An interview with Jürgen Pansy, Chief Innovation Officer and co-founder of smart lock manufacturer Nuki (link). The company from Austria has announced a model with the Matter standard for 2023. What to expect from the product, and why some developments for Matter take so long, Pansy explains in this conversation with matter-smarthome.
Dies ist die Übersetzung eines deutschen Interviews.
Zum Original bitte hier entlang.
It is well known that Nuki is working on a new Smart Lock with Matter. Now you have explained that you succeeded in establishing a third locking state within the Matter standard: in addition to unlocking and locking, also pulling the latch. Why was that important?
Jürgen Pansy: A standard that not only unlocks and locks, but can also unlock without pulling the latch is something we’ve always wanted, already in HomeKit. Because this third state and the knowledge that a lock supports it, we think is important for European doors. But HomeKit was an Apple-only event and the product managers are all based in the US. They have little idea about European doors, which is why it was difficult for us to get through there.
Matter is now HomeKit for everyone, so to speak. Apple has brought its experience with HomeKit into Matter. Google has done something similar with the radio protocol Thread, which they kind of bought when they took over Nest. Other companies like Amazon are joining in, and we can now also get involved because the whole organization operates according to democratic principles.
How should we imagine this in practice?
Pansy: We have made a proposed change to the Matter specification. If such a proposal finds enough parties to accept it, it goes on the list for the next Matter version. In our case, that is Matter 1.2, which will be approved in the fall. There are deadlines within which an objection can be made, and if everything goes through, the feature will be included in the standard.
But that’s not all because the Connectivity Standards Alliance (CSA) doesn’t just produce written paper, in other words the specifications. There are also test plans and tools for the certification program, and a software development kit (SDK) with sample applications that can be installed on the various platforms. For a feature to really make it into the standard, it has to be considered in all these places. That’s what we’re working on right now. Months can pass over all the objection periods and resolving objections, but that’s the way things are in a democratic process.
„A standard that not only unlocks and locks is something we’ve always wanted, already in HomeKit.“
Who supported your application, even competitors?
Pansy: I can’t give any names, just this much: They weren’t lock manufacturers. We looked for people who could implement this with reasonable effort because none of our competitors are engaged in the CSA. Nevertheless, everyone benefits in the end because all the work on test cases and the SDK is for the public. Everyone can use the results later and build a Matter-enabled smart lock with them.
Such non-profit activities are not everyone’s cup of tea, firstly, and secondly, you only do them when you have time. We don’t have as many resources as big companies, so we limit ourselves to restricted areas like door locks and try to push the standard at this point. Also because no one else is doing it.
Behind it all is the hope that a common language will move our entire industry forward. A notable example here is Apple. They took their HomeKit, which they spent ten years shaping and developing to a certain level of maturity, and put it into open source. There are Apple employees who write code for Matter to advance the standard. And they’re not bad. In the CSA, you are sitting at the table with the same people who developed the Bonjour protocol that is used everywhere today.
So the new locking state comes with Matter version 1.2?
Pansy: Theoretically yes, but having the feature in the standard is only half the story. Matter controllers must also implement and use it. The Door Lock Cluster is a good example. To understand: There is a so-called cluster for each application in the specification, such as the Lightning cluster, the On/Off cluster for sockets or the Lock cluster.
The specification for door locks is 40 pages long. It covers things like creating new users, keypad login to the lock, and time-limited access. Why? Because the CSA, which is also the standards organization for Zigbee, made a point of including existing Zigbee clusters in the Matter standard. Now all the features that have ever been defined for Zigbee locks are also in Matter. I deliberately say that they are only in there because Apple, Google & Co. have essentially taken two commands from the 40 pages: unlock and lock.
Establishing innovations such as unlocking without pulling the latch is not trivial because the user interfaces do not currently provide for such a feature. The door is either open or closed. The fact that I want to unlock without pulling the latch requires an additional control element in the app. You can imagine what it means to get a new UI element in Apple Home, for example, that doesn’t require another device yet. Even more so for an optional feature that practically only Europeans are interested in.
“Innovations such as unlocking without pulling the latch are not even provided for in the user interfaces.”
Does that mean it may take until next year for the feature to show up in Matter ecosystems?
Pansy: It will pretty much take that long. For one thing, a smart lock can only handle the functional scope of Matter 1.2 once Matter 1.2 has been released. It is quite possible that we will launch the lock on the market and Matter 1.2 will only follow later via an update.
And yes, then the Matter controllers have yet to incorporate the feature. Whether they do that is entirely up to them. Things could go a bit faster with the Home Assistant, currently the only controller on the market that itself is open source. There could quickly be a reference implementation for this platform. It will certainly take a while for the big ones because the UI problem is not that easily solved.
What about other features in the Door Lock cluster? Will they ever find their way into the platform apps?
Pansy: This is a chicken-and-egg problem. We could, of course, support all these features from Matter, but if there are no controllers to implement them, they are useless. On the other hand, there are strategic decisions about which features we want to offer with our own app and don’t want users to use them via Matter at Apple, Google & Co. . One conceivable way would be that I can install a Matter-enabled Nuki via the Matter platforms without using our Nuki app. But for more advanced functions, users then download our app.
With light bulbs, this path is virtually mapped out because they are relatively “dumb”. Their basic functionality does not require an app of its own. The Eldorado that Apple and the other big players have in mind: You go to a supermarket and pick up a random lamp with a Matter code from a massive selection – whether it comes from China or elsewhere doesn’t matter. That’s the reason why the big players support Matter, we have to be honest.
Security has always played an important role in the development at Nuki. Is the connection via Matter as secure as the previous technology?
Pansy: In any case, it is not more insecure. After all, the Matter standard assumes that devices are hanging around somewhere and can be attacked from all sides. High security requirements are one of the reasons why the introduction of Matter is taking so long. The situation is somewhat reminiscent of the launch of HomeKit anno 2014, when it was almost impossible to get it onto an existing chip because the encryption was so computationally intensive. We tried with our first-generation smart lock, but pairing and unlocking took far too long in practice.
So we made the Smart Lock 2 with a better chip and a better library. And now it’s similar again. The Matter SDK is huge in scale. To validate whether a device is certified, there is communication with a blockchain. Many certificates are used, which is all computationally intensive, and means that the existing hardware is often not sufficient.
The current Nuki Pro works with Wi-Fi. Is that an option for Matter or do you go for the more energy saving-wireless protocol Thread?
Pansy: During the development of the Matter SDK, it has always been assumed that battery-powered devices will operate with Thread and hard-wired ones with Wi-Fi. The current Nuki Smart Lock 3.0 Pro proves that it can be different. It’s a little marvel of technology that can stay connected via Wi-Fi for months, even though the batteries also have to power the motor and turn the key. I never thought we’d get this far.
It owes its efficiency not only to the hardware, but also to our protocol behind it. Unfortunately, the Matter protocol for Wi-Fi is not as efficient. There is much more communication, which comes at the expense of battery life. That’s why Thread is the better choice for us. Apart from the advantage that I only need a Border Router and the rest magically organizes itself. I don’t need a bridge, like for Bluetooth. We actually want to get rid of those.
“Unfortunately, the Matter protocol for Wi-Fi is not as efficient. That’s why Thread is the better choice for us.”
Does this mean that there will be no Matter Bridge for the current Nuki Smart Locks? This would allow Matter to be retrofitted on them.
Pansy: I’m not saying there will never be anything like this from us, but basically no one wants a bridge. It was always just a tool required for IP connections. An ideal smart lock doesn’t need a bridge. And we want to build the ideal smart lock.
Does Matter change anything about how the lock is connected to smart home systems? Currently there are different ways: HomeKit, the online service Nuki Web and your HTTP APIs?
Pansy: That would actually be our hope, that at some point everything would work with Matter. But to achieve this, we would first have to get the many features of our APIs into the Matter specifications – so that they are accessible via a Matter connection. And Matter, with its QR code and other things, is currently even harder for system providers to integrate than an HTTP API. That will change as it becomes more widespread. Because more and more people are writing instructions and modules that can be integrated. This reminds me of the early days of HomeKit. Apple came up with a specification with 300 pages, and there were test cases with 500 pages, and everything had to be considered.
Today, you buy a chip from a vendor like SiLabs and get a library with a lot of things already pre-certified. That makes things a hundred times easier. So maybe we’ll get to the point where we can phase out our own APIs for basic functions. When that will be, or whether it will even come that far, is not foreseeable at the moment.
Currently, the standard struggles with incompatibilities. Things work in one system, but not in another. Does that surprise you?
Pansy: No, I’m not surprised. It’s the kids’ shoes. It was the same at the beginning with HomeKit and it can’t be avoided when so many devices come together. Certification doesn’t change that. Because what the CSA does is only check the individual product, such as our smart lock. How it behaves in interaction often only becomes clear in practice. There are so many scenarios that cannot all be tested. Not even at the regular CSA test events. In addition, there are bugs because the solutions are very new. And all of this is then met with high expectations that were stoked at the beginning.
Some manufacturers would like to see platform operators Amazon, Apple, Google and SmartThings more involved in certification. How do you see that?
Pansy: The platforms can pick and choose from the portfolio what they want to support and what they don’t want to support. Yes, that’s the case. But I think it’s unrealistic to demand something like that when you look at the market reality. Companies are too different to be expected all to do the same thing. In some of them, cost-cutting programs are underway, entire departments are being laid off – and with them employees who worked on the Matter standard. Not everyone can devote the same resources as Apple.
“Entire departments are being laid off – and with them employees who worked on the Matter standard.”
After all, we wouldn’t like it if all smart locks had to offer the same features via Matter. Providers should decide that for themselves. The best thing is for the market to regulate it: If customers complain to Amazon that something that works with Apple doesn’t work for them, that will have an effect at some point. I think it’s better if more decisions are made on the free market and there aren’t too many requirements – except for a common language and the associated possibilities.
But – and this is the most important thing – you have to give it time. After a start, there’s always a valley of tears, and when you’re through it, you’re happy that things are going up again.
Mr. Pansy, thank you very much for this interview.
Share this information:
