How Big Tech's AI Integration Could End the Startup Boom

00:00:00: - I got a theory because with all those AI services and models, there have been some opinions,

00:00:08: - some voices who basically said that it's a bit like the internet, a new platform on

00:00:13: - which a bunch of new business ideas can come up and grow so that people could use AI and

00:00:21: - those AI models to create new business models, startups, and therefore, well, challenge the

00:00:31: - existing companies and business models, just as it happened with the internet, which also

00:00:36: - gave us a brand new platform to build up on.

00:00:39: - And I don't think that's the case.

00:00:41: - I think that many of those AI services that were created over the last year, which essentially

00:00:51: - all or most of them use open AIs, APIs, the ChatGPT API under the hood, that most of them

00:00:59: - might not survive the next couple of years.

00:01:04: - And I think that this will happen because all these services, all these business ideas,

00:01:14: - for example, a website that allows users to upload a PDF document to automatically summarize

00:01:20: - it, I think that they will be replaced by the products that already exist for a certain

00:01:28: - job by Adobe Acrobat, for example, if we stick to the PDF example, because I think that the

00:01:35: - creators of those established products will simply integrate AI into their products and

00:01:42: - integrate those features that are solved by those extra AI services into their products,

00:01:50: - therefore making many of those AI services redundant.

00:01:54: - That's what I think.

00:01:56: - So you would say that although AI looks like this great chance for small new companies

00:02:03: - to kind of challenge the big ones, the big ones will take the cake in the end because

00:02:11: - it doesn't make sense from a user perspective.

00:02:14: - Well, it does make sense to try these things out, as you said, uploading a PDF to a specific

00:02:19: - site that provides that service.

00:02:21: - But in the end, in our daily workflow, we want to have our products that we use.

00:02:28: - As you said, we take the Creative Cloud by Adobe, for example, which comes with Photoshop

00:02:32: - and so on.

00:02:33: - So you want to have an integrated workflow and because of that, these small startups

00:02:38: - don't have a chance.

00:02:39: - Yeah.

00:02:40: - If I think about it, I would agree at first sight because personally, I see myself not

00:02:47: - using ChatGPT that much anymore because it's not intuitive for me.

00:02:51: - I use the tools I'm used to.

00:02:54: - What we, for example, do use though is or are extensions for Premiere Pro, for example,

00:03:01: - which add features to Premiere Pro but into this existing product.

00:03:07: - In this case, though, we use a third party extension.

00:03:11: - So here in this case, we actually have a situation where the startup wins at the moment.

00:03:17: - Yeah.

00:03:18: - I think that's only a matter of time in many use cases until, again, to stick with the

00:03:24: - Adobe example because we use it a lot for the various tools they offer, until they implement

00:03:29: - those key features into their core products.

00:03:32: - Because the plugin you mentioned, for example, helps us with generating captions and so on.

00:03:37: - FireCut, by the way.

00:03:38: - FireCut.

00:03:39: - Yeah.

00:03:39: - It's a really great plugin.

00:03:41: - We're using it because it, for example, helps with creating captions and it does that at

00:03:47: - least for us at this point of time better than the built-in caption creation mechanism

00:03:54: - that is already offered by Adobe Premiere Pro.

00:03:57: - They do have a feature to create captions but those do not have the same quality as

00:04:01: - those of the plugin.

00:04:03: - But I think this will change and the same will happen, I think, for many other features

00:04:09: - that can be driven by AI.

00:04:14: - At the moment, we still have some services that are using open AIs, APIs, or some of

00:04:19: - those publicly available models to deliver real value and benefits but I think it's only

00:04:26: - a matter of time until the big players, the big companies will catch up if they haven't

00:04:31: - already and simply integrate those solutions into their products and people are convenient.

00:04:36: - Why would I use some third-party service for which I have to pay to summarize a PDF document

00:04:43: - if I can do it from right inside my PDF reader?

00:04:47: - Why would I do that?

00:04:48: - That's true but I see a point where these smaller startups still have a great chance

00:04:52: - because in contrast to the big companies, which tend to be slow and not very innovative,

00:04:58: - the smaller companies have the drive and the energy to create new products.

00:05:04: - Let's name Firecard once again.

00:05:05: - We don't get paid for that.

00:05:06: - We just like the tool.

00:05:07: - It is very easy to use, right?

00:05:10: - It's so simple and what Premiere Pro isn't is simple.

00:05:13: - It's a complex tool with, I guess for many of us, way too many features.

00:05:17: - I don't even understand 10% of its features probably after editing videos for seven years.

00:05:22: - Still, I think that these small companies still have the chance to create these awesome

00:05:29: - AI products because what happens in the end is Adobe, for example, will think, okay, we

00:05:35: - can create this on our own or we just buy it.

00:05:39: - We just integrate the existing product somebody else created, buy it and integrate it.

00:05:45: - So of course, the person who created that plugin or that extension doesn't work on this

00:05:50: - product anymore but still was able to create something very helpful and get some money

00:05:55: - in the end.

00:05:55: - So I think that's maybe something I could imagine.

00:05:58: - Still, I think that the innovation coming from these small companies is very important

00:06:04: - in the AI segment now.

00:06:05: - I agree that this might be an exit opportunity for some of those companies and therefore

00:06:12: - for the founders of those companies, it's great but of course, the end result is the

00:06:16: - same that we go back to having the big companies who have all those features integrated no

00:06:24: - matter if they bought it or developed them themselves.

00:06:28: - The end result is the same and regarding the speed of innovation, of course, small company

00:06:33: - startups are quicker but at the same time, if we take a look at Microsoft, which for

00:06:39: - example rolled out a new co-pilot almost every month last year, not all publicly available

00:06:47: - yet but they have, I guess, the Microsoft co-pilot.

00:06:50: - They have co-pilots or AI features integrated into Microsoft Excel and then Word and for

00:06:57: - us developers, they have GitHub co-pilot.

00:06:59: - So they have all these extra tools and they are really pushing forward aggressively with

00:07:05: - integrating AI into all their products and Adobe, for example, is doing that as well.

00:07:10: - They have their own AI solutions already.

00:07:13: - They're just not that great in all scenarios yet but I think that'll be a matter of time

00:07:19: - so it might not even happen that they need to buy a lot of startups but of course, it

00:07:25: - could.

00:07:26: - But in the end, I don't think we'll see a lot of big new players with brand new business

00:07:31: - ideas because I guess my main point is AI might not open up that many new business ideas

00:07:40: - as the internet did.

00:07:41: - Instead, it opens up some new business improvements that can be reaped by those companies who

00:07:50: - run the main business that can be improved.

00:07:52: - That's I guess my main point here.

00:07:55: - I see the point though, if we talk about Microsoft, do you remember last November when I tried

00:08:00: - to find out how this co-pilot thing can be helpful for us?

00:08:04: - I spent two hours, I guess, just to find out that it's not available for normal companies

00:08:09: - or normal people.

00:08:10: - The general Microsoft co-pilot.

00:08:12: - Yeah, it was like pain to read through 300 documents, blog posts, whatever.

00:08:17: - So my point is that I agree that the big companies might take the cake but I still think it takes

00:08:24: - more time because they always try to provide the big package but smaller companies can

00:08:31: - offer those smaller solutions in the end, which are sometimes more helpful for people

00:08:38: - like us, for example.

00:08:40: - Like you mentioned, the capture thing or this multi-cam editing thing, if we edit this podcast,

00:08:45: - for example, we don't have to edit the two perspectives on our own.

00:08:49: - This is done via AI in the meanwhile.

00:08:52: - And these are small things that at least I want to have in small and simple and fast

00:08:57: - tools and not in this big overall tool that comes with 10,000s of features.

00:09:03: - So I see that niche, but I also see your point to be honest.

00:09:07: - So again, the question is basically, does AI help the majority of people in the end,

00:09:12: - right?

00:09:12: - That's your point, I guess.

00:09:13: - So if we talk about the business perspective, it takes jobs, but let's say it this way,

00:09:19: - it takes more jobs than it offers opportunities for new jobs to be created, or how do you

00:09:24: - see that?

00:09:25: - Yeah, that's not the main thing I meant.

00:09:26: - I guess that's a different question regarding the impact it has on all the people being

00:09:32: - employed and various jobs.

00:09:34: - I was really just looking at the business idea and business model thing here because

00:09:40: - it might look like AI opens up a lot of new potential business models.

00:09:46: - And I guess my main point is that if you take a closer look, a lot of those new business

00:09:51: - models are actually not new, but instead just improved versions of existing models, and

00:09:55: - therefore the benefits will be reaped by the companies that have those existing models

00:10:01: - unless they totally sleep and they fail to integrate AI.

00:10:06: - In that case, some AI-based startup might come up and might actually build the rest

00:10:12: - of the product around their AI core and then deliver a better product experience maybe

00:10:17: - than the established player.

00:10:20: - But whilst that might happen in certain cases or scenarios, I don't think that it will happen

00:10:26: - a lot.

00:10:27: - And therefore I think that a lot of those AI services might not survive because they

00:10:33: - don't have a unique selling point in the end.

00:10:37: - But isn't then the implication that although AI is very big and offers many opportunities

00:10:44: - to make things easier or better, it is quite limited in the way it can provide additional

00:10:51: - value?

00:10:52: - Let me tell you what I mean here.

00:10:53: - If we think about YouTube, right?

00:10:56: - YouTube, who owns YouTube?

00:10:57: - Google.

00:10:58: - So Google could just say, we create all videos on our own because it's our platform, we are

00:11:03: - the big player.

00:11:04: - That's simply not possible because the content on YouTube lives from its creators because

00:11:09: - there are an endless amount of topics that can be covered.

00:11:13: - And a big company can't just cover this on its own.

00:11:16: - It's simply not possible.

00:11:17: - If we say that AI can add hundreds of thousands of features to Adobe, for example, to Premiere

00:11:25: - Pro, then it would be hard for Adobe to implement all these features on its own, right?

00:11:30: - Because there are limitations regarding the workforce, for example.

00:11:34: - So you would need other companies to step by step at this, this and that feature.

00:11:38: - Actually AI, if you think about the Adobe example again, has a limited amount of features

00:11:43: - that can be added.

00:11:45: - Isn't that the point?

00:11:46: - That you say, how many features can be added to make video editing easier?

00:11:51: - I think these are not hundreds of features.

00:11:54: - Am I right?

00:11:54: - Yeah.

00:11:55: - Yeah.

00:11:56: - I would say so because of course, yeah, YouTube is probably not going to start generating

00:12:03: - videos with AI in a couple of years, even if AI might be capable of generating videos.

00:12:09: - It already is with Sora and in a couple of years, it might be able to generate longer

00:12:13: - videos.

00:12:14: - But even then, YouTube is probably not going to do that because as you say, they, they

00:12:19: - earn money because they have all these creators, these creative people who generate anything

00:12:25: - from documentaries to pure entertainment to tutorials.

00:12:30: - And that's all fine.

00:12:31: - But that's also not a feature of YouTube.

00:12:34: - The videos are not a few feature.

00:12:36: - They are like the core product in the end that's wrapped by YouTube or presented by

00:12:41: - YouTube, you could say.

00:12:43: - But in case of like tools like Premiere Pro, as you said, I think we're only talking about

00:12:50: - a couple of features, maybe a hundred if you would say like this, but I don't know the

00:12:56: - exact number.

00:12:57: - It doesn't matter.

00:12:58: - There are a fair amount of features that could be enhanced by AI, but there are some features

00:13:03: - that could really be enhanced tremendously, like helping with automatically removing errors

00:13:09: - in the footage, multi-cam editing, captions, as we already said, potentially also things

00:13:16: - like removing the background in a video, even if there's no green screen.

00:13:20: - These are all things that can be done and that probably will be done.

00:13:24: - And here, of course, I want to have that integrated in Premiere Pro because then I

00:13:30: - don't have to leave the tool.

00:13:31: - I don't have to pay for some extra third-party service.

00:13:34: - I can do it from right inside there.

00:13:37: - And that's why I think that this is the approach or the solution that will win for many products

00:13:45: - in the end.

00:13:45: - So basically you say that AI cannot be compared to the internet because the internet was something

00:13:53: - new that offered endless opportunities and options, whereas AI just changes the internet,

00:14:02: - so to say.

00:14:02: - Is that kind of the point?

00:14:04: - Okay, I got you.

00:14:05: - I think kind of.

00:14:06: - There might be and there definitely will be certain new business models that are now possible

00:14:13: - with AI that weren't possible before that are not just the improvement of an existing

00:14:21: - product.

00:14:22: - So think about things like planning furniture for your flat or something like that, where

00:14:30: - you kind of want to take some photos of your apartment and you want to upload that somewhere.

00:14:35: - And then you maybe want to take some photos of furniture you're thinking about putting

00:14:40: - there.

00:14:40: - And then maybe some AI model generates a video where you can walk through your flat with

00:14:46: - that furniture in there so that you get an idea of how it looks.

00:14:50: - Maybe that is something new, though even here you could argue that this could maybe

00:14:54: - be offered by the companies which sell furniture as an extra service to make that easier for

00:14:59: - you.

00:14:59: - But there might be certain business models that can arise from AI that simply didn't

00:15:07: - exist before because the AI part is their main feature.

00:15:11: - But I don't think there are too many of those.

00:15:14: - Basically AI is limited, if we think about it this way, to do boring tasks for you.

00:15:21: - For example, creating captions which you could do on your own if you want to, like this multi

00:15:25: - cam thing, for example, or to bring your imagination into reality.

00:15:31: - Furniture example, if we once again take mid-journey, these are all things that you have in your

00:15:36: - mind that you want to see on the screen in a video or an image.

00:15:40: - Actually that's it, videos or images.

00:15:43: - And the features related to that are quite a few and can be implemented into quite many

00:15:50: - areas.

00:15:51: - At IKEA, for example, they can implement that.

00:15:54: - But still at the moment, and here I agree with you, I don't see any additional revolutionary

00:16:02: - ways to create new businesses with that AI concept, right?

00:16:08: - Yeah.

00:16:08: - That's a good point.

00:16:10: - Maybe if we would see them, we would simply build them and not talk about them.

00:16:14: - And I do think, as I said, that there are opportunities, there always are.

00:16:20: - I'm just not sure if it's that big of a revolution as the internet was from that business model

00:16:26: - startup perspective.

00:16:28: - It definitely is a huge revolution, it changes a lot, but I'm not sure if it opens up a lot

00:16:33: - of brand new business ideas or opportunities for new companies.

00:16:38: - And we will see basically, because what AI definitely does when it's implemented and

00:16:42: - used correctly, it speeds up things.

00:16:45: - So we can do more work once again.

00:16:49: - Exactly.

00:16:49: - But it speeds up things we're already doing and we're already doing things with tools

00:16:53: - that we're already using.

00:16:55: - So in the end, it speeds up those tools and therefore it can be implemented in those tools.

00:16:59: - That's I guess kind of the main point I was trying to make here, that it's exactly doing

00:17:05: - that, as you said, it speeds up things and therefore it's those things, the tools we're

00:17:10: - already using, that need to integrate AI and that probably will integrate AI and that will

00:17:16: - therefore win.

00:17:17: - Because of course, what's quicker?

00:17:19: - If I have to use some extra tool to generate captions, for example, or I use the tool I'm

00:17:25: - already using for editing the video and I also generate captions with it.

00:17:29: - I mean, let's be honest, we use mid journey a lot to create images, thumbnails and so

00:17:33: - on.

00:17:34: - If you have a thumbnail for a YouTube video and you need an icon, for example, then you

00:17:38: - have to go to Discord, create that image, create that icon in mid journey.

00:17:42: - That's not a logical process.

00:17:44: - It would be better to do this directly in Photoshop, in your project.

00:17:48: - You get back the image, you paste it and they're done.

00:17:50: - So I totally agree.

00:17:52: - As I said, I still believe there are still chances in the next months or years for smaller

00:17:57: - companies which come with innovative features to bring these to the market before the big

00:18:02: - ones got it.

00:18:03: - But if it's a good feature, chances are that the big one also implements it soon or you

00:18:08: - get simply bought.

00:18:09: - Yeah.

00:18:09: - And there is also another interesting twist here and that is, if we take a step back,

00:18:18: - it's one thing to have all these companies that build up on top of the APIs offered by

00:18:24: - OpenAI, which might be bought or which might be replaced by established players.

00:18:31: - That's one thing.

00:18:32: - But what about OpenAI itself?

00:18:35: - I think it's important to keep in mind that OpenAI doesn't have a lot of products.

00:18:43: - They have really amazing AI models and they have chat GPT.

00:18:48: - But I think if OpenAI does not stay as innovative as it is and maybe other players and companies

00:18:58: - like Google catch up as they already might be doing with Google Gemini and their latest

00:19:04: - AI models, even OpenAI might become less relevant in some future.

00:19:10: - I think they're still far ahead in many areas.

00:19:13: - But if we think about it, Midjourney is really good for images, in my opinion, often better

00:19:18: - than Dolly.

00:19:20: - Definitely.

00:19:20: - Google Gemini looks really promising, might be better than chat GPT.

00:19:24: - And guess what?

00:19:25: - Most people are already using Google for searching.

00:19:28: - So if some chatbot is integrated there, that might be a thing.

00:19:32: - Or most people have a Google account.

00:19:34: - So if Google finds a way of upselling them to their chatbot, a lot of people might do

00:19:40: - that.

00:19:41: - Whereas chat GPT is something you have to sign up for.

00:19:45: - It's a different company.

00:19:46: - And well, with text to video, we might see something similar that in some years maybe

00:19:52: - we get some other companies, established companies like Google or some other company who's doing

00:19:58: - really well there.

00:19:59: - And therefore, I think that the irony is that OpenAI started that all and they're still

00:20:06: - at the front of that entire development and they're still pushing ahead and really impressive.

00:20:11: - And I think GPT-5 will be super impressive as well and all that.

00:20:16: - But ultimately, even they could theoretically be replaced if the other companies catch up

00:20:24: - and deliver great experiences or maybe even just good enough experiences integrated into

00:20:31: - their products.

00:20:32: - Yeah.

00:20:32: - Because OpenAI has the feature or the product in general, so the AI models, but doesn't

00:20:40: - have the platform to market it in the end, right?

00:20:43: - Yeah.

00:20:43: - They have chat GPT.

00:20:44: - And of course, lots of people are using that.

00:20:47: - It was the fastest growing service in history.

00:20:50: - But I don't think chat GPT is a good platform if you compare it to a Google product, to

00:20:54: - be honest, from the usability part.

00:20:56: - I don't think it's nice to use just from my personal thing.

00:20:59: - I would prefer, as you said, a Google integrated flow that allows me to write prompts or something

00:21:05: - like that and get back these answers concisely integrated into my normal Google browsing

00:21:11: - workflow.

00:21:11: - I would definitely use that, but chat GPT, I don't know.

00:21:15: - So I agree what you say.

00:21:17: - It can become difficult for OpenAI because they present the features, others copy that

00:21:24: - features and might even improve these features and have the people locked in into their platforms,

00:21:30: - into their services, so that the people then use their AI product they create.

00:21:34: - And I mean, Google showcased, I think, at their last I.O. what their vision for search

00:21:40: - is, that search is becoming more of a thing where you can ask real questions and get a

00:21:47: - complex answers because it's using AI under the hood.

00:21:50: - And I think it's easy to imagine how Google could integrate AI into all its various search

00:21:55: - products and Google Maps, but of course, mainly also into the web search and therefore replace

00:22:01: - the need for chat GPT.

00:22:03: - Because if I can go to Google and ask it, hey, give me the code for a tic-tac-toe game.

00:22:10: - Well, why would I go to chat GPT if Google gives me a good answer here?

00:22:14: - It sounds so kind of strange because using chat GPT just means opening or browsing a

00:22:19: - website, but we are all comfortable people, right?

00:22:23: - We want to have this in a seamlessly integrated workflow.

00:22:26: - So as I also said, if you Google something and if you can instead get back the code for

00:22:30: - a game, you want to stay in that environment.

00:22:33: - So yeah.

00:22:34: - Yeah.

00:22:34: - I might not even Google it in the future.

00:22:36: - I might also just use like GitHub Copilot, which is already integrated into my IDE and

00:22:44: - offers a chat functionality already, which does use the GPT APIs by OpenAI under the

00:22:50: - hood, but that could change in the future or even if it does, of course, OpenAI will

00:22:56: - then still earn money because their APIs get used under the hood, but I'm not visiting

00:23:01: - their site.

00:23:02: - I use GitHub Copilot in my IDE and if Microsoft decides to end their partnership with OpenAI

00:23:11: - because they got their own AI models, which are similar in power or if they buy OpenAI

00:23:17: - simply, then of course the outcome would be great for OpenAI, but ultimately I, the end

00:23:24: - user, I don't care about the model or the company who's used under the hood.

00:23:30: - I care about my user experience and well, I think we're still far away from OpenAI becoming

00:23:38: - redundant.

00:23:39: - It's just an ironic thing here that the company that started it all could be the company which

00:23:46: - in the end maybe doesn't profit the most from that entire AI boom.

00:23:51: - And this kind of ironically ties back to the beginning of this podcast where we said that

00:23:56: - or you said that small companies that might have opportunities related to AI, to generate

00:24:05: - products will not have a big chance because the big companies will basically take the

00:24:10: - cake.

00:24:11: - Yeah.

00:24:11: - Same thing could happen to OpenAI, which now is a big company, but actually was a small

00:24:16: - company that became big.

00:24:18: - But in the end, the big ones just take their features, their innovation as motivation and

00:24:25: - as driver for their own AI products they create and in the end, OpenAI might become maybe

00:24:31: - not obsolete, but might not be the big winner.

00:24:35: - The big winner.

00:24:35: - Yeah.

00:24:36: - So that's the long term.

00:24:37: - Yeah.

00:24:38: - I mean, nobody knows, right?

00:24:39: - Also, we don't know, but it's a very good point because at the moment, OpenAI seems

00:24:44: - to win it all.

00:24:45: - As you said, we have Gemini from Google, so we have other services that are also getting

00:24:49: - bigger and bigger.

00:24:50: - But generally, I think for the majority of people, AI is OpenAI at the moment.

00:24:57: - But it might be different in the near future.

00:25:00: - Yeah.

00:25:01: - It's hard to tell.

00:25:02: - And I guess the company who also seems to win is NVIDIA, though that takes us down a

00:25:08: - totally different podcast episode probably.

00:25:12: - Talking about the stock market also, I guess.

00:25:15: - Yeah.

00:25:16: - Yeah.

00:25:16: - Because also there you could, of course, argue that maybe those other companies start making

00:25:20: - their own chips and so on, but it's a different story.

00:25:23: - But I guess you can dive deeper and deeper.

00:25:26: - And you can.

00:25:27: - And the difference is, I guess, that we are talking about the tech area here, and Google

00:25:32: - and Microsoft are the giants here, so they have lots of capabilities to develop their

00:25:37: - own digital product, so to say.

00:25:40: - I guess that's where this NVIDIA example would be more complex and also something we are

00:25:45: - not really experts in, so as I said, it's a different topic.

00:25:48: - But as you said, you could argue that the same problem exists also there.

00:25:52: - Yeah.

00:25:52: - Yeah.

00:25:54: - Well, the main takeaway is AI is amazing.

00:25:58: - Open AI started it all.

00:26:00: - They might not win in the long term.

00:26:01: - It's hard to tell, but I definitely don't think that all those small startups will win.

00:26:07: - Some might come up with really great new ideas and, yeah, might become big companies, but

00:26:14: - I think that for a lot of business models and ideas, it will be the established players

00:26:20: - who will win in the near and definitely long-term future.

00:26:24: - Still, I think if you got a great AI idea, try to make it reality at the moment, because

00:26:31: - now I think it's still the right time to do it.

00:26:32: - Absolutely.

00:26:33: - And you could get bought or maybe you have like a couple of years or months at least

00:26:38: - where you can make good money off of some ideas.

00:26:41: - So absolutely, it's a good idea because you can't lose much.

00:26:45: - I just don't think that the next Google or Microsoft will – is just around the corner

00:26:52: - because of AI.

00:26:53: - It might.

00:26:54: - I might be overlooking some totally obvious great business idea, but I'm not convinced

00:27:00: - that this is the case.

00:27:01: - Yeah.

00:27:02: - I would agree with you here that that's – it's a good point and we'll see.

00:27:07: - That's the other thing.

00:27:07: - Yeah.

00:27:07: - We'll of course see.

00:27:08: - We never know.

00:27:08: - So if you got great AI ideas, share these with us.

00:27:11: - Send them to us.

00:27:12: - Send them to us.

00:27:13: - We'll see what we can do.

00:27:15: - And no, but that's our take on this whole how AI is the new internet thing, I guess.

00:27:23: - And with that, we got the next episode.

00:27:26: - Yeah.

00:27:27: - Hopefully see you back in the future.

00:27:28: - Indeed.

00:27:29: - And share what you think about AI and bye bye.

00:27:32: - Bye bye.