Why Automatic Nesting Still Needs a Human Touch (And How to Make That Fast)

A few weeks ago, someone testing Lapas sent me a message that stuck with me. He was moving away from a much more expensive CAM-nesting package and wanted to know, basically: is this actually going to save me manual re-work, or just repackage the same problem in a cheaper tool? Fair question. I want to answer it properly, instead of giving the usual software-company answer.

Here’s the honest part: automatic nesting doesn’t remove manual adjustment. Nothing I’ve seen does, ours included. What it changes is how much of that adjustment you still have to do, and how painful it is when you do.

Where the algorithm runs out of information

Nesting software optimizes geometry. It’s very good at that: packing 20-something irregular shapes into the tightest arrangement it can find, much faster than anyone could do it by hand.

But there’s a whole category of information no DXF file carries, and no algorithm can invent. Where your table clamps actually sit, and how much clearance a part needs so it doesn’t get pinned during the cut. What your cut sequence needs to look like so heat doesn’t warp a thin part before the last pierce. Whether the sheet in front of you has a scratch in one corner, or is actually a remnant with a chunk already missing. Whether a part’s grain direction matters for the finished piece.

None of that lives in the file. It lives in your head, or on a sticky note next to the machine.

So when people tell me the auto-nest “looks tight on screen, but we still move things around before we cut”, that’s not necessarily a complaint about the software being bad at its job. It’s what happens when a packing algorithm meets a real shop floor.

I hear some version of this from almost every shop that’s used Radan or SigmaNEST for a few years and is starting to look around.

What I won’t tell you

I could write this post as “upload your file, get a perfect layout, done.” It would read nicely, but it wouldn’t be true. Shop owners who’ve been burned by that kind of copy before can spot it, and honestly, they should.

What I can tell you is what we actually built the last 5 to 10% of the workflow around: once the automatic nest gives you something close, moving or rotating a part to clear a clamp should take seconds, not become a whole new task. You don’t re-export, fix it in another tool, and re-import. You adjust it where it sits, and the export reflects exactly what you changed.

To make that concrete, here’s a job we ran recently. No cherry-picking, just the last real file I had open: 26 parts, a 48 x 96 in sheet.

The initial estimate was 3 sheets. The automatic nest packed all 26 parts across 2, at 86.0% utilization on the first sheet and 81.3% on the second, in about 13 seconds.

From there, if a part needs to shift because it lands too close to a fixture point, that’s a few seconds of dragging, not a full re-run.

That gap, between “needs manual adjustment” and “manual adjustment takes almost no time”, is the thing worth evaluating. Not whether a tool claims to be fully automatic. None of them are, whatever the homepage says.

The number that usually gets left out

Software that handles this well at the enterprise level, Radan and SigmaNEST and similar, typically runs $5,000 to $20,000 a year, plus setup time that assumes you’ve got someone to help with the installation.

For a shop with two to twelve people, that’s a real conversation with whoever signs off on the budget, not a rounding error.

Lapas runs in a browser, nothing to install, and costs $39/month billed annually, which is $468 a year.

Same core idea: let the algorithm do the packing, and make the last bit of manual work fast instead of expensive. Just without the five-figure price tag.

If you’re actually comparing tools right now

Don’t trust a utilization percentage from any vendor’s homepage, mine included.

Bring your own file and ask three things:

1. Does it run on a job I actually have, not a demo file the vendor picked?
2. When the auto-nest isn’t quite right, how many clicks and how much time does fixing it take?
3. Does the export match what I changed by hand, or do I need to double-check it before it goes to the machine?

Those three questions will tell you more in ten minutes than any spec sheet will.

If you want the wider field first, I’ve written up the nesting tools worth knowing about in 2026, including the free ones.

---

Upload one of your own jobs and see where the automatic nest gets you, then see how fast the rest is to fix. Run a nest in the browser, free, no credit card required to start.

Rokas builds the nesting engine at Lapas and spends more time than he’d like reading shop-floor feedback and looking at edge-case DXF files. If you’ve got a file you think will break the algorithm, he genuinely wants to see it: [email protected].