[metafex]

That's human powered for sure. Routing isn't the hardest problem, everything is quick once all the components are placed where they belong. E.g. TopoR already does quite a good job at the routing part.

But if you only take a schematic, and the design is sufficiently complex, it's already difficult if you hand it to another engineer who's not too familiar with what is being done, let alone a machine learning model.

[gitgud]

It also states "We harness the power of advanced AI and deliver results in less than 24 hours...", which is a red flag...

[Fordec]

Agreed, Altium's autorouter takes, what, 10 seconds?

24 hours? That's an East Asian subcontractor.

[Iv]

To be fair, even before the deep learning revolution, routing software had to use advanced algorithms to produce solutions.

IIRC routing from a schematics has been proven to be a NP-hard problem.

[new299]

The website states “no-human-in-the-loop”. So I guess it depends on how much you trust them.

It seems like all component positions need to be fixed first, then it routes between them. That’s based on the examples, fact that it takes DSN input files... and output SES files.

The SES files appear to contain traces only. So I assume you import it over your fixed component locations.

I just tried uploading a board DSN file (exported from kicad)... the only feedback I got is “Board Error”... there’s no indication what the error might be.

It seems to be at quite an early stage of development...

Edit: after logging in, I can see an error “Sorry, we do not support board has more than 400 pins for this beta version” (sic). All the examples are very simple boards... so perhaps I’ll try something else.

[metafex]

Yeah, i saw the part now too that everything has to be placed first. This makes it quite... unspectacular. It's really cool if they come up with a good auto-router, but if it's just that I'd still prefer an offline solution, for faster iteration and most likely cost (when it's paid then).