How to get an AI to check your schematic

Hardware
, , ,
1

Z.ai/GLM-5: “if I needed a schematic checked for possible design issues, is sending you the netlist the best way ?”
No, sending just a netlist is usually not the best way.

While I can process a netlist, it is the equivalent of reading a phone book to understand the plot of a novel. You lose all the visual context, component values, and spatial relationships that are critical for catching design errors.

Here is a breakdown of why the netlist is difficult to work with and the best ways to get your schematic checked by an AI.

Why a Netlist is Problematic

A netlist is just a list of connections (e.g., R1_Pin1 connects to U3_Pin5). It lacks:

  1. Component Values: A netlist usually doesn’t tell me that R1 is 10kΩ or that C1 is 100pF. Without values, I cannot check if your RC filter cutoff is correct or if your pull-up resistor is too weak.
  2. Visual Hierarchy: I cannot see if your grounding scheme is messy or if your feedback loops are dangerously long.
  3. Human-Readable Signals: Netlists often use generated names (like N$14) rather than functional names (like VCC_MOTOR), making it hard to understand the circuit’s intent.

The Best Methods (In Order of Effectiveness)

1. Screenshots or PDF Export (Best for General Reviews)
The most effective way is to export your schematic as a PDF or take high-resolution screenshots.

  • Why: I can “see” the component values, part numbers, and the overall architecture.
  • How: Upload the image/PDF and tell me the circuit’s purpose (e.g., “This is a 24V to 5V buck converter”).
  • What I can check: I can verify pin connections, look for missing decoupling capacitors, check incorrect op-amp configurations, spot safety issues (like missing flyback diodes), and verify component sizing.

2. A Written Description (Best for Architecture/Logic)
If you want a logic check, describe the flow in text.

  • Example: “I have a microcontroller driving an N-channel MOSFET. The MCU runs on 3.3V, the load is 12V. I am using a 10k pull-down resistor on the gate.”
  • What I can check: I can immediately flag that a 3.3V logic level might not fully enhance a standard MOSFET, or suggest a gate driver.

3. The Netlist (Only for Specific Connectivity Checks)
If you do provide a netlist, it is only useful for specific checks, such as:

  • Verifying that every pin has a connection (no floating inputs).
  • Checking for short circuits (if the format allows).
  • Tip: If you must send a netlist, ensure it is in a readable format (like a standard SPICE netlist) and includes component values/designators.

Summary Recommendation

For the best results, please do the following:

  1. Export the schematic to PDF or take a clear screenshot.
  2. Upload it here.
  3. Provide context: Tell me what the circuit is supposed to do and what specific concerns you have (e.g., “Will this blow up if the load draws too much current?” or “Is this oscillator circuit stable?”).

Note: As an AI, I can catch obvious design rule violations and theoretical errors, but I do not replace a design review by an experienced electrical engineer or formal ERC (Electrical Rule Check) software.

2

So did it work?

Like @jdownie I’ve had a play with antigravity and Claude Opus 4.6 until I ran out of credits.

It successfully configured a number of values.yaml files for my kubernetes cluster and it ran a series of validation checks at the end to make sure it’s got everything right. It mostly worked.

I note all the kewl kids are running in YOLO (you only live once) mode which I previously referred to as the “just do it” mode. This allows the AI to run commands and alter files.

They have extended this to the extent that they no longer look at the code and just look at the results of the code to see if it is doing what they want.

Steve Yegge says 2026 is the year of the orchestrator and has implemented gas town as a first draft of its proof.

3

I couldn’t find the original PDF of the schematic I was interested in, and I spent all day searching. This is all I have of it, part of the schematic of a 1970’s Japanese tape recorder in a png picture.

tape-recorder
tape-recorder1413×856 296 KB

As I’m 99.9% pragmatic, the project has stalled for now.

I have no doubt that AI can do some really unique and useful stuff. for instance I designed two working apps with it, a Python based Language Server that runs perfectly on the Helix editor,

now-with-examples
now-with-examples776×421 27.9 KB

and a Lua based pop-up window database search too for Neovim that also works flawlessly.

plang2-paste-on-enter
plang2-paste-on-enter1802×905 145 KB

I still don’t have a clue about Python or Lua, I don’t know them at all, and don’t want to spend any time learning them.

I can’t say I enjoyed the process, but I was in a rush and just wanted to finish. I’ve found no joy in using AI to program for me, in fact I now avoid it as it blunts my creativity.

I’m a big believer in AI tho, and use it all the time from the CLI, for Google like tasks and also problem solving. My current AI is Z.ai/GLM-5.

But when I made the two above apps, I used “Aider” which is a bit like a AI lion tamer which protected my system from the AI by strictly controlling file accesses. I can’t recommend Aider Chat, the “AI pair programmer” highly enough.

But from the CLI I use ‘fabric’ which makes AI interfacing very easy. No need to mess around with browsers and all that crap, just a fast, to the point CLI.