This Guy Attached 21 Chef's Knives to a Slicing Robot Arm to Determine Which One Is Best

WIRED

15-07-2025

Jul 15, 2025 7:00 AM Certified kitchen knife nerd Scott Heimendinger used a robot arm on multiple chef's knives to collect 100,000 data points about which blades cut most efficiently. He's ranked them from best to worst. Scott Heimendinger's slicing robot arm prepares to apply the blade of a chef's knife to a helpless piece of bread. Courtesy of Scott Heimendinger
People occasionally ask me if AI is coming for my job. I'm pretty hands-on, which gives me a bit of a feeling of security. But that feeling dropped a bit when I saw a robot with a knife in its gripper, testing its edge.
The robot is a side project from Seattle Ultrasonics, a tiny operation run by Scott Heimendinger, an alumni of Modernist Cuisine and Anova, and cofounder of the beloved Sansaire sous vide company. Before all that, he was at Microsoft, working on software that became Power Query and Power BI.
'I'm pretty ridiculously good at Excel,' Heimendinger says, referring to his time at Microsoft before pausing. 'I'm very good. I was a whippersnapper.'
More recently, he's been working on creating an ultrasonic knife for home kitchens, due out later in 2025. It looks like a regular chef's knife, but it vibrates almost imperceptibly, allowing it to cut incredibly well.
The knives tested in Heimendinger's lab. Courtesy of Scott Heimendinger
But it's Heimendinger's data-gathering side project, which he calls the Quantified Knife Project, that has grabbed my attention. For the experiment, he bought 21 chef's knives, attached them to a robot arm, ran them through a battery of tests, then crunched the gobs of data to calculate a "food cutting rank."
To do all of his testing, he did a weekday's worth of ingredient shopping, queued up the Real Genius soundtrack, and ran his robot arm hard over a full weekend to collect the data. Five different foods sliced five times each, times 21 knives. That's 525 individual robot cuts, from which Heimendinger accrued 100,000 data points for his rankings.
Dedicated knife nerds will be happy to know that he also ran each knife through a BESS sharpness test that measures the force needed to cut through a synthetic wire and sent the blades away for CATRA edge-retention tests. He then incorporated that data into the cutting rank.
I suppose I should have been scared, or even jealous, of what he and his robot arm were doing. But mostly, I was intrigued. There were things I was curious about, and while I might have made little modifications to the tests or chosen different knives, mostly it just felt exciting.
The Excel nerd now had reams of data. Spreadsheets galore! By crunching the data, he could rank the knives. I wanted to see for myself how he was testing, learn more about the knives he thought were best, compare our testing styles, and see how my favorite knife compared to his. A Cut Above
I put that favorite knife, a Glestain with enormous dimples, in my knife roll and went to visit Heimendinger at his downtown Seattle lab. There's a lot to take in in his tiny space: a drill press, 3-D printers, specialized photo and video gear, a tiny shop vac, optical and digital microscopes, his 'crazy handy' oscilloscope. There's also the robot arm, a uFactory xArm 6, which he has dubbed Dr. J. Robot Choppenheimer.
You can move this "cobot,' or collaborative robot, around while it records the movements, then it will mimic those movements over and over again. "It means it knows how much force is being applied and can stop if too much resistance is being met," he says. In other words, if you've got the settings right, it won't chop through your table or your arm.
Scott took my knife, attached it to the robot's gripper, and started collecting data, running through the same battery of tests as his other 21 knives. First, it cut through five tomatoes, one at a time, through the equator. He mounts everything he's cutting to a 3-D printed plate with spikes that keep it from rolling or sliding away. That sits on a scale connected to his computer, allowing him to create a graph of how much force is being applied as the blade moves through the food. Tomatoes, for example, require a knife edge to exert a relatively large amount of pressure to break through the skin, then much less to complete the cut.
After the fifth tomato, I had to know how my knife did. Without judgment, Heimendinger announced that it placed "second to last in the cohort." I felt responsibility for this knife, and a little embarrassment, until I remembered his knives came straight out of the box and into the testing process with an unblemished factory edge. My knife, on the other hand, had been in regular use in my test kitchen for more than six months receiving only the occasional honing. Plus, I noted protectively that these tomatoes appeared bigger than the ones he used during his main battery of testing.
He continued through the other four foods: potatoes, cheese cubes, baby carrots, and King's Hawaiian rolls, the latter chosen for their extremely uniform interior.
A test knife slices through a piece of cheese. Courtesy of Scott Heimendinger
Bread rolls were also used in the testing. Courtesy of Scott Heimendinger
Understandably, my knife didn't fare particularly well, but I was able to get an appreciation for Scott's testing and data-gathering process.
Even with the robot, collecting this amount of data took a lot of time. Every piece of food needed to be loaded and unloaded from the scale, the knife wiped, cleaned, and dried after every stroke, the room kept cool, the whole thing happening during that monotonous bender of a weekend, Don Henley and Tears for Fears playing over and over.
Once he got all that data and made dozens and dozens of charts and graphs, what did he learn?
'How scattered the results are.'
Per his testing, three chef's knives were fairly blazingly fantastic, doing well across the board: a Shun Classic Hollow Edge, a Moritaka Hamono, and a Tojiro Professional. Number four was weird: The $300 Wüsthof Amici (very similar to their fantastic Classic but with a different bolster and handle) aced everything except the carrots, at which it was quite bad. The last two slots, 20 and 21, were also well secured, by a Henckels Classic and the $18 Zwilling Solution Fine Edge.
Yet the stuff in the middle—slots five to 19, more than two-thirds of the test group—were what he was referring to when he said 'scattered,' performing well in one category and poorly in another.
"You would think that a great tomato knife would make a great potato knife," he said before noting that wasn't necessarily the case. "It's bananas."
Tomatoes are a real test of slicing ability. Courtesy of Scott Heimendinger
Same with potatoes. Courtesy of Scott Heimendinger
Those three knives at the top of his rankings feel like safe bets. I'd even feel pretty good about lumping that Wusthöf in there. And if I had been considering purchasing one of those two at the bottom of his list, I'd abandon the idea. But those 15 in the middle? What about them?
For his part, Scott appreciated their lack of predictability.
"Nobody's done this sort of evaluation. This might be the first time we're understanding that what matters for a tomato is different than what matters for a potato," he said. "When you get these kinds of answers in science, these are the most exciting."
So what is best for a tomato?
I thought he was going to generalize about blades, but instead he said the Wüsthof Amici, which, thanks to a particularly well-honed apex, just sliced right through.
"It's all about breaking the skin!" Personal Touch
To me, his testing and the way he chose to do each test leaves a fair amount of subjectivity to the whole thing. Then again, the way I test probably feels subjective to other reviewers. There's stuff in his criteria for everyone obsessed with knives and robot arms and databases, and there are other things that other testers would change.
If it were my test, I'd make Dr. Choppenheimer's tomato motion more of a slice. I'd use bigger hunks of cheese in the cheese test, I'd throw out bread as a criteria because I don't use my chef's knives as bread knives, and maybe I'd reposition the carrot to be closer to the heel of the knife where I usually cut them.
Something I'd want to feel more sure of are things this testing doesn't take into account. How comfortable is the handle? Is the top of the spine rounded where pinch-grip user would lay their index finger over it so they get blisters while chopping up a bag of onions? Is the blade tall enough so the fat fingered among us don't whack their knuckles on the cutting board? Is it weighted in a way you like? Some knives are center-weighted while others feel heavier toward the front of the blade or back in the handle; years ago I tested a Shun nakiri that was super weight forward, which, I learned, was not my jam. What you like about a knife is very personal.
If you don't take stuff like that into account, the possibility that you'll like your new knife is kinda up in the air. For Heimendinger, this is by design. He intentionally leaves out what he calls "objective and personal stuff."
I took my Glestain to a professional sharpener then, a few days later, brought it back to Heimendinger to run the tests again. I'd noticed his ability to be dispassionate about his wall of beautiful knives he'd amassed for testing, but I still I hoped he'd be interested, even excited about mine, with its monster dimples. But no. Mostly, it was just the 22nd knife he ran through its paces, though he did say "Nice! These will make for better numbers," when he ran it through the BESS tests.
As someone who wears his knife-sharpness pride on his sleeve, I was happy to see that the Glestain leapt into eighth place in his rankings after it had been sharpened.
Not that the two of us had staked out opposing positions on that first day in the lab, but I started appreciating how this data would make a nice addition to my kind of testing. He certainly understands the value of hands-on time with whatever blade you're researching. If you buy a knife that feels good and is backed up by his data, I dare say you'll be ready to slice and dice to your heart's content.
"There isn't just one test or one answer to how a knife performs. Hold one. See if it speaks to you. You could hold the worst performer and love it," he says. "I hope you are happy together."