Pro edition · coming
First of all: AI does not touch the calculations
Everything Wiring Lab produces — metres, conductor sizes, impedances, losses in dB, fill ratios, bills of materials, drawings — comes out of physical formulas and rules from standards written into the code, applied to your project's data. There is no generative model in between, and there will not be.
On a system that has to stand up, a number has to be repeatable and checkable by hand. That part does not change, and the AI link does not touch it.
What does change: the project can talk to your assistant
The link does not bring artificial intelligence into Wiring Lab. It does the opposite: it takes the project's data out, to the assistant you already use, so that you can ask it for what the software does not do.
You use your AI, not ours
We are not selling you an engine of our own at a high price, one that would be out of date in six months and that you would pay for twice. You connect the subscription you already have, the one you work with every day and whose strengths and weaknesses you know. Change assistant and you change the link: the software stays where it is.
It talks to the real data, not to a photograph
The link runs over an open standard for communication between software and assistants, the same one the most widely used assistants speak. The AI does not read a screenshot, does not summarise a PDF and does not guess from what you pasted in: it queries the project and receives the data that is inside it, the data you worked on.
What it is actually for
The client wants a table the software does not produce. Another designer asks for the summary in their own format. The purchasing department wants the bill of materials grouped differently. You need a description of the system in words for a report, or a graphic no menu offers.
These requests are endless and different from client to client: they cannot all live inside a piece of software, and the day they did they would still be the wrong ones for the next client. With the link you settle them with your own AI and whatever approach you think fits — but starting from the project's real data, not from something retyped by hand.
Where our guarantee ends
The data the assistant receives is the project's: for that we answer. What the assistant then writes on top of it is its own work. A summary, a text or a drawing produced by a generative model has to be re-read before you send it to a client, exactly as you would re-read a colleague's work. The difference is that it starts from correct numbers instead of invented ones.
Which assistants it works with
With every one that speaks the standard. Today those are:
- Claude — Anthropic
- ChatGPT — OpenAI
- Gemini — Google
- Microsoft Copilot — Microsoft
- Le Chat — Mistral AI
- Visual Studio Code, Cursor and the other development environments with an assistant built in
The list grows month by month and it is not up to us. The link is written once and serves them all: when a new assistant appears, if it speaks the standard it works without us having to add anything and without you having to update the software.
One thing to know about your data
Wiring Lab works offline and the project stays on your disk. But the moment you connect an assistant that runs in the cloud — and nearly all of them do — the data you ask it to work on reaches that provider's servers, exactly as when you paste text into its chat.
That is why the link is optional and switched off until you switch it on: if you never enable it, nothing leaves your computer. If you do enable it, you decide which projects the assistant may open. And if you use an assistant that runs locally, the data does not leave at all.