromagnetic, and thermal simulation, and checks fit and clearance in the enclosure before you review the design.
Yes. In Flux, you can ask the agent to place and route a board, change parts, and model the enclosure, brackets, and other mechanical parts in the same project. The electronics and the mechanical parts are checked together.
Try it
Skip the long division. Open a project and tell Flux what you want to build. For example: “Design a battery-powered soil moisture sensor with LoRa, in an IP67 enclosure that mounts on a stake.” Start Free Trial.
Related Content
Your AI Agent Can Outgrow Its Editor: Hardware Design Workflows Beyond EDA and CAD Features
Automated PCB Manufacturing Needs Automated Design
Cloud AI Agents: Your Project, Not Your Whole Laptop
How to Choose Decoupling Capacitors with PDN Simulation, Not Rules of Thumb
Gerber File Check: The Flux Agent Checks Your Files Before Fabrication
Stack Overflow for Hardware Engineering: Ask AI Inside Your Design
Let’s start building.
Let’s build your idea together. All you need to do is describe what you want to make, or what problem you want to solve. Flux will work with you to make it a reality.
