Construction projects are full of repetitive, tedious, but necessary tasks, like filing daily reports, administering orientations, and so on. Nyfty.ai creates 'bots' that handle critical tasks like these, and leverages powerful natural language understanding to make interacting with these bots fast and easy. As more and more processes digitize, understanding how to automate these processes will be critical.
Matt Edwards
Hugh Seaton: [00:00:00] Welcome to Constructed Futures. I'm Hugh Seaton. Today I'm here with Matt Edwards co-founder of Nyfty.ai. Matt, welcome to the podcast.
Matt Edwards: [00:00:12] Thanks Hugh great to be here.
Hugh Seaton: [00:00:14] Great. So let's start with what Nyfty does. I'd like to just sort of set some context and we can talk about this fantastic journey you've been on.
Matt Edwards: [00:00:22] Yeah, sure. So we make what we call these little robot helpers. It is a natural language understanding based product, that we try and complete tasks that would otherwise have to be done by say a superintendent or a project engineer. So we kind of almost think of ourselves as a skilled labor hire company, but we do it through software.
Hugh Seaton: [00:00:48] And let's break apart some of that. But before we get into what you meant by that, talk to me about how you guys started. Cause I knew you as a voice company and you've evolved over time. And I'd love to hear about that.
Matt Edwards: [00:01:01] Yeah, sure, sure. So back in July 2019, and we released an app, a voice app, which is a companion application for Procore for the daily log in Procore. Completely voice-based, you touch the microphone and said, Hey, log this manpower for me, this company had 10 guys is here, that worked for so many hours doing this on, in this location. And Nyfty would pull all that information out of the voice information and then log that in Procore for you. And you could do lookups say, Hey, how many dumpster pulls have there been this week? And things like that. And we could very quickly give an answer back and use voice as the main medium for that.
Which was really cool. I mean, the, the reason David and I, my co-founder David started the business was because we, we really just kind of love the idea of voice and using natural language instead of clicking user interface. And that was good, we really got some great traction. But we'd always thought of this third part of the system.
If you'd like, the voice app did two things, you could ask questions and then you could also log information, but we asked ourselves, what else can we do with this technology? How could we take it 10X higher for that. And we said, well, what if you don't have to do the task at all? What if Nyfty could do that for you?
And that, that single kind of idea, that concept almost blew our minds. Really. It was just instead of a different user interface that people can access. What if we just said to them, you don't have to do that task anymore. We're going to do it for you. That's really where we said, well, hang on a minute. Voice is awesome. We love it. I hope I'm going to get back to it one day, but we switched completely from voice to SMS and to building these automation bots, which we call the little robot helpers. So that's the start of the journey, I guess.
Hugh Seaton: [00:03:01] And I think, both of those two ... and obviously there's a ton of work in between speak to the problem that you're looking at, right. Which is out in the field. How are we making it so that. Busy folks in a very complex environment that with a lot of problems to solve, don't have to worry about how to punch in or type with their fingers, in the sun or whatever it might be.
How are you streamlining that? First it was their inputs. Right? First it was what's the, what's the most natural, smooth, and easy way to enter information. Is that, is that how you looked at it in the very beginning was, was we can use this technology to. One part of entering what is frankly a tedious part of their job, less tedious.
Matt Edwards: [00:03:46] Absolutely. I mean, that was the central idea was you're busy, you're mobile. You want to be hands-free, you don't want to be stuck at a desk. How can we do some of that admin stuff while you're on the go? And, you know, I love the app and I talk about the app in past tense, and I just want to be clear, because we killed the app. We killed the voice to make the bots, which is, I'll get to later. So I'm talking about in the past and I have these kinds of fond memories of it because there's videos on LinkedIn of me testing voice stuff next to an excavator, and next to a kind of hammer drill and things and achieving some amazing things, using voice and still logging information.
So absolutely it was about real life, noisy environments. Hands-free, still getting the stuff done while you were in situ while you're looking at the thing you needed it to log. We still think that's a great idea. I really think that's awesome. Anything to do with the voice I love.
Hugh Seaton: [00:04:47] Yeah, the reason I wanted to stick there for a minute is to speak about a trajectory.
Cause what I'm, what I'm hearing you talk about is that you've watched what's going on and really asked yourselves, okay. If the problem we're trying to solve is how do we remove some of the tedium and, you use the word administration and admin, I love that. Cause nobody really loves, well most people don't really love doing the admin part of their job.
And certainly you hear trades people often will say that, you know, that's sort of, and superintendents and so on, that that's not the part of the job that they signed up for. It's necessary and you can't avoid it. But the journey I hear you on is how are we pulling away some of the admin so they can find time to focus on what they're really there for, which is putting work in place.
Matt Edwards: [00:05:34] Absolutely. So we started with, you know my history. I had a previous startup to this, which I kind of sold out of a couple of years ago. And in, in that world, we'd taken paper and gone digital so that you had a digitized version of a piece of paper on a screen.
And it was a little bit smarter than that. There was lots of really cool things it could do, but essentially it was just digitizing paper. And when we started Nyfty and we were looking at the voice app, we were saying, well, Do away with the paper all together and take this a step further so that you don't have to do paperwork in any way, not with a pen, not with an iPad, can we just do it with voice? And that really excited us. That idea, but then when you then hit, well, what if we can just eliminate that, altogether for you. What if you don't have to do it anymore? Cause we can do it for you. That really got us very excited. And that's the kind of the journey we've been on.
And I think that the journey technology will eventually go on. And you, you would know, from kind of basic risk management theory, that the first, best way to reduce a risk is to eliminate the task in the first place. And we, we kind of take that same approach with how do we save you time and how do we get things done for you? And it's not just about making it easy. It's not just the alternative user interface, or an alternative way of talking essentially to an app. It's what if we can just eliminate it altogether? And that's what excites us and drives us.
Hugh Seaton: [00:07:08] So give me an example of what sort of tasks you guys automate, what do one of the little robot helpers do?
Matt Edwards: [00:07:16] Well, I'll talk about our first one, and this is really the transition from voice. So I mentioned earlier that we built a voice app and you could talk to it and log Manpower. And so you say, you know, Jackson electrical, I've got 10 guys here, they've worked eight hours and they're working on level one doing first fix install.
And that was what the app could do. So we built a bot using the same technology that can understand all of these kinds of verbal inputs and, and understand construction vernacular. But do it over SMS and do it without, the superintendent having to do it.
What if we could just reach out to the foreman from that electrical contractor and just say, Hey, what's your manpower log for the day. Just tell me and I'll log it for you. And that's essentially what our first bot was. And we released that in October, 2019,the manpower bot. And it's a super simple concept, extraordinarily complex, natural language understanding that drives that, the ability to understand the myriad of different ways someone can reply to that text message naturally, in natural language
Hugh Seaton: [00:08:22] I've built stuff like this, and it's funny how we take for granted as software creators, how people might say something and then real people don't talk like that. Or some percentage of them decide to come at it with Yoda- like language.
Matt Edwards: [00:08:39] Yeah. Yeah. And we have this gamut of the beautifully simple, we had 10, 10 people for eight hours doing this activity, in this location. This is just a beautiful, simple input. But sometimes we get these almost essays that come back into the bot, too.
You know, we had one guy on level one did this and then three guys for four hours doing this. And then we had, 17 guys in the morning and...
Hugh Seaton: [00:09:06] then Jimmy, Jimmy took a bathroom break and then you get people thinking with their mouth, right. Like that's, what we do often is we're, kind of answering the question in our own mind with while verbally, before we arrive at a final answer and the computer has no idea that maybe we discount all that and wait and see see where he ends up.
Matt Edwards: [00:09:29] Absolutely. And, you know, a lot of people are still replying with their voice when they're doing this. Not, not everyone's replying using the keyboard on their phone. You can see sometimes from the response that someone's hit the dictate button, or has told Siri to reply and they've just talked. And that's fantastic, because one of the reasons we killed off the voice app. And I'm sorry, I'll jump around a little bit, was, we have this strong belief in there are already too many apps in construction technology. There's just been this, this proliferation of apps. Everybody has got an app to solve some problem for some person in the field.
And if you're a worker coming onto a job site, you know, depending on where you go, there's various different apps you're gonna have to download. And we felt that adding yet another app wasn't necessarily being helpful. So SMS gives us the ability to fit into the way that person already interacts with their phone.
So if they're already used to using the voice app on their phone, like the dictate button or the, or to telling Siri to reply, we just do that natively with them and they just reply over SMS and we just get the text input through SMS and then take it from there. We still essentially process voice and we still think of ourselves as a voice-based business.
We just don't actually take the audio of the voice. We take the natural language text input.
Hugh Seaton: [00:10:55] And that's why, I mean, you launched in the beginning of this you said that it's a natural language understanding, which is important for people to distinguish, because I think if people don't spend a lot of time with voice and natural language processing, which most people obviously don't, there's a difference between the audio file and then what it turns into, which is what the computer's really trying to figure out.
So the fact that you guys have maybe abstracted and said, look, we're not going to worry about whether the natural language is coming in because you've used texts or whether you've used, Siri's voice or, Alexa or whatever it might be.
Matt Edwards: [00:11:30] Yep, absolutely.
Absolutely. I mean, and also one of the things we did in testing voice on construction sites was found that the existing technology was actually pretty good. Now there's a whole construction context that it didn't quite get, and we can still repair that, but we don't need to do that with the audio file.
We can do that with the text information. So it's really the transition to SMS just made a lot of sense to us and gave us this ability To, like I say, eliminate the task altogether. It's is, can we do that for you? And that's what the bots are all about and why I say, we think of ourselves as this kind of skilled labor hire business that you can rent some workers from us and we'll come and do a bunch of things.
And we started with the manpower bot. And now I'm kind of finally circling back to your original question. We do manpower, we do site orientations, we do help people get inspections done and chase down inspections and help people to get them completed. We've done a lot of... So many health surveys during the COVID pandemic.
And there was that - the pandemic has been one of those things that spurred on the development of our technology to help people solve some pretty severe problems that cropped up during that time.
We do kind of attendance who's onsite. We do even just simple onboarding into Procore. but all of these bots just do a task that otherwise would have to be done by one of the superintendents or one of the project engineers and Nyfty is there to kind of take that burden off them, do it for them and report back to them when it's done.
It's kind of like an apprentice for the superintendent.
Hugh Seaton: [00:13:17] Yeah, and I want to talk a little bit about this skilled versus not skilled. And that is that you're taking away some of the least skilled parts, but also, that's just what computers are good at, is doing the same thing over and over again. They won't forget to do it. They won't put it off. It'll get, it'll get done. When you either, when you schedule, I'm assuming you're scheduling or you can, you can order it to be done. But the thing, the broader point is it's like an assistant that never takes coffee breaks.
It can't put work in place. It can't do any of the things except for what you've asked it to do. So it's a narrow.
I mean, speak a little bit about how you've made them distinct bots instead of one big assistant. Is that a decision that you're you made on purpose or because, or is it because you're building one thing at a time?
Matt Edwards: [00:14:01] I think it's actually a very conscious decision because each person has a different pain point they want solving, and it's great for them to discover us that way. They don't have to implement a big chunk of software. You can come and say, yeah, I don't care about all of the things you do. I just need a, I just need an orientation bot that's going to just come and take care of orientations.
Because as a superintendent said to me, just the other day, I'm just so sick of people not turning up on time and me waiting around to deliver orientations. If I can get this bot to do it for me. Fantastic. You know, so, that person just wanted orientation solving and didn't care about, doing inspections or manpower or anything else. It's just, take that problem.
So we really liked that aspect of it is it kind of pick and choose the problem. So each bot works completely standalone, can work in its own right. But we do find that we get a whole bunch more automation and a much more intelligent interaction with people when we do link the bots together.
And that's something we've spent the last kind of 6, 9 months doing a lot of work on. So once we know who's on site. And we have an attendance bot to do that where somebody just scans a QR code, sends a text message to Nyfty and it remembers who they are from the first day they turned up and says, great, welcome back, Dave, in you come.
And depending on their role, do they know who they are, it can then start to automate, Hey, we've never met before. I need you to come and do an orientation or a, Hey, you're a foreman. So I need you to complete this job hazard assessment for me. And then, sometime later in the afternoon, it's going to text them again and say, Hey, I just need a manpower log from you, just let me know what you did. So yes, each individual bot needs to work in its own right. And be justifiable in its own right. But when you link them together, you really get this great configuration and great kind of intelligence and automation across your whole job site.
Hugh Seaton: [00:15:56] And I guess, you know, we're always talking about how software and the data that it creates and uses. You know, the next step is for people to begin using things together. But I think automation is probably we the sort of automate process automation you're talking about is probably one of the most obvious places where that's true. Right? Cause you're hoping that you can automate a sequence or a kind of a cluster of requirements that have to happen. But you know, they're sometimes a pain and sometimes they're a distraction. Are you finding that that's some of it is that, more than is sometimes the case for other software, the ability and the function software does that the ability to really link data from one bot to the other, is more true in this context than it might be in others.
Matt Edwards: [00:16:43] Yeah. So far we've really focused on what's happening on the job site. And if we know which foremen there, you know, every worker, but there's a whole bunch of things we can do in that particular context, we also think there's a whole bunch of other stuff that we can do outside of that particular world as well, which is maybe more project based and not necessarily the actual construction site based. So there, we can pull in different parts of information and whether that information is sourced through one of our bots or whether we source it through one of our partners' information. And so currently we are very deeply integrated into Procore, and obviously we will be adding more soon, but we, we started out , just as the, you know, in that startup sense, there's a big bunch of Procore customers. Let's just work with those guys for now and then we'll spread sideways later. So we can pull information from all sorts of different sources within that Procore application and help to automate a task.
In all sorts of different contexts, I'm sorry. I'm sounding kind of slightly hairy fairy there but, we really think that it's the difference that the our automations can make , is to really, really deeply understand the context of what's happening. Who's there, who's doing what, who needs to do what, at what point.
And if we can understand those things, we can really, really intelligently automate processes. And that's the key. So standalone is awesome. But, but linking together and pulling source information from other systems to help that automation is really where it gets very exciting and very intelligent from, if you're the superintendent, our goal for the superintendent is for you to just not have to worry about it.
Nyfty has got your back. You need that doing, it's going to get done. And we'll tell you if there's any issues along the way. And that's really the goal. We don't want to bother them.
Hugh Seaton: [00:18:37] Yeah. And again, it's the idea that the system will notice that somebody's new, and from that say, well, therefore the following things are also probably true that they need to be onboarded or they need to do a daily report later or whatever it might be.
That's really, that's really cool because again, it allows the superintendent or the foreman or the trades people to focus on putting work in place instead of dotting I's and crossing T's that are often a frustration.
Matt Edwards: [00:19:05] Absolutely. And it's a real key for us is that Nyfty looks after itself.
It automates itself. When we first built our first bot, there was quite a bit of interaction with the superintendent. Hey, who's on the site today. Who do we need to go and collect this information from? And we've moved so far away from those original early days. From interacting with superintendents, finding out what irritated them, what didn't irritate them.
And just by, by linking information that we could say, okay, we no longer need to ask you about that because we already know from this bot or from this place over here. And that's a key part of automating tasks for us.
It's not necessarily related to the intelligence of the automation, but the, the other key part that we think and why I keep banging on about us being kind of skilled labor hire business is because we, as a business, we believe that what you want us to do is do what you would have to do.
But if a superintendent is going to use our software, he shouldn't have to come to our software to find out what's going on. So in our case, we're currently integrated into Procore. We do the work in Procore, we will write the logs in Procore. We'll complete the inspections in Procore.
We'll add that person to the project directory in Procore. You'd never need to come into our software or very, very rarely need to come into our software and anything we need to get from you, we can text you and just say, Hey, just had an issue with this. Yes or no on this thing and make it really super in interactions.
So it's not another silo of data and there's two really kind of key things that drive us is not another application and not another silo of data. You know, we're just a resource that goes and does tasks and does it and logs it where you want it logging. And we think that's, that's a really crucial part of what we do.
Hugh Seaton: [00:20:51] I love that, this idea that you're not asking people to go to another place you're saying no, no, no, stay where you are and we're going to help you. And if we needed you to do something that, that environment, whether it's, Procore something else can't handle, we'll just text you and ask you a very narrow question and get help.
That's amazing. Well, let me ask you the other side of it then, and that's the human side. So you're now asking somebody. Who probably hasn't spent a lot of time thinking about AI or natural language understanding or robotic process automation, which we haven't talked about, that this is obviously certainly a Venn diagram of that kind of idea.
How are they trusting this? What do you guys go through to make people say, all right. I'm going to let my hands off the wheel and let this thing run it and trust it. And as opposed to checking on it every 15 minutes, which doesn't actually take work away, it adds, you know what I mean?
Like the hard part is, someone has someone has to say, good, I trust this and now I'm going to go focus my attention elsewhere.
Matt Edwards: [00:21:47] Yep. Yeah, it's a great question. It's no good me telling a superintendent, Hey, you just trust it. You know, we're great at this where most people hear about us is through their trade partners or through other construction vendors who who've interacted with Nyfty.
And so, so come to something like 90% of our growth comes through word of mouth. And the biggest chunk of that is a trade partner turns up on a job site and says, Hey man, why do I have to do this? Why aren't you just using Nyfty? And I can just reply to a text message and you get this thing done. Or the other part is just kind of technologist sharing it with a technologist and saying this stuff is wowing our superintendents.
So the trust comes from it being embedded. And when we see these kinds of growth patterns happening where, you know, recently we've had a bunch of growth around the Boston area, just because one of our customers is really deployed across all their projects. And then we're getting a whole bunch of other customers coming on board because other people have told them about what they're seeing and interacting with.
So that's the strongest validation we can get, really is to other people telling them about it. That is our whole kind of product led growth ethos as well. So, we don't have a sales team. We don't have much of a team we're an extraordinarily small business, doing an extraordinary amount of tasks across the U S and Canada.
I guess trying to circle back to that question of why do people trust us? It's just that they're seeing it happen. They're seeing it work and, and our whole business is based on that happening.
We don't want to invest lots of money and spend lots of money trying to persuade people to use our product. We think the product has to be so good that people will go and tell other people about it and go, man, you've got to check this thing out and if it's not that good, we don't really want to be in the business of doing it. You know, that means we've still got more work to do as product managers.
And that's why we kind of really, infuse on the product led growth strategy as a whole of business concept.
Hugh Seaton: [00:23:59] I love that. So somebody comes in and you've already got credibility because they've heard from word of mouth or they've checked with somebody one way or the other they've heard from somebody who they trust.
There's still an onboarding process. How does that tend to work.
Matt Edwards: [00:24:15] Yeah, again. Most customers will just come to our website login. They use their Procore credentials to log in. So I say from that point of view, we're a trusted partner of Procore and I guess that gives them some comfort. And then you come into our console and we have these in-app guides that'll walk you through and say, Hey, this is what we do. This iswhat Nyfty is all about. Let's give you a quick overview and then let's get a bot setup. And we have a free plan that is available for people to play and test. And we've got some people who were just using fully in production still on, on free.
So you can just experiment and play with it. Now we are always available and nearly every new customer that comes on board gets a personal email from me just saying, Hey, this is who we are. If you need any help, I'm happy to jump on a call and help you otherwise, let us know if you have any issues or need any help.
So people onboard themselves through the whole process. Unless they have questions or feedback
Hugh Seaton: [00:25:08] That's kind of cool. So in other words, it's very meta of you is our automated product has an automated onboarding...I'm kind of being cheeky there.
Matt Edwards: [00:25:18] Yeah. So that's absolutely, but that is absolutely how we see the business that we automate everything in our backend.
We are extraordinarily small as a business and at the same point we are in this business, my previous business, we had 25 employees. We've got two and a half employees in Nyfty.
Hugh Seaton: [00:25:39] Isn't modern software, amazing and obviously you've chosen something that doesn't require quite the same level of...
so I want to end with something you hinted at earlier, and that is where you see this going, both in a medium term, but also,an idea you might have about where it could go in five years.
Matt Edwards: [00:26:00] Yeah we think it's constantly coming back to that at skilled resource. So, you know what, and the skill we have is the ability to communicate with humans and to, understand vast quantities of data very quickly. So when we combine those two things, we can read everything in your project management system, and then we can communicate with the person. When we think of it from that point of view, there are just so many interactions that happen today in this industry that are human to human. And we think there are a lot of them can be automated to some degree or another. So we've started in the field and we think there's still a ton of work for us to do in the field.
We think there are so many other communications- type issues that are happening in construction that we can help to automate. And one of the things we believe very strongly in is we don't do notifications and we don't do reminders. We complete tasks, so there's a whole bunch of stuff that are in the project management sphere, you know, kind of things like submittals, RFIs, action plans and ITPs and things like that,that we think have a very heavy human involvement today. And we can help to reduce the burden on the superintendent and the project engineer who were having to manage those kinds of things and also reduce the burden on the recipient of a million notifications or a million reminders, to get concise, streamlined interactions that can help get a process done, get a task completed.
And so I'm kind of being very generic there, but there is a ton of stuff. You know, construction is a human to human business.
Hugh Seaton: [00:27:50] And in the context of ongoing labor shortages that are not going to get better, just demographically, we don't have enough people in the pipeline, so to speak for, I mean, you can look at the demographics and know that this current cohort of millennials is the largest generation that we will have, like period. We're just not having enough kids and so on. Soit's not going away. And we haven't been since the fifties. I think it is, certainly in the US and I don't think it's particularly different across a lot of the developed world.
So the idea that you're going to take mindshare and tedium and things that distract people from the hard stuff of putting work in place and surround them with little bots that just support them with process, this and process that, is I think really an attractive way of thinking about where this could go, right.
Is that, that you've essentially got a bunch of little assistants instead of one assistant that kind of tries to do too much, and also starts doing stuff you didn't want it to do, like, you can choose what it's going to help you with.
Matt Edwards: [00:28:59] Exactly and thinking in terms of the labor shortage and skill shortage, we think of two things. One is there is already, an extraordinarily skilled workforce out there and we need to get as much out of them as possible. And that means not having them do really, reductive tasks that they are far more skilled, than chasing somebody up for a manpower log, something as simple as that, you know, they're far too skilled for that kind of menial task.
The other part of it is how do we supplement or augment a junior superintendent or a junior project engineer to make sure Nyfty has got their back to just cover them on the basics. And I can't remember if I've told you this story before Hugh, but a bit when I was a project engineer in construction, every Friday night and I was a junior at this point, and it probably, to be honest, carried on through my career.
Every Friday, when I drove home from the job site, I used to have these , I now recognize were basically panic attacks that I was just, I knew I'd dropped the ball somewhere. I just didn't know where yet. And I knew sometime next week it's going to hit me. I just don't know what it is yet. And you know, it's that kind of it's Friday. The only way I could get rid of it was to have a couple of beers when I got home, which, which is a pattern that remains today, let me tell you.
But you know, that's to me, that kind of junior engineer being augumented to say, look, relax, I've got your back. I've got these basics covered. Now you do the stuff that you're meant to be there for, the actual engineering, the building of buildings, the interacting with humans in a way that they should be interacted with, not just chasing them up for stuff or just trying to get them to complete a JHA or something like that.
You know, that's I think, is how we think of it, getting more out of the really skilled guys and augmenting the less skilled guys.
Hugh Seaton: [00:30:55] And I'll end with one observation. And that is that I've heard a number of different contractors talk about how they are looking at technology also, not just for what it does on its own, but for how it helps them attract talent, where if you are, if you're a mid-sized contractor or larger, you're in a war for talent, especially people that have been there for a while or people that are, you know, younger. And people will leave some contractors that they think are not, you know, supporting them or not a little bit more kind of technically current, they don't even have to be advanced just technically current. So I would imagine that one of the things that attracts customers is that this is part of the package that they offer to people that they're trying to bring into their teams, is we are also going to support your process and your we're going to make... I doubt that they're going to give examples of panic attacks on a Friday, but most people get the idea. Most people get the idea that there's so many little things going on. There's so many little bits of paper that we have to manage.
That if something can take some of that burden away, you just have... you enjoy your job better.
Matt Edwards: [00:32:03] Absolutely. And, a thought on that note, I got a LinkedIn message about two weeks ago from a superintendent that had just joined a new company. And he was one of our very, very early customers in a previous company he worked at.
And then they'd gone off into the wilderness for a little while and he, he, he just joined a new company. And he messaged me to say, man, he was so happy to know they had Nyfty in that company. It was wonderful for me to get the message as well. You know, you know, I like those personal interactions and, you know, that's, it's exactly that kind of thing that, that company was telling people.
Yeah, we, we supplement our superintendents with this kind of technology and that was a big factor for him.
Hugh Seaton: [00:32:43] Love it. Well, Matt, thank you for being on the podcast. I've really enjoyed this conversation.
Matt Edwards: [00:32:46] Hugh it's always a pleasure chatting to you, mate. Any chance for me to waffle on about Nyfty I love.
Hugh Seaton: [00:32:54] Excellent.