Cutting List Software vs AI: Which Gives a Carpenter a Reliable Cut List?

Use cutting list software for the layout and AI for the words around it. A panel optimiser nests parts, allows for saw kerf and can lock grain direction; a chat assistant can write a plausible list that needs more sheets than it claims. Test either on one job you have already cut, using three checks: area, piece count and grain.

The reason is simple. A layout is a packing problem, and software built for it tracks every piece, every cut and every offcut. A chat assistant predicts likely text, which looks the same whether or not the pieces fit. Where AI does help is before the layout, turning drawings and notes into a parts table, and after it, checking what the optimiser produced.

Follow me on Instagram@sagnikteaches

Four panel optimisers and what their pages say

These four rank for the search, so they are the obvious ones to try first. The details below come from each tool's own pages, read in October 2026. Where a page said nothing about grain or kerf, the table says so rather than guessing, because that is exactly what you will test.

Connect on LinkedInSagnik Bhattacharya

A page that is silent on grain does not prove the tool ignores it, and a page that lists a setting does not prove it works on your board. Both claims get settled by the dummy-part test later. One more detail matters for a one-person shop: the free tiers cap parts, so a large wardrobe job can exceed a cap that a kitchen-base job never touches.

Subscribe on YouTube@sagnikteaches

Why a chat assistant gets cut lists wrong

A language model writes the most plausible next words. For a cut list, plausible means a tidy table with sensible-looking dimensions and a confident total. It does not mean the pieces fit. Four failure patterns show up again and again, and all four can be checked by hand:

1. The area does not add up. The list claims two sheets when the parts need nearly three. A model can miss this because it never adds the areas.
2. A row is longer than the sheet. Four pieces of 716 mm look like they belong together until you add them.
3. A leftover strip is a few millimetres too narrow. A 100 mm part is tucked into a 97 mm gap, which reads as close enough to a language model and impossible to a saw.
4. Grain is ignored. A part marked as grain-along-length is silently turned through ninety degrees to make a row fit.

Some assistants can run code. If yours can, a request to write and run a packing script, then print each sheet's pieces with positions, is far better than a request for a list. The output still needs the same checks, because a script can pack parts wrongly too. If you prefer to bring the whole question into one tool, the AI in carpentry tutorial shows where cut lists sit among takeoffs and quotes.

One test job: four base units from 2,440 by 1,220 mm sheets

Here is an illustrative job to try on any tool. Four base units in 18 mm veneered ply, cut from 2,440 mm by 1,220 mm sheets with a 3 mm saw kerf. Sides and doors are visible, so their grain must run along the 2,440 mm length of the sheet.

That is 28 pieces. The areas add up as follows: sides 3.226 square metres, doors 1.707, bottoms 1.263, shelves 1.218 and stretchers 0.451, a total of 7.865 square metres. One sheet is 2.9768 square metres, so the job needs 7.865 divided by 2.9768, which is 2.64 sheets. You cannot buy 0.64 of a sheet, so the minimum is three, and no tool or assistant can legitimately report fewer.

At three sheets the waste is fixed whatever the layout: 1 minus 7.865 divided by 8.930 (three sheets) is 11.9 per cent. Three is also achievable with the grain locks and 3 mm kerfs respected, and you can check this layout by hand. Rip sheet 1 into two 596 mm strips: one takes three doors and two stretchers turned across it (2,363 mm of the 2,440 length, kerfs included), the other one door, two sides and two stretchers (2,371 mm). Rip sheet 2 into two 564 mm strips, each taking two bottoms, two shelves and two stretchers, all turned so their 564 mm edge runs across the strip (2,418 mm). Cross-cut sheet 3 into three panels, each 720 mm long and the full 1,220 mm wide, and take two sides from each.

The layout decides something else: whether that 11.9 per cent comes out as usable offcuts or as slivers. In this one, about a third of it sits in a single 271 by 1,220 mm piece at the end of sheet 3, enough for four more stretchers on the next job; the rest is narrow strips and saw kerf.

Check any cut list in five minutes

Run these checks in order on whatever you are given, from software or from AI. Each one catches errors the others miss.

1. Area check. Add up the part areas, divide by one sheet's area, round up. The list must not use fewer sheets than that number. Here: 2.64, so at least 3.
2. Waste check. Recalculate waste as 1 minus part area divided by (sheets x sheet area). If the tool says 7 per cent for three sheets, either its sheet count or its part list is not what you think. A figure that includes saw kerf or edge trim will be slightly higher; ask which.
3. Count check. Total the pieces by name against the job sheet: 8 sides, 4 doors, 4 bottoms, 4 shelves, 8 stretchers, 28 in all. Missing pieces and duplicated pieces both happen.
4. Row and width check. Along the 2,440 mm length, add the pieces plus a kerf for every cut. Three sides: 3 x 720 = 2,160 plus 6 for two kerfs is 2,166, so it fits with 274 mm spare. Four doors: 4 x 716 = 2,864 before any kerf, so it does not fit. Across the 1,220 mm width, two rows of 560 mm parts use 560 + 3 + 560 = 1,123 mm, leaving 97 mm, and only about 94 mm once the last cut takes its kerf. A 100 mm stretcher will not go in that strip.
5. Grain check. Every locked part must have its grain edge along the sheet's long side. A door that appears as 596 by 716 in a layout whose x-axis is the sheet length has been rotated.

A spreadsheet can total the first three checks for you; if you want a head start on formulas, the ChatGPT and Excel guide shows how to have an assistant write them. Ten minutes of setup repays itself across every job after the first.

What an AI answer to the test job can look like, and what to fix

Now paste the parts table into a chat assistant with this prompt:

Here is a parts table for 18 mm veneered ply (2440 x 1220 mm sheets, 3 mm kerf).
Sides and doors are grain-locked: the first dimension must run along the 2440 mm
edge. Produce a cutting layout that uses the fewest sheets. For each sheet, list
every row with its pieces and their lengths, then give the total sheets and waste.

An illustrative answer might look like this:

Sheet 1: Row 1 holds the four doors side by side. Row 2 holds four sides, rotated to fit. Sheet 2: four sides, four bottoms, four shelves, and the eight stretchers in the leftover strip. Total: two sheets, about 7 per cent waste.

It reads well, and every check breaks it. Area: two sheets hold 5.95 square metres and the parts need 7.865. Waste: three sheets cannot waste less than 11.9 per cent, so "7 per cent" cannot be right. Row: four doors need 2,864 mm in a 2,440 mm row. Grain: sides "rotated to fit" violate the lock. Width: stretchers in a leftover strip are 3 to 6 mm too wide for it. Five errors in four sentences is not unusual for a text-only answer, which is why the checks exist.

The fix is not to argue with the assistant. Either ask it to write and run a packing script and print positions, then run the five checks again, or feed the parts table to an optimiser. Either way, the parts table was the useful part of the exchange.

Where AI earns its place in cut-list work

The useful work happens at the edges of the layout, where language matters more than geometry.

From messy notes to a parts table

Before: "3 base units 600 wide x 720 high x 560 deep, 18 ply, sides and doors face grain up, doors 2 mm gap each side, one shelf, a bottom and two rails each, plus a fourth unit the same as the others." After, a draft table from an assistant, with an error you should spot:

Side 720 x 560 x 6; Door 716 x 596 x 3; Bottom 564 x 560 x 3; Shelf 564 x 540 x 3; Stretcher 564 x 100 x 6.

The notes said four units in the end ("a fourth unit the same"), and the assistant counted three, so every quantity is a quarter short. The count check catches it immediately: 8 sides, not 6. The assistant did the dull work of working out the sizes from the carcass width, but you supplied the number of units. The tutorial on building a cutting list and material order from a drawing goes deeper on this stage.

Edge banding and order text

Ask the assistant for edge-banding lengths, then verify them. Doors: each has a perimeter of 716 + 716 + 596 + 596 = 2,624 mm, and four doors need 10,496 mm, or about 10.5 metres. With a 10 per cent allowance for starts and trims that is 11.5 metres. The same assistant can turn the parts table into an order for your supplier: "Please supply three sheets of 18 mm veneered ply, 2,440 x 1,220 mm, and 12 metres of matching edge banding for the doors, for delivery on [day]." Check the sheet count against the area check before you send it.

Reading the optimiser's output

Paste the optimiser's result into an assistant and ask it to find rows that exceed the sheet length, grain-locked parts that appear rotated and parts missing from the count. This is a good use of AI, because the model checks arithmetic against text you supplied rather than inventing a layout. A prompt that works:

Below is a cutting layout exported from my optimiser and my job sheet.
Sheet size 2440 x 1220 mm, kerf 3 mm. Grain-locked parts: sides and doors.
1. Total each part name and compare with the job sheet; list any difference.
2. For each row on each sheet, add the lengths plus one kerf per cut and say
   whether the row fits within 2440 mm. Show the sum.
3. List any grain-locked part whose long edge is not along the 2440 mm side.
4. Do not suggest a new layout. Only report what you find, with the numbers.

An illustrative reply: "Count: sheet 2 lists 3 shelves, the job sheet says 4, so one shelf is missing. Row check: sheet 1, row 3 holds three sides and a bottom, which is 3 x 720 + 564 = 2,724 mm before any kerf, 284 mm more than the 2,440 mm length. Grain: no locked part is rotated." That reply earns its keep twice over: it found a missing shelf and a row that cannot be cut. You still confirm both findings against the printout, because the model can misread a table, but you are now checking two named rows instead of hunting through three sheets.

The dummy-part test for grain, kerf and trim

Any tool can state it supports grain locks. Prove it with a job built to fail. Add a part that cannot be placed without rotating it, and lock its grain; add a second part sized to fit only if the kerf is ignored; add a third that fits only if no edge trim is taken from the sheet. A good tool will either refuse, add a sheet, or flag the part as unplaceable. A bad one will quietly rotate, overlap or ignore the trim.

Then compare the tools on the same four-base-unit job, noting the sheet count, the waste, the number of cuts, the time to set up and any error you found. The table below is a filled-in illustration of how that comparison might look, not a measurement of the products named earlier.

In this illustration Tool 1 wins, even though Tool 2 was quicker: one rotated door is a ruined veneer face, and the full-width offcut is four stretchers for the next job. Your own comparison will show different numbers, but the method is the same.

Your first 30-minute test, step by step

1. Pick a finished job where you remember how many sheets you really used. Pull the old parts list from your notes or invoice.
2. Confirm the real sheet size. Write down the sheet size from your supplier's delivery note, not the default in the tool. A sheet that is 2,500 by 1,250 mm on the screen and 2,440 by 1,220 mm in the yard gives a layout you cannot cut. Note any trim allowance your supplier's edges need.
3. Measure your kerf with a test cut and callipers. Enter it. Do the same for the other saw if you use two.
4. Enter the parts with quantities and grain locks. Time yourself: the setup time is a real cost that goes into the comparison.
5. Run the five checks from earlier on the result and write down any failure.
6. Add the dummy part for grain, kerf and trim and run again.
7. Compare with what you did last time: sheets used, offcuts you kept, and the minutes the old method took. Keep the table.

The sheet-size slip in step 2 is worth an illustration. A carpenter leaves a tool on a default sheet of 2,500 by 1,250 mm. The plan stacks two rows of 560 mm parts and a strip of 100 mm stretchers across the width: 560 + 560 + 100 plus three 3 mm kerfs is 1,229 mm, which fits in 1,250 mm. The merchant delivers sheets of 2,440 by 1,220 mm, which are 9 mm too narrow for that arrangement, so the stretchers need another sheet. Nothing in the software failed. It did what it was told with the wrong input.

What a bad list costs against what a tool costs

Say an 18 mm veneered sheet costs you $95; that puts the stakes in proportion. A list that makes you buy a fourth sheet you did not need costs $95 once, which is nearly twenty three-day passes at $4.90 each. A rotated door can cost a new sheet and a day's delay. Money is not the only cost either: a missing part found on site means a trip back to the workshop. Against that, the free tiers and the low-cost options cost almost nothing, so the sensible question is not whether to use a tool, but which one passes your dummy-part test.

Time is the other half of the sum. If laying out a kitchen's worth of parts on paper takes 45 minutes and the tool takes 12 including entry, that is 33 minutes saved, or $22 at $40 an hour. Four such jobs a month return $88, and that is before counting the sheets you stop wasting.

For jobs that repeat, such as weekly kitchens, a desktop program with saved stock sizes and labels pays back over weeks. For a one-off wardrobe, a pass or a free tier is enough. If you run a shop with several tools and want the whole chain from quote to order, the cabinet shop software tutorial compares what that costs.

Choosing by the kind of job you are doing

Whatever you choose, keep the last job's list and the sheets you actually used. That record is your benchmark: the next list, from any source, should use no more sheets than the last comparable job, and you will see within a few jobs which source can be trusted.

Questions carpenters ask about cut lists

Is there a free cutting list tool good enough for one carpenter?

Possibly. As listed in October 2026, OptiCutter has a free tier capped at 5 stock sizes, 20 part sizes and 500 parts in total, and MaxCut offers an always-free Community Edition for Windows. cutlistoptimizer.com limits free use and sells a $4.90 three-day pass. Compare each cap with your biggest job before you rely on one.

Can ChatGPT make a cut list from a photo of my drawing?

It can often read clear dimensions and produce a parts table, but misreads happen, especially with handwriting, fractions and repeated units. Ask it to list every assumption, check each quantity against the drawing, and then paste the table into an optimiser for the layout. Treat the table as a draft.

What kerf should I enter in the software?

Use your own blade's cut width, measured on a test cut with callipers, not a figure from a forum. A table saw, a track saw and a panel saw cut different widths, and the error adds up across a sheet. If you use several saws, set the kerf per job.

How do I stop the software rotating veneered parts?

Look for a per-part setting that locks orientation or grain, then prove it with a dummy part: enter one locked part that would fit better rotated and see whether it ever comes out turned. If it does, your version is not honouring the lock, and you should not trust it on veneered board.

Further reads

• Best AI App for Carpenters: Measuring, Quoting and Invoicing Compared — Measuring, quoting and invoicing apps compared for carpenters.
• Cabinet Design Software: What AI Adds for a Small Cabinet Shop — Compare cabinet design software by price band, CNC output and cut-list accuracy, and see what AI renders and drafting really add for a small shop.
• Joinery AI App Checklist: What to Check Before You Pay for One — A buying checklist for a joinery workshop: test files in and out, a hand-checked cut list, full monthly cost, export and ownership on one finished job first.
• AI Tools for Welding Shops: Quotes, Drawings and Job Scheduling — A welding shop owner's guide to AI on the office side: a drawing-based quote check, cut lists, a schedule stress test and where weld sign-off stays human.
• AI for Carpentry: A Site Measure Checklist From Notes and Photos — What to record at a measure visit, how to label photos so AI can use them, and how to turn voice notes into a sheet that flags gaps before you leave.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: cutlistoptimizer.com, opticutter.com, cutlistplus.com and maxcutsoftware.com feature and pricing pages, read October 2026; area and packing arithmetic worked by the author for the illustrative test job.