Speaker 1: 0:00

Welcome to another episode of the Accelerometer. Today we have Nick with Showzone. Nick is building an AI presentation software. And Nick, why don't you tell us a little bit about your work?

Speaker 2: 0:15

Yeah, thanks, matt. So we're a startup that have existed since January. We're an Etage spin-off and we're creating an app that helps presenters deliver. So our raison d'être is to enable presenters to get their message across, and we have a software that helps with the delivery. So not necessarily with the creation of slides and content, but once the people are in the room, so in person, once they present, then that's where we come in.

Speaker 1: 0:49

How exactly do you help people with that delivery?

Speaker 2: 0:52

So we have an app that enables them, as a studio app they have on the notebook, so they can show up with their notebook as typical, add in their content quite easily, just drag and drop in their slides and they can start a show at that point which starts essentially like a live stream, starts everything up in the background automatically and we enable them on one side, to keep the audience engaged.

Speaker 2: 1:17

So typical on an audience in a presentation will kind of lose track with what's being said within 10 minutes and usually a presentation, or a class for that matter, could go on for much longer. And we try and enable this interaction back so that with the app the audience can stay engaged. What we also try and help is that after the presentation has ended, that the information that they tried to get across is prolonged, and we do that by generating a summary of everything that happened during the presentation so that the audience, when they leave, they essentially get a summary in their pocket can take that information along. Because what we've seen, and what we've seen in research, is that even after, like immediately after, a presentation, people can't remember up to 80% of the content. So if you immediately ask them what was talked about in the presentation you just watched. They can talk about maybe 20% of it, but the vast majority of it is lost. So we're trying to get that problem solved as well.

Speaker 1: 2:21

So is that your main KPI right now? The. I guess how much people remember after their presentations.

Speaker 2: 2:28

That's our KPI for us, for the presenter, is to somehow measure if their message actually got across or not. Yeah, so this would be some sort of a call to action that they can measure. So they would see OK, now 50 people were in the audience and of those 50 people, you know, 40 per people followed through with some sort of call to action or their key argument, their message. And if we can measure that and they can say, ok, using show zone, they're superior in reaching that call to action versus not having it. Then that's, you know, that's when we did our job. Well.

Speaker 1: 3:06

How much have you improved that so far?

Speaker 2: 3:09

So far it's hard to measure. We're an early stage startup, so we're still proving that part in the product market fit. From our observations so far, with a limited data set, there is an improvement definitely there, but still need to, you know, look at long and hard at the numbers. If not, then of course we'll start removing bits and pieces. So this is that product market fit area we're in right now, which is quite exciting. So we have, you know, the first stable version of the product and it's out there, it's being used, and right now we're looking and seeing. Ok, what's the feedback we're getting? How are people interacting with it in the real world, which is always surprising to see after having a product in development for quite some time. But that's the measurement we're doing there.

Speaker 1: 3:56

So you're in the arena trying things Exactly.

Speaker 2: 4:00

We have the software developed in the first version, so it's very beta, let's put it that way but it's stable and it's essentially it's good enough for a presenter to want to use it, because that's their livelihood. They have that message that they want to get across and we don't want to give them something that doesn't function at all. So you know, we reached that quality level. But now we can also observe from a far, see how it's being used, see how people are interacting with it, and that's quite interesting. Did you dog test this internally?

Speaker 1: 4:34

Are you like now a company that has just an extremely large number of meetings in order to like dog food, your product? We came out of a personal need actually.

Speaker 2: 4:43

So we're definitely, you know, dog fooding. So when we started, this was during the pandemic, so I was a lecturer and during the pandemic I'd be sitting at home in my office space, having, you know, a Zoom call, with 30, 40 students staring back at me and I'd be trying to hold a class and trying to make it interactive and I always thought, you know, it's got to be, there's got to be something better than this. And throughout the pandemic, I got better and better at it, to the point where my office became kind of half TV studio and half Ikea At the same time. I co-founder Aaron. He's or. He was a student at ETH and was on the other side of things, so he would have cases where he'd be sitting in his room all day watching these classes and the interaction we fully lost. So we had that like initial need the two of us and that's actually how we found each other.

Speaker 1: 5:44

Have you kept your post as a lecturer through building the startup?

Speaker 2: 5:50

I decided to give that part up, actually just to fully focus. So we're now three co-founders and in January we decided to commit 100%.

Speaker 2: 6:02

Congrats no, thanks a lot, no extra jobs, no side gigs, just 100% all in one, and gave us, you know, like a time span of a year plus minus to achieve what we wanted to. Was that jump scary for you guys? Somehow, for me it felt a bit less scary. I would say it varies between the co-founders also seeing like what's their personal risk, so is their family involved? So Aaron, as a student kind of already was used to like kind of the low income, low expenditure lifestyle. My other co-founder, simon, has a family so he has a bit more pressure there. But I didn't necessarily feel scared. But I also had like that support for my wife. She's working 100% and if I promised to take over all the household work then I could go on right. So for me it felt I still have that sense of like kind of security and I don't truly mind not having the money all the time because the work itself is so fun.

Speaker 1: 7:21

One VC, when I was raising money, called me a mission-driven founder. Would you say that you're a mission-driven founder?

Speaker 2: 7:29

I think so. Yes, because, as you know, when a startup like there's got to be something driving, there's got to be something there, like why go through the pain of building a product there? Are masochists out there, that's true too. I'd say more mission-driven than masochist in my case.

Speaker 1: 7:51

Yeah, at what point did you and your co-founders like look at what you're doing and decide like this right here, like let's make the jump right now, let's quit our jobs.

Speaker 2: 8:01

We started as kind of a student project, or it started as a student project, so it started without me and a few months in I joined in and shortly before, actually, my other co-founder, simon, joined in. So he's coming from the business side of things. He was working for EY previously and had the business expertise and joined into the project and we found ourselves on the basis of what that mission is, which is quite unique, I guess. So we didn't know each other beforehand. How would you distill that mission down? The mission is to enable presenters to get their message across.

Speaker 1: 8:41

To enable presenters to get their message across.

Speaker 2: 8:44

Yes, so if you think about a classroom, then structure wants to educate and there's that content that needs to get into the heads of the audience. Yeah, and we want to optimize that communication form and make it better. Attaching a vision to the end of the mission would be, in a few years time, the next person that has that great idea out there. They have that great new thing they want to express. If they express it through show zone, then we'll ensure that their idea gets out there and gets heard. Couldn't this be a?

Speaker 1: 9:17

dangerous mission. Are you worried that you could enable the wrong person to get their message across? I haven't thought of it that way.

Speaker 2: 9:26

I was always thinking positively like yeah, that must be a good mission. The positivity of that message has never been questioned so far.

Speaker 1: 9:35

When I started actually I was emailing people trying to get people to use their product and says on our website our mission is to democratize AGI. And somebody messaged me and they were like how could you say something like that? Like that is such an evil mission. I encourage you to rethink your values. And this was I'm not going to name it, but this was the CEO of a large company and that was the point where I knew I was doing something right. I agree.

Speaker 2: 10:00

I worked at it, started before and the founders there were kind of my heroes, and one of them did say, when stuff started to happen, right, you know, you get kind of that shit storm on Twitter, right, the naysayers yeah, to me it was like, oh, did something wrong here, and but he was saying, if that wouldn't happen, we wouldn't be doing something of value. So looking at it that way, you know, it's an indicator that you're doing something that people care about.

Speaker 1: 10:30

Did you ever have like a moment like that in your startup journey where you just had like an extraordinary amount of self doubt about what you were building?

Speaker 2: 10:37

I'm more optimistic there, but within our team we have the different projects that we're doing. In our team we have the different personalities and I wouldn't call itself doubt, but it would be more like criticism. Yeah, it'd be like highly critical, like, is this the thing we're building? And that criticism is often brought up, which is healthy and we kind of need to discuss it through. I'm generally feeling quite positive of what we're building.

Speaker 2: 11:04

Like I still, you know, every morning get up and think you know, let's do it, yeah, and yeah, just go out and program the thing and build the thing. So yeah, so far, so good. Is this your first startup? It's my first startup of my own. So I've been previously with the Etage spin off catcher guide, which is a travel company, very highly successful, and I've been along with their journey as an early employee For about six years to the point where they became a unicorn, which was that incredible fast growth journey. And that's where I kind of learned to step out of the comfort zone, just get, to try to get that mentality of building something, putting it out there, seeing if it works, tear it down, build it up all that iteration.

Speaker 2: 12:01

And that's been incredibly helpful. Now, with my own startup, is to just accept that those facts. But it's a very different thing to be in a growth startup versus, you know, a zero to one startup.

Speaker 1: 12:19

What's been the most challenging thing for you about having your own startup?

Speaker 2: 12:23

I would say resources. We're a software company and to build a software of value and not just a small tool, right, is something that needs to provide a value, have a functionality, need to be thought through, have a good user experience, and all this it takes time and we're two engineers in a three-person startup and that pace is too slow and that's the key challenge as a software startup. So what we expect is kind of like that plateau where we hit version 1.0 of the software, getting it out there, paid customers being able to really like, dig in with the marketing and scale, which is different to, you know, another startup. You know, had we been selling cookies, we could have had the first cookie customers within days, right, so, but not for us. So that's that's for me, that key challenge of like that long stretch as a software startup to get to something of value that customers are willing to pay for, just with that limited amount of resources available and time, of course, how do you, how do you stay motivated?

Speaker 1: 13:33

I mean, I guess you're very optimistic, but how does your team stay motivated?

Speaker 2: 13:37

We're as a team. We gel quite well, surprisingly well. I could have interviewed, like you know, 100 people and maybe not have found that same team that just somehow works and fits together. So it's like all personalities kind of extend each other and we just work, and so there's there's no personal drama, there's, you know, there's real talk, there's real discussion and I think that works well to the stay motivated. So the answer to your question would be just pure luck at this point.

Speaker 1: 14:15

How do you think about company culture To?

Speaker 2: 14:17

me, culture is the basis and it's highly important. So would I look for a job now? Would I go out there and look for a job? The number one thing for me would be to evaluate if it's a good company, culture or not. Over wage, over location, over, possibly, the work itself, I put it at number one. So it's kind of that basis that everything else is built upon.

Speaker 1: 14:53

So how did you, I guess, first get introduced to generative AI?

Speaker 2: 14:57

Through the outside world, but also my co-founder, so the initial student project we joined had two data scientists there, which with a deep passion for AI and joining there, that's kind of my first touch point. Interesting thing there, though, is that I didn't join because of the tech. I joined because of kind of that problem or that need or that itch worth scratching. Nonetheless, ai opened up the entire new possibilities, so it's kind of like connecting the two, but I would say, for me, it really got into the nitty-gritty early last year.

Speaker 1: 15:42

That's really cool. What was it like for you first seeing AI?

Speaker 2: 15:47

Amazing. So of course there's always that chat, gpt, initial experience that was awesome, but then also building the product and using it, and then sometimes it goes weird and then you see it not working. But then it's hard to wrap the head around. I'm using this AI and something's not working and it's just going off track somehow.

Speaker 1: 16:18

And then you start hallucinations In a transcription.

Speaker 2: 16:21

We've had cases where I would just start repeating a word infinitely and I'm like, oh okay, gotta fix that. Or one case where I was developing and just kind of talking into the microphone and then when I finished my test presentation I got a summary back that I thought was hardcoded by my colleague and but then I ran it again and it wasn't there anymore. So that, like the thing it gave me back based on my kind of my testing talking into it, was so good I thought it was like a hard-coded text it was giving back. So that was kind of an amazing, amazing experience there.

Speaker 1: 17:04

Using AI, do you ever worry that the hallucinations are going to cause problems for your clients? Not just in that you know it's not giving back the right result, but it will hallucinate in the wrong way.

Speaker 2: 17:18

Yes on some occasions. So a transcription like modern transcription, like a whisper, is fantastic right, where really high probability of getting the right words. We also have the benefit of having some context, where you know there's slides, there's text on the slides, there's all of this and then there's context of a class or a presentation, so it gets it quite right. Nonetheless, people are afraid of presenter that it could be transcribed wrong and some output could be wrong, and then they start mistrusting it a bit because they're putting their professional career in a way, kind of on the line right. So to these business professionals, it's a key way of communicating and it has to work and if there's something wrong in there then they'd make or break the product right, where people say no, I definitely want to turn off a transcription because of, like, those few words that could be just misinterpreted.

Speaker 2: 18:23

So, as a product that we can work towards it and have, for instance, that we generate this summary of everything that happened during a presentation but it needs to be confirmed by the speaker before the audience sees it. So the transcription is there, kind of as a subtitle, and it's there and then it's gone. So it's not there forever. It's just to help people, possibly with a disability, but then the summary needs to be specifically confirmed, so these like small corrections can still be made at that point. So this is kind of like we can build some safeguards kind of into the product and checks. So Microsoft always talks about like in that co-pilot, and I agree with that concept, so you can't fully hand off everything to the others. It needs to be like a check and one is still personally the pilot, but one has that like great co-pilot along the way and I like that notion and you can build a product kind of around that notion too.

Speaker 1: 19:24

So, other than the correctness of the product, what have been some other challenges when using AI in your product?

Speaker 2: 19:33

Compute resources, also, in our case, real time. So this is the software is running in real time or in a presentation occurs, so the show is started and we're picking up an audio signal, sending it to our server. It's transcribed in real time and there's, like, those compute resources necessary. So we need to spin up those resources so we don't have a pool of heavy duty machines just waiting there to be used. So we need to start them up and tear them down again. So all of that setting up, tearing it down, so the compute resources part and the real time aspect of it.

Speaker 1: 20:15

So what's some AI technology that has been particularly exciting for you that's come out in the last few months.

Speaker 2: 20:23

It's, I would say, the advancements and on that front, specifically like with the transcription, so you know, we found out that it works incredibly well, not just picking up, you know, like an English language, but picking up any language and then possibly also translating it in real time. So you know, having some flow of data coming in but then outputting it into a different format entirely that the AI is extremely good at that and that's that's actually quite exciting. And, of course, possibilities with generative AI. We're not doing anything with images.

Speaker 2: 21:01

So far, we didn't see the need for that in our product, but in terms of summarization and all that, what I'd like to see more is to be able to provide context easier. So you have a trained model, yeah, but we have much more data. We have content from slides. So maybe you know there's a person's name on there, or maybe it's a presentation about mathematics and we can say, yeah, this is the context we have and we kind of like got it out of the presentation, or you know, these are the questions that were asked live by the audience. To be able to provide that back to the AI in an easier way, yeah, I'd love to see that.

Speaker 1: 21:48

Have you considered training your own models to do something like that?

Speaker 2: 21:52

The problem there is that so if we're a general product for presentations and it's happening real time, yeah. I don't think we could train the model that quickly. Right, it would need to be set up as a computation.

Speaker 1: 22:10

But train the model like a model ahead of time to be good at this sort of context management.

Speaker 2: 22:17

That's a great point. I probably passed that one on to my co-founder, aaron, which is really the go-to person on that front.

Speaker 1: 22:28

I feel like you should be scared to about to waste a billion dollars. It's a very expensive thing training.

Speaker 2: 22:38

Yeah, that's kind of that thing. So you know, a transcription and generating text content is manageable, but then training a model on the fly, that would be definitely like a future that I want to see. Can that become fast enough and easy enough to use? That would be amazing and it would also make a true difference in what we're aiming towards.

Speaker 1: 23:06

What's something that excites you about the future of AI?

Speaker 2: 23:09

I'm very positive about all the new possibilities. So a few days prior to this recording, we were at AIX Summit, which was a big conference here in Zurich hosted by the AI Center, and just the vibe and the excitement that conference was something I haven't seen in quite a while and just that whole innovation of like moving forward and positivity and people visiting each other and saying like, oh, you're doing this and that and that's great and you know, not necessarily holding back, kind of like that close source aspect, but you know, out there and trying to build something new. So that's very exciting. There's like that pace, the possibilities and for me specifically, you know we're building a startup based on this. Is this client, this product we want to get in front of people is what are all those innovations where we start connecting the technology we have at hand to all different aspects of our life. So there's, you know the important. You know like medical research and all of this. You know like the important topics, you know to reduction all this, which definitely that the New technical capabilities have huge potential there, but also in smaller products.

Speaker 2: 24:37

Like we are, we're not solving the big world's problems, we're solving a business problem. I'd like to see more and more products truly use AI in the correct way. There, of course, I think the technology is moving faster than the consumer, especially in if you think about where we are with presentations. So a PowerPoint took years and years and years to get people off of the overhead projector, and not just that pace. Maybe you see it yourself in academia. You still have the old school people doing things the really old way, and their acceptance for doing something new with an AI is likely not there. It does need to be right, there's enough customers to go around, but I think that pace is much slower than the technology.

Speaker 1: 25:33

What was it like just building a startup in Zurich?

Speaker 2: 25:37

It was actually quite good so far, Because Zurich has that innovation hub and also has a lot of people that genuinely want to help. So it's not people saying like, yeah, I'll be your mentor, whatever, and then move on, but we've had people helping us out that they want to see us succeed. The difference offers and startup space and office space and all that is super easy. Almost the opposite. Where people want us to come to their office and we say no, no, thanks, we're already in the other office, so we're the prettiest on the block, so to speak, and they want to entice us to join. But on the flip side, what I've seen in terms of investment is not on par with other places, and I'd like to see that happen.

Speaker 1: 26:37

When I was speaking to Swiss VCs one of the early experiences literally like we'd been building the company for maybe a month and a half. We had this demo product out and we went out to raise money because we were dumb and young and we spoke to a Swiss VC that knew that we'd only been around for about a month and a half and we're in the fifth talks with them because I think they, like everybody else, were mostly interested in us because of our backgrounds. But five talks in the full team was there sitting there, meeting ended and they were like we really like you guys who want to invest. But for the next steps, we need you to come up with your five-year financial projections. Our company's been around for a month. You expect us to know how much money we'll be making and how much money we'll need in five years.

Speaker 1: 27:30

I have sat there on the other side of the table and looked at the startups to invest in myself. I was curious about how to do it and that is not like you've been there. You've been building a startup. You know that anything at a one-month startup tells you is utter made-up nonsense. You don't care about that. You care about who the founders are If the idea is good, if they have traction or stuff like that, not about their projections.

Speaker 2: 27:56

Yeah, we've been looking at it too. It's always funny From my perspective. It's like yeah, we got the team, we got the product Traction. Okay, you can discuss that as a sort of what traction means for us, right.

Speaker 1: 28:10

Let's reinterpret traction.

Speaker 2: 28:13

And so when we get criticism, I look at it as like, yeah, I just couldn't convince them what I believe on my side, which is an interesting way to look at it, even in terms of the team.

Speaker 2: 28:28

I'm fully invested in the team, I know the team is awesome and if I couldn't convey that, it's just like me not being able to bring that to them in a way that they can believe it. Yeah, the traction part, specifically for a software product that has any sort of complexity to it, you know that's beyond like a simple tool, and that can't be a simple tool like fast enough. That's kind of like a conundrum there, where on the one side, you need to, like prove that there's traction in some shape or form, but at the same time, you need to spend the time to build a product that can be sold or used in some way, and that for me personally is a bit of a struggle, like where's that middle ground? Because investors will come and say, well, you know that's not enough traction. Or you know, like, come back when you're, you know, making this amount of money already.

Speaker 1: 29:29

The product is supposed to build itself right. You get enough. You get enough users and the users build the product for you.

Speaker 2: 29:37

Yeah, yeah, I mean in hindsight. So if I look at successful software companies of recent years, also with Y Combinator, you know like the Twitch, or you know that's just the pain they went through in the like the years of building something until it finally like clicked into place and took off they stuck with it and that's commendable.

Speaker 1: 30:06

You know, like with a lot of the product.

Speaker 2: 30:08

You know like you got the Figma or Zoom as well. You know like they didn't just, you know, start one month in and be like this is Zoom now and here's the investment. They you know you need to build it up to reach something first, and so they're kind of my heroes in that respect, right, like they did like this march of death through the desert of software development and came out the other side and kind of survived and built something awesome and built something great and made money with it, you know, and solve the problem.

Speaker 1: 30:40

So thank you very much for coming on the accelerometer. Thank you, matt, it was really great having you. Thanks a lot.