Clay Tech

"clay-works make things real"

translated from clazytech.com

Judging Japan's Monozukuri Hardware Cup Taught Me What Judges Really Do

I had the opportunity to serve as a judge at the Monozukuri Hardware Cup (held online) on February 25th. https://hardwarecup.monozukuri-startup.jp/

It’s positioned as the Japan qualifier for an event whose final is held in Pittsburgh. (This year the whole thing seems to be held online.) This was my first time judging a pitch event rather than presenting at one, and having actually done it, my first honest thought was “presenting is way less nerve-wracking than this.” lol A judge, even more than the presenters, has to come up with questions within a limited time (usually only one or two questions are allowed) that anticipate to some degree how the presenter will respond, that lead to a productive exchange, and that are also useful for the judging itself. It took a lot of mental effort. But it was tremendously interesting, and I’d be glad to help out again if I get the chance.

Since I have the opportunity, I’d like to introduce some of the startups that presented and add my own perspective along the way. (In the actual event I commented only on the third-place startup, but I have things to say about a lot more companies than that.) Through these comments I hope to share some of my own essential points on how to shape the direction of a business plan or vision.

Let’s start with Water Design Japan and Magic Shields. https://www.waterdesign.tokyo/ https://www.magicshields.co.jp/product-/

Water Design Japan eliminates the need for cleaning water pipes by attaching a nozzle that generates a special kind of bubble. The bubble technology itself already exists, and products on the market take the form of things like high-end external devices or units built into washing machines, but their product just requires swapping out the pipe’s nozzle. Magic Shields makes a mat from a specially structured absorbent material that normally behaves like a hard floor but turns into a cushion the moment a fall occurs.

These two companies genuinely opened my eyes. The hardware industry these days has, in a sense, become extremely software-like, and products built around massive platforms and arrays of sensors, with big data and AI thrown in, are everywhere. What’s striking about both of these companies is that an extremely simple hardware product is simply solving a problem. To be clear, that’s not something we judges were specifically evaluating. No one said anything like “this is too software-leaning, so it’s no good.” But in an industry where it’s easy to let your guard down and assume everything’s already been done, it was impressive to see that simple hardware technology can still deliver this much social significance, market potential, and room for horizontal expansion. It made me want to start racking my brain again from tomorrow. Another strength both companies shared was pricing that matched the market. When you build a hardware product meant to solve some minor inconvenience, it often ends up costing $200 or $300 and risks becoming a product only early adopters find fun enough to buy. But because their products are simple, they can price them in a range that genuinely competes with existing, ordinary products on the market, and I think that’s a real strength.

Next, CYBO. https://www.cybo.jp/

It’s fair to call them a startup that came out of ImPACT and the University of Tokyo. They’re developing an integrated system aimed at early detection of cancer cells. I think “integrated” is the key word here, and in a sense it’s a bet. As I understand it (as a layperson), the detection process involves imaging, analysis, and sorting: looking at cells (hardware), judging them (software), and separating them out (hardware). The process itself isn’t built around some wildly novel idea; the point is the efficiency, speed, and precision the integrated system achieves. With something like the AI component, there’s generally a lot of competition, and it’s very hard to simply rank who’s better. But in terms of performance as an integrated system, there’s a real chance of reaching something no other company can match. The key, I felt, is how much you focus not on being somewhat better, but on reaching a point where other companies simply can’t follow.

Next, LANGUALESS, the company that went viral with Inupathy. https://www.inupathy.com/langualess/

It’s a product that picks up heart rate information via microphone and uses it to estimate a dog’s emotional state. Technically, I imagine they’re using something like an HRV-based approach, which inevitably raises questions of accuracy. There are two possible stances: “it’s an entertainment product, so why worry about accuracy,” versus “no, improving accuracy further will expand the range of applications.” In their presentation they argued the latter, but personally I find the former more appealing. The accuracy required of ECG products used in medical devices, for instance, is extremely high. Having a big vision like “if only we could detect illness in pets” is great, but looking at it soberly, there’s a gap that, technically speaking, has to be called flatly impossible. They may need to rethink the technology they’re using from the ground up. Depending on the case, they might even need to reconsider the “pet” target itself (though at that point, would it even be the same company?). Given that they’ve set their sights on the pet market, whose scale is actually not to be underestimated, I think they should double down in that direction instead: expanding to other animals, or building a platform that collects data linking pets’ heart rate to their condition (in a sense, training data). I felt it would have been better to take a stance of wringing everything possible out of the market using what current technology allows. I imagine ideas along these lines are already part of their plans, but it’s often said that deciding what not to do matters just as much as deciding what to do, and I felt their situation was exactly that.

I can’t cover everyone, so last is Edgecortix. https://www.edgecortix.com/

They’re aiming at an IP business in silicon to enable edge AI. As fellow judge Murakami-san pointed out, this is a space the big players are charging into hard, so the natural question is: what makes you think you can win? In other words, how are you going to manage this when you’re outgunned on funding, organizational strength, and sales power? That’s a tough question to answer. It’s also a very fundamental question that should be directed at almost every startup. A few model answers come to mind:

and so on. It depends on the situation, so I can’t make a blanket statement, but personally I’m not very confident in the, in a sense, clichéd pattern that leans heavily on patents and personnel. The reason is obvious: it’s extremely common for the team to suddenly fall apart, and that’s a major risk. A company that stakes its position on a patent needs to make full use of that patent’s advantage, and it’s engineers who do that work. The fantasy that simply holding a patent translates into money doesn’t exist. You need a reference product that demonstrates its effectiveness, you need to shape the proprietary technology into a form partners can easily use, and you need to keep providing FAE-style support as those partners adopt it. Engineers do most of that work, so unless you have engineers who can make full use of the patent, it might as well be meaningless. Then there’s personnel. Nothing is as hollow a claim as “personnel.” Startups are especially prone to high staff turnover, and because teams are small, they can fall apart over the slightest thing. One exception is when the key technical person is a co-founder, or when the founder and CEO is themselves the researcher or inventor. I asked a question along these lines, and since they answered in terms of “members” (the CEO himself has no background in silicon), I have to admit that worried me a little. That said, the figures they presented in their pitch, like reductions in power consumption, were impressive. But that kind of claim is hard to deal with in a pitch contest. There’s no way to debate the validity of those numbers on the spot. So when an idea about the target market comes up in response to a question, that’s the kind of thing that really grabs my attention.

This ended up being a somewhat abbreviated write-up, but I’ll close here. I hope this is of at least some help to anyone who’s trying to launch a startup, or who has already launched one and is preparing to take on a pitch contest.


Originally published in Japanese at https://clazytech.com/2021/03/396/. Translated with LLM assistance and reviewed before publication.