Let’s Talk About Tech

Flynn AI's CEO explains how regulatory AI automates MedTech compliance and creates competitive moats

32 min Listen to the original Episode page
MedTechRegulatory AIStartup StrategyHealthcare AutomationVenture Capital

Summary

Tibor Zechmeister, Head of Quality and Regulatory at Flynn AI, discusses using software to automate regulatory paperwork for medical device companies. He argues that AI cannot fix broken processes and must be benchmarked against human experts for validation. Zechmeister claims EU certification provides a multi-year competitive moat for startups over large incumbents. He also critiques the European MedTech ecosystem for home bias and lack of business expertise among technical founders.

Episode sections

  1. 0:00 – 0:51

    Introduction and Sponsor

    Host introduces topic and sponsor Venina Capital.

  2. 0:51 – 5:55

    Guest Background and Motivation

    Tibor Zechmeister shares personal surgical experience and 15+ years of experience.

  3. 5:55 – 10:05

    Entrepreneurial Lessons and Book

    Discussion of five prior companies and release of book 'Subtract to Ship MDR'.

  4. 10:05 – 12:02

    Flynn AI's Value Proposition

    Software automates regulatory paperwork for large MedTech companies.

  5. 12:02 – 14:11

    Process Improvement vs. Automation

    AI cannot fix broken processes; speed is not efficiency.

  6. 14:11 – 16:06

    Regulation as Competitive Advantage

    EU certification creates a multi-year moat for startups.

  7. 16:06 – 19:50

    AI Validation and Benchmarking

    AI must be benchmarked against human experts in regulated industries.

  8. 19:50 – 28:01

    European MedTech Ecosystem Challenges

    Gaps in matchmaking, home bias, and long adoption timelines.

  9. 28:01 – 28:54

    Accessibility of Regulatory Knowledge

    Democratizing regulatory info through educational resources.

  10. 28:54 – 29:25

    Promotion of Educational Platform

    Promotion of 'subtract to ship from from you' platform.

  11. 29:25 – 32:08

    Timeline for Advanced Technologies

    Stem cells and artificial hearts are 15-40 years away; AI for aging is a frontier.

Claims worth verifying

Tibor Zechmeister is the Head of Quality and Regulatory at Flynn AI.

Unidentified speaker ·0:45

How to verify: Check LinkedIn profile of Tibor Zechmeister or the 'About Us' page on the Flynn AI website.

Venina Capital is the sponsor of the podcast and is an entrepreneurial partner for venture capital and private equity.

Unidentified speaker ·1:32

How to verify: Check Venina Capital's website or marketing materials for their stated role and partnership with the podcast.

The guest had a routine throat surgery at the University Hospital of Graz that resulted in severe pain and a lack of apology from the doctor.

Unidentified speaker ·3:01

How to verify: This is a personal anecdote; verification would require independent confirmation from the guest or public records if the incident was reported, though primarily it is a subjective claim.

The guest's first startup, Tremetors, was certified in multiple countries and continents but failed due to regulatory and commercial mistakes.

Unidentified speaker ·8:18

How to verify: Search for 'Tremetors' in business registries or news archives to confirm its existence, certification status, and eventual failure or acquisition.

The book 'Subtract to Ship MDR' was released on September 2nd.

Unidentified speaker ·7:55

How to verify: Check Amazon, Goodreads, or the publisher's website for the release date of the book 'Subtract to Ship MDR' by Tibor Zechmeister.

Flynn AI's software automates regulatory paperwork for companies managing 20,000 device types across 160 countries.

Unidentified speaker ·10:21

How to verify: Review Flynn AI's product documentation, case studies, or website features list to confirm the scope of their automation capabilities.

OpenAI and Anthropic spend billions of dollars on AI development but still experience hallucinations.

Unidentified speaker ·16:26

How to verify: Check public financial reports of OpenAI and Anthropic for R&D spending, and review technical reports or news articles documenting AI hallucination incidents.

Medical device adoption rates in Europe can take up to 10 years to reach the average hospital.

Unidentified speaker ·22:41

How to verify: Search for industry reports or studies on medical device adoption timelines in European healthcare systems.

The guest has spent over 12 years at the intersection of medical device development, regulatory affairs, and entrepreneurship.

Unidentified speaker ·2:05

How to verify: Check the guest's LinkedIn profile or professional biography for career history and tenure.

Regulatory certification in the European Union provides a two-year competitive advantage over large multinational competitors.

Unidentified speaker ·15:33

How to verify: This is a strategic claim; verify by analyzing case studies of medtech startups that entered the EU market before large competitors and comparing their market share or time-to-revenue.

The cost of acquiring knowledge to understand the medtech industry should not be around 50,000 euros.

Unidentified speaker ·28:04

How to verify: Check market rates for MedTech regulatory consulting, certification courses, and specialized legal advice to see if 50k EUR is a standard high-end benchmark or an outlier.

Only 0.01% of the population knows the secrets of how to get a medical device onto the market.

Unidentified speaker ·28:14

How to verify: This is a rhetorical hyperbole; verify by checking the number of certified regulatory affairs professionals globally versus the total population to assess the scale of expertise.

The speaker posts regulatory content on LinkedIn in a beginner-friendly way.

Unidentified speaker ·28:30

How to verify: Visit the speaker's LinkedIn profile (identified as Tibor in the transcript) and review the complexity and frequency of his posts regarding MedTech regulation.

Stem cells and cybernetics (artificial hearts/lungs) will not reach patients in the next five years.

Unidentified speaker ·29:58

How to verify: Review FDA/EMA approval timelines and clinical trial phases for artificial organs and stem cell therapies to determine current commercialization status.

Realistic timeline for artificial organ replacement is 30-40 years.

Unidentified speaker ·30:04

How to verify: Consult expert forecasts from organizations like the WHO or major medical journals regarding the commercial viability of bioengineered organs.

AI and medical science progress could lead to slowing down aging.

Unidentified speaker ·30:04

How to verify: Review current research papers and clinical trials focused on senolytics, geroscience, and AI-driven drug discovery for age-related diseases.

The speaker's name is Tibor.

Unidentified speaker ·29:04

How to verify: Cross-reference the name 'Tibor' with the podcast guest's official bio or social media handles mentioned or linked in the episode description.

Do your own research

  • Verify Flynn AI's current customer base and revenue model via their website or press releases.
  • Check the publication date and content of Zechmeister's book 'Subtract to Ship MDR' for specific regulatory claims.
  • Review EU MDR certification timelines for comparable MedTech startups to validate the 'multi-year moat' claim.
  • Search for academic or industry reports on human error rates in regulatory compliance to benchmark AI performance.
  • Investigate the adoption rates of MedTech in European vs. US markets to assess the 'home bias' argument.
  • Look for conflicts of interest regarding Zechmeister's educational platform 'subtract to ship from from you'.
  • Verify the timeline claims for stem cells and artificial hearts against current clinical trial data.
  • Check Venina Capital's portfolio for any existing investments in Flynn AI or related MedTech firms.
Full transcriptexpand
0:00
Unidentified speaker: Imagine you're in a big medical device company, really a big one, a multinational corporate, and you have like 20,000 different medical device types here in 160 countries in America.
0:09
Unidentified speaker: Each of these device types needs to be regulated, meaning you need to do a lot of paperwork.
0:14
Unidentified speaker: When I'm talking about a lot of paperwork, I mean you can sit there for two years to just do the paperwork from one device and then multiply that by 20,000, for example.
0:22
Unidentified speaker: So this is where automation and film comes into play, where we say, look, you have excellent human people there and experts who are great regulatory affairs people, but should they really be doing repetitive manual jobs where honestly also an automation could do that or AI could support for that?
0:38
Unidentified speaker: And that's the basic idea.
0:39
Unidentified speaker: So reducing bureaucracy in this sense, that is really helpful.
0:43
Unidentified speaker: I mean, this is basically saving lives.
0:45
Unidentified speaker: Today I am joined by head of quality and regulatory headlink AI.
0:51
Unidentified speaker: I want to save patients, not only me, but other ones.
0:55
Unidentified speaker: I don't want them to become sick because we have so many issues with a sick society where people just get older, we don't know how to deal with certain diseases like Parkinson's and so on.
1:04
Unidentified speaker: I really want to see technology medical devices preventing these things, and if that's impossible, saving us and protecting us.
1:11
Unidentified speaker: This is why I'm in this industry after altogether more than 15 years now, we're gonna also have the talent and that's also the case in the med tech sector.
1:20
Unidentified speaker: The gap is between innovation and what are the companies that successfully cross that gap doing right?
1:26
Unidentified speaker: Personally, I think there are three reasons for that.
1:29
Unidentified speaker: Of course, there are many more, but three are a little bit outstanding.
1:32
Unidentified speaker: Number one, hello and welcome to Let's Talk About Tech provided to you by Venina Capital, the entrepreneurial partner for venture capital and private equity.
1:53
Unidentified speaker: Today I am joined by Tibor Zechmeister, head of quality and regulatory at Flynn AI.
1:59
Unidentified speaker: Welcome, Tibor.
2:00
Unidentified speaker: Thank you so much for having me.
2:02
Unidentified speaker: Really excited to be here on the show.
2:04
Unidentified speaker: Of course.
2:05
Unidentified speaker: I mean, you have spent over 12 years at the intersection of medical device development, regulatory affairs, and entrepreneurship, Tibor.
2:13
Unidentified speaker: And uh you begin, I mean your journey began uh as you studied biomedical engineering, actually, and you ended up founding a medical device company that got approved in three continents and was acquired, correct?
2:25
Unidentified speaker: So walk us through your journey.
2:27
Unidentified speaker: Um, I mean, your career development uh and uh how you you know ended up now at Flynn Ai.
2:32
Unidentified speaker: Yeah, absolutely.
2:33
Unidentified speaker: More than happy to do so.
2:35
Unidentified speaker: Thank you so much.
2:36
Unidentified speaker: Yeah, maybe let's start with the topic of why you got into medical devices and into the medical device industry.
2:41
Unidentified speaker: So, of course, I could now give you an example of yes, I always loved biology and technology, so the two were absolutely logical.
2:49
Unidentified speaker: But I would like to tell you a little bit of a wake-up story I had a few weeks ago.
2:54
Unidentified speaker: And this wake-up story reminded me again why I'm in this industry after, yeah, altogether more than 15 years now.
3:01
Unidentified speaker: I had a regular surgery at the hospital of Graz, at the University Hospital three weeks ago, four weeks ago.
3:08
Unidentified speaker: And it was really a routine throat surgery that should have been done like a 10-minute event.
3:13
Unidentified speaker: Nothing serious, everything should be easy.
3:15
Unidentified speaker: It didn't go well.
3:17
Unidentified speaker: So I didn't get the treatment during the surgery under anesthesia.
3:21
Unidentified speaker: The doctor seemed to be very nervous, shoved on a medical device into my throat, she slit open my throat and everything.
3:28
Unidentified speaker: I had severe pain for one week.
3:30
Unidentified speaker: Wow.
3:31
Unidentified speaker: And the biggest problem was I didn't get a sorry or excuse me.
3:34
Unidentified speaker: It was like, yeah, yeah, I did my best, and that was okay.
3:37
Unidentified speaker: What I don't want to tell with the story is that doctors are bad or they have bad intentions or something like that.
3:43
Unidentified speaker: But the issue is they're overworked, the system is making them being overworked, tired, and then these mistakes are happening.
3:50
Unidentified speaker: And this really reminded me, once again, why I'm into medical devices.
3:54
Unidentified speaker: On the one hand, I want to see medical devices that prevent us from getting actually into the hospital, meaning they should be preventive, they should be helping us, not creating any chronical diseases or something like that.
4:07
Unidentified speaker: They should really prevent us from any surgeries that could be happening.
4:10
Unidentified speaker: But if you get into the hospital, the medical devices should be so safe that even if a doctor might make a mistake and tries to shove down a medical device a patient's throat, that shouldn't be happening.
4:21
Unidentified speaker: A medical device is smart enough detecting that, maybe sounding an alarm or being not even making it possible to get into that situation.
4:30
Unidentified speaker: And you know, when I was waking up, it was hurting super bad.
4:34
Unidentified speaker: And I was like, but this was not the medical devices problem.
4:37
Unidentified speaker: This was human error.
4:38
Unidentified speaker: So this is why I'm in this industry.
4:40
Unidentified speaker: I I want to save patients, not only me, but other ones.
4:43
Unidentified speaker: I don't want them to become sick because we have so many issues with a sick society where people just get older, we don't know how to deal with certain diseases like Parkinson's and so on.
4:54
Unidentified speaker: And I really want to see technology, medical devices, as I said, preventing these things.
4:58
Unidentified speaker: But if that's not possible, saving us and protecting us.
5:01
Unidentified speaker: Yeah.
5:02
Unidentified speaker: During surgeries, during recovery, during treatment, and so on.
5:05
Unidentified speaker: Definitely.
5:06
Unidentified speaker: I mean, this does not sound like a normal procedure, actually, because he said it was like a common thing, but like slitting your throat over.
5:12
Unidentified speaker: Yeah, it was not too funny altogether.
5:14
Unidentified speaker: Yeah, it happened.
5:15
Unidentified speaker: But again, I don't want to make the point here that that I blame the doctor personally, that she intentionally did that.
5:21
Unidentified speaker: I mean, she was definitely stressed and said, hey, that's not working out.
5:24
Unidentified speaker: Let's try it again a little bit harder to to to get to the to the area for treatment.
5:29
Unidentified speaker: But then again, I know we are all human, we are tired, we are making mistakes, but there is a difference if I make a mistake wiring some money through a banking account or yeah, operating on a patient for something could go wrong.
5:42
Unidentified speaker: Yeah, and you're saying, I mean, it's a problem in the system.
5:45
Unidentified speaker: I mean, you are a serial entrepreneur and you have built five companies actually building or improving uh this system and each one informed by the last.
5:55
Unidentified speaker: So, what does that compounding experience actually feel from the inside?
5:58
Unidentified speaker: And I mean, what did the earlier ventures teach you that you could uh not have learned any other way?
6:04
Unidentified speaker: Yeah, really good point.
6:05
Unidentified speaker: I would say that of course, most of these companies were in the medical device industry.
6:11
Unidentified speaker: And of course, if you have an entrepreneurial journey, you're making so many mistakes.
6:14
Unidentified speaker: So if if I'm looking back for through the last 15 years, oh man, for every right decision, I probably made 100 wrong decisions and incorrect decisions.
6:22
Unidentified speaker: So somehow that's the that's the curse of being an entrepreneur that you're just making as many mistakes as possible.
6:28
Unidentified speaker: But this is also the good learning thing.
6:30
Unidentified speaker: And to answer your question, interestingly enough, of course, I had a steep learning curve.
6:35
Unidentified speaker: After my first startup, I knew a lot more than being a first-time founder.
6:40
Unidentified speaker: And of course, now with the current venture, it's much easier to assess things than with the first or second venture.
6:45
Unidentified speaker: But interestingly enough, every venture taught me one extreme distinct thing.
6:49
Unidentified speaker: For example, the first company you mentioned before, that taught me about the founder fit.
6:54
Unidentified speaker: No matter how good your product is, no matter how perfect the market is, if you're not having the right co-founders and partners to build everything, it will go wrong.
7:03
Unidentified speaker: You can have a terrible product, but an excellent team, you will be somehow able to bring that onto the market and maybe even be successful.
7:10
Unidentified speaker: But that was my first learning.
7:12
Unidentified speaker: For the second company, for example, I saw that product-market fit was the biggest issue.
7:17
Unidentified speaker: There you had, okay, good team, we have good people, there is some liquidity and funding there, but we saw that the initial market that didn't work out properly.
7:24
Unidentified speaker: We we had to pivot to another market.
7:26
Unidentified speaker: And in the third startup, for example, it was about timing.
7:29
Unidentified speaker: I saw the tech was right, there would be money and everything, but I saw okay, the timing is not working out.
7:34
Unidentified speaker: So of course you learn a million things, but for me, it was really just burning onto my, I don't know, onto my forehead.
7:41
Unidentified speaker: Every startup told me no matter what you do the next time, you need to have a checklist and ask these things, otherwise it's not working out.
7:48
Unidentified speaker: I mean, it sounds like you went through basically every stage of failure, so let's just say of the common failures in the startup world.
7:55
Unidentified speaker: But I mean, you turned that experience into a book subtract to ship MDR.
7:59
Unidentified speaker: I mean, this is out since September 2nd.
8:02
Unidentified speaker: So uh what is the core argument there?
8:04
Unidentified speaker: Exactly.
8:05
Unidentified speaker: So with my first startup, we had the situation that it was, as you mentioned, certified in multiple countries, all of Europe, free continents, as you mentioned, but still we didn't have at the end of the day enough runway.
8:18
Unidentified speaker: Why?
8:18
Unidentified speaker: Because I made so many mistakes I mentioned before, regulatory mistakes, commercial mistakes.
8:24
Unidentified speaker: And you know, this story has haunted me for 15 years because I saw other startups doing the same mistakes.
8:30
Unidentified speaker: And if I had the possibility to have a conversation with them, I told them, look, don't do this, go there, or this is the way.
8:37
Unidentified speaker: But then I saw, okay, this is just a limited possibility.
8:40
Unidentified speaker: So together with Felix Leinhardt, we we discussed about this topic.
8:44
Unidentified speaker: And I said, Felix, we somehow need to communicate that to people so that they don't, I don't know, have their next bad experience and I think that they're incompetent or something like that.
8:53
Unidentified speaker: Yeah, so we said, you know what?
8:55
Unidentified speaker: We take this experience of this first company Tremetors, collect everything that went wrong there, and give you basically a playbook how not to become the next Tremetus that failed at the end of the day, but how you can have the maximum success rates with your medical device startup.
9:10
Unidentified speaker: Not only on the regulatory side, meaning how what does it mean to get onto the European market, for example, but also check your commercials.
9:17
Unidentified speaker: Is the market big enough?
9:19
Unidentified speaker: Who is the actual buyer on that one?
9:20
Unidentified speaker: Do you really have enough runway, for example, or not?
9:23
Unidentified speaker: So this would put together all yeah into one book stating look, you don't need 20,000 euros of expert consulting or something like that to get these basics.
9:33
Unidentified speaker: Get the book, get the copy of that, and then you get all the necessary information for at least an 80-20 approach when you start off.
9:39
Unidentified speaker: Yeah, yeah.
9:40
Unidentified speaker: And also, I mean, it's such a common cliche or phrase, but like failing is learning basically.
9:45
Unidentified speaker: Absolutely.
9:46
Unidentified speaker: And uh it's it's uh very good, I would say.
9:48
Unidentified speaker: I mean, with subtract to ship that uh startups now or founder are not turning to themselves and say I'm incompetent, but things are happening in the startup world and you can fail.
9:59
Unidentified speaker: And I mean, now you are at a company, Flynn AI, that actually does great, I would say.
10:05
Unidentified speaker: So, what is Flynn actually building and what problem in the medtech world is it solving?
10:09
Unidentified speaker: Yeah, that's really a great question.
10:11
Unidentified speaker: So I will try to formulate that in in a way so that we we don't dive into medical device regulation jargon or something like that.
10:21
Unidentified speaker: So imagine you're in a big medical device company, really a big one, multinational corporate, and you have like 20,000 different medical device types.
10:30
Unidentified speaker: They're in 160 countries and everything.
10:33
Unidentified speaker: Each of these device types needs to be regulated, meaning you need to do a lot of paperwork.
10:38
Unidentified speaker: So when I'm talking about a lot of paperwork, I mean you can sit there for two years to just do the paperwork for one device and then multiply that by 20 for 20,000, for example.
10:48
Unidentified speaker: I mean, one option you can now do is say, okay, I'm just hiring 20,000 more people to work on that thing, but we know reality works a little bit differently.
10:56
Unidentified speaker: So this is where automation and Flynn comes into sp into play, where we say, look, you have excellent human people there and experts who are great regulatory affairs people, but should they really be doing repetitive manual jobs where honestly also an automation could do that or AI could support with that?
11:14
Unidentified speaker: And that's the basic idea.
11:16
Unidentified speaker: We take the existing regulatory processes, all this paperwork, documentation they need to do, and provide our software so that it gets more automated, they are faster, and more importantly, the I would say the boring part of their job is really eliminated or automated so they can focus on the more important things, like as we heard before, patient safety.
11:38
Unidentified speaker: Yeah.
11:38
Unidentified speaker: We should think about could this medical device hurt people if it's out there on the market?
11:43
Unidentified speaker: Could the surgeon maybe misuse the medical device in a certain situation?
11:47
Unidentified speaker: But right now they cannot do that because, as I said, it's just paperwork after paperwork after paperwork they are doing.
11:53
Unidentified speaker: Yeah, so reducing bureaucracy in this sense that is really helpful.
11:57
Unidentified speaker: I mean, this is basically saving lives, you know?
12:00
Unidentified speaker: Exactly.
12:00
Unidentified speaker: And uh yeah, this is fundamental.
12:02
Unidentified speaker: Um, so moving on, I mean, how far can software actually go in fixing a regulatory system that is fundamentally a systemic and human problem?
12:12
Unidentified speaker: Isn't that a paradox to some extent?
12:13
Unidentified speaker: Yeah, I really love that question.
12:15
Unidentified speaker: This is really a great one.
12:16
Unidentified speaker: The reason is, and this is really, as you say, a little bit a paradox or ironic.
12:22
Unidentified speaker: We oftentimes visit companies that would like to use our software, but when we check their processes, we see that the processes are broken.
12:30
Unidentified speaker: They took some processes from the 80s, it was never properly updated, it was just patched together, and then somehow you have now a patchwork process.
12:37
Unidentified speaker: And then they say, Okay, your software is a little bit expensive.
12:41
Unidentified speaker: Let's uh I hope it will solve everything.
12:45
Unidentified speaker: But then we step back and say, look, no AI, no automation, and no software can fix a broken process.
12:52
Unidentified speaker: So, first, what we need to do, let's have a workshop, let's have a conversation.
12:56
Unidentified speaker: How could we improve your process or or reduce the burden there?
13:00
Unidentified speaker: And then we can do automation.
13:02
Unidentified speaker: Because I think we have so many examples, other examples in life where we say if it's broken, you don't need to automate it.
13:08
Unidentified speaker: I mean, if your email account is getting 99.9% spam emails, then maybe the problem is not automating the spam email triage.
13:17
Unidentified speaker: Maybe the alternative would be to get a new email account and say, hey, this one is broken, we we get a new one where you can triage.
13:24
Unidentified speaker: So great question.
13:25
Unidentified speaker: And I think this is a general societal problem that people believe that AI and automation can solve everything.
13:31
Unidentified speaker: It's like a magical black box where you say exactly I I just give it to AI and then it will solve it.
13:36
Unidentified speaker: But then again, if you're in the business, especially, I mean, you you cannot fix uh exactly.
13:42
Unidentified speaker: And I I love that you're saying that because I think it's a common misconception also that we we think that velocity and speed, you know, which is enabled by artificial intelligence, actually, yeah, um is like uh efficiency.
13:56
Unidentified speaker: And we sometimes I think in my humble opinion, we we have a misconception here.
14:01
Unidentified speaker: I mean, as you as you said, taking the time to do the workshop, you know, and fixing the broken process and then you know accelerating with AI, yeah.
14:09
Unidentified speaker: You know, this is the the way to go, I think.
14:11
Unidentified speaker: And you're going in that direction and actually also covering uh the MDR, so the medical device regulation.
14:18
Unidentified speaker: And this is demanding, but that's not in dispute.
14:21
Unidentified speaker: But uh, you've argued uh before that rigorous regulation can be a competitive advantage.
14:26
Unidentified speaker: So how does a company turn compliance into something that helps them win in a market?
14:32
Unidentified speaker: Absolutely great question.
14:33
Unidentified speaker: So I think the first thing, especially startups think about when they see all these regulations, and let's be honest, you have the same thing in the US, you you have the same thing in Asian countries, Australia.
14:43
Unidentified speaker: So you have regulations for medical devices everywhere.
14:46
Unidentified speaker: The first reaction is this is slowing me down.
14:49
Unidentified speaker: This this is killing innovation and so on.
14:51
Unidentified speaker: And yes, we can discuss in a nuanced way how much regulation uh destroys more innovation or protects patients, but let's be honest, at the end of the day, we are talking about human lives.
15:01
Unidentified speaker: Yeah.
15:01
Unidentified speaker: So again, this is not a bank account where we make a 50 euro mistake with a wiring where they say, whoopsie, sorry, here's your money back.
15:08
Unidentified speaker: We are talking about a patient that might die if it's not properly tested, this medical device.
15:13
Unidentified speaker: So, how can we use this as a competitive advantage?
15:16
Unidentified speaker: Imagine you're a small startup and you get actually a certification for your medical device in the European Union.
15:23
Unidentified speaker: You have some diagnostic software or something like that.
15:26
Unidentified speaker: The interesting thing here is yes, it will be brutal.
15:29
Unidentified speaker: It will take time, it will take money and a lot of resources.
15:33
Unidentified speaker: But as soon as you get the certificate, even if a huge competitor, multinational competitor says, Oh, we want it too, and we want to crush that startup, they still need to go through the same slow procedure as the startup, meaning you have a two-year advantage, even in comparison to a big corporation, another startup, and so on.
15:52
Unidentified speaker: So, this is what I really like to communicate.
15:54
Unidentified speaker: Yes, it's definitely hard to get there, but it's a perfect competitive mode to be in that position.
16:01
Unidentified speaker: Yeah, I mean, uh sounds uh sounds uh about right, I would say.
16:06
Unidentified speaker: Um, I mean, let's talk about artificial intelligence more in depth.
16:11
Unidentified speaker: Uh, I mean, AI is basically in your name and that's deeply embedded in Flint's product.
16:15
Unidentified speaker: But how do you actually validate AI itself?
16:18
Unidentified speaker: Like, how do you prove something that learns and changes?
16:21
Unidentified speaker: Yeah, so I think this is one of the biggest technological questions everyone is asking right now.
16:26
Unidentified speaker: Because even if you if you look at huge providers such as OpenAI or Entropic and so on, no matter how many billion dollars and euros you are spending there, sometimes the thing is still hallucinating, sometimes the thing is still inventing some facts or getting a little bit slow.
16:42
Unidentified speaker: And this is what brings me to this whole concept of validation.
16:45
Unidentified speaker: I think we we we need to stop interpreting a machine or AI as a perfect entity.
16:51
Unidentified speaker: Because for some reason, and one of my favorite examples is um driving driving a bus or flying a plane, for example.
16:59
Unidentified speaker: So society is super terrified by having a machine drive a bus or or flying a plane, but a tired human again who who had to work through 30 hours and is now super tired, and now they need to drive on the highway for another 16 hours because they missed, uh didn't get some sleep, and so somehow we have trust in there.
17:19
Unidentified speaker: Yeah.
17:19
Unidentified speaker: So again, doesn't mean that the human is having bad intentions, but we have this comparison of yeah, if the human is making mistakes, that's absolutely fine, but the machine needs to be perfect.
17:30
Unidentified speaker: So I think for the whole validation topic, what people need to understand, there is no 100% reliable technology or automation or AI altogether.
17:39
Unidentified speaker: However, we can do benchmarking.
17:40
Unidentified speaker: What does that mean?
17:42
Unidentified speaker: If we check, for example, uh taking the Flynn case, regulatory experts, we did that actually.
17:47
Unidentified speaker: We asked experts, hey, look, here is an exercise.
17:50
Unidentified speaker: Tell us how you would make a classification or a decision for a certain regulatory question.
17:56
Unidentified speaker: And we checked it again against AI, for example.
17:59
Unidentified speaker: And funnily enough, sometimes even the experts, because we took the test on one day and then one month later again the same test, sometimes the experts contradicted themselves after one month.
18:10
Unidentified speaker: Sometimes they agreed with each other, sometimes not.
18:13
Unidentified speaker: And in most of the cases, it turned out that even the experts said, oh, what the AI said could be actually even a better decision.
18:19
Unidentified speaker: So that would be interesting.
18:21
Unidentified speaker: So to really cut it short, if we talk about validation, A, we need to accept it's not 100% reliable.
18:28
Unidentified speaker: We should set up a benchmark what is good enough for the AI.
18:31
Unidentified speaker: And if we know if this is 80%, 90%, or 99% or something like that, then we can talk about actual validation.
18:39
Unidentified speaker: Like, do we have a ground truth testing where we as a human know what the right answer would be, and we check if the AI is making the same answer?
18:47
Unidentified speaker: Or if you have a situation where even humans disagree, you could have uh majority votes where you say three humans are making a decision, two out of three, that will be the gold standard answer, and then we check the AI against that.
18:59
Unidentified speaker: One more comment on that, because I also admit I'm not a data scientist, and the data scientists at Flynn AI are brilliant geniuses, where I'm a caveman in comparison to that one, if I'm looking at what they are capable of doing.
19:12
Unidentified speaker: But I think that the last point you mentioned about constantly learning, that's the important point.
19:16
Unidentified speaker: In regulated industries like the medical device industry, you don't just let it learn continuously, you train it, you test it, you freeze it, you release the frozen version, which is then hopefully reliable.
19:28
Unidentified speaker: And internally you can already develop the new version, but then again, of course, again you train it, test it, and only then do you release that.
19:35
Unidentified speaker: So for me, this is the smartest way to go forward, and of course, also the safest way to not have some random experiences where the AI is drifting or starting to hallucinate or something.
19:46
Unidentified speaker: Yeah, there will be a necessity in the regulatory space to go through like that.
19:49
Unidentified speaker: Exactly.
19:50
Unidentified speaker: Yeah.
19:50
Unidentified speaker: Um, I mean, let's talk about the industry and you as a startup founder in Austria or basically Europe.
19:56
Unidentified speaker: Um, you are experienced with that.
19:59
Unidentified speaker: I mean, uh a very interesting point in the media all the time.
20:04
Unidentified speaker: But I mean, we as Europe and Austria have the talent and the technology, and that's also the case in the uh medical in the med tech sector.
20:12
Unidentified speaker: Um, the gap is between innovation and market.
20:15
Unidentified speaker: So, what is your take?
20:17
Unidentified speaker: Um, what are companies that successfully cross that gap doing right compared to the others that are not doing it right?
20:24
Unidentified speaker: Yeah, this is this is a really great question, and you know, of course, we now could be discussing about the things everybody is discussing about.
20:32
Unidentified speaker: We don't have enough capital, the regulation is too hard.
20:35
Unidentified speaker: But let's be honest, there are also other places where you don't have a lot of money and a lot of regulation, and still companies and people are succeeding.
20:42
Unidentified speaker: So I wouldn't focus on that one.
20:45
Unidentified speaker: Personally, I think there are three reasons for that.
20:48
Unidentified speaker: Of course, there are many more, but three are a little bit outstanding for me.
20:51
Unidentified speaker: Number one, I'm somehow missing in Austria and in Europe this logical matchmaking of the right skills and the partners.
20:59
Unidentified speaker: If I go to San Francisco in the US and there is a brilliant CTO who had the best ever invention, you get the CEO, you get the go-to-market expert and everything, and they're joining forces and saying, look, of course, this has to be working together.
21:12
Unidentified speaker: In Austria, for example, you have a spin-off from university, three brilliant engineers and scientists, top of their field, probably worldwide.
21:20
Unidentified speaker: They have never written a business plan.
21:22
Unidentified speaker: They have no idea what the product market fit is, but still they stay in their isolated ecosystem saying, but what if somebody else steals my idea?
21:30
Unidentified speaker: Or what if they're not good enough?
21:33
Unidentified speaker: Or or they're lacking some skills.
21:35
Unidentified speaker: So I think this ecosystem thing is something I fear we are still a little bit lacking because instead of just joining forces and saying, look, yeah, okay, then they get 25% of the company.
21:45
Unidentified speaker: It's okay.
21:46
Unidentified speaker: But it's better to have 25% of a successful company than 100 or or 99% of a failing company.
21:52
Unidentified speaker: I think that's one missing point.
21:54
Unidentified speaker: The the second point is this home bias, which we have in Europe.
21:59
Unidentified speaker: We are an I'm an Austrian startup.
22:01
Unidentified speaker: My first market has to be in Austria.
22:03
Unidentified speaker: Um, I don't know, Germany, German startup, and so on.
22:06
Unidentified speaker: But sometimes your technology might be absolute medical device technology, might not be even needed in Austria, for example, because the density of specialist doctors for your certain case is so high that nobody needs the technology.
22:19
Unidentified speaker: But look at other markets.
22:20
Unidentified speaker: Maybe it's a completely different story in the US or in India, for example.
22:24
Unidentified speaker: So the second factor is this is this I know it's cliche, but think a little bit bigger.
22:29
Unidentified speaker: Think outside of your home market, what could be really your target market?
22:33
Unidentified speaker: And the third thing is connecting to the second thing, and this is really about people need to understand adoption rates in Europe can be slow.
22:41
Unidentified speaker: So even if you have a new breakthrough technology and medical technology, that could take up to 10 years until it lands in the average hospital and in the average treatment.
22:51
Unidentified speaker: What does that mean?
22:52
Unidentified speaker: Think about the runway.
22:53
Unidentified speaker: So don't pitch for 500,000 euros.
22:56
Unidentified speaker: If you know you need 7 million euros to reach the market properly, please pitch for 7 million euros, even though people are super afraid about these numbers and this is too much.
23:05
Unidentified speaker: And on the other hand, if you know it takes you seven years or 10 years to get onto the market, again, think about beachhead market somewhere else.
23:12
Unidentified speaker: Meaning, okay, Germany takes me seven years, but I don't know, in Hong Kong, we could already launch in two years.
23:18
Unidentified speaker: So in one year, make some revenue and use that revenue to survive as a startup.
23:22
Unidentified speaker: So again, I know there are a million other reasons why why this couldn't be working, but these are some repetitive patterns I'm definitely seeing in the med tech industry.
23:30
Unidentified speaker: You're saying like thinking bigger, you know, like actually globally.
23:35
Unidentified speaker: Um resilience, as I understood, a big part of that, you know, because as you said, sometimes it takes seven years actually in in some markets.
23:43
Unidentified speaker: And also, and this is something I want to circle back a little bit, um, matching, like I would say, specialists in the field regarding technological specialists, for example, or researchers who are highly sophisticated and connecting them with uh people with business acumen.
24:01
Unidentified speaker: Yeah, this is like basically on that of head.
24:05
Unidentified speaker: Because sometimes you do have a unicorn, you may might have a scientist who is also a serial entrepreneur who says, Okay, I I know what to how to figure it out.
24:13
Unidentified speaker: But let's be honest, we have so many great spin-offs.
24:16
Unidentified speaker: And I I talk to so many of these scientists, again, in comparison to them on their field, I'm like a caveman saying, Okay, you know, 10 times more than me on that topic, even if I would be learning five years about that.
24:27
Unidentified speaker: But that's exactly the problem.
24:28
Unidentified speaker: You cannot be an expert and a perfect expert in every field.
24:33
Unidentified speaker: Be the perfect scientist, be the chief research officer, be the chief technological officer in the project, but understand that the best technology you're producing won't get onto the market if you don't have a go-to-market strategy, if you don't have a marketing approach, if you don't know how sales work.
24:50
Unidentified speaker: I mean, even simple things like unit economics, you you need to calculate.
24:55
Unidentified speaker: If your if your product costs you 50,000 euros in manufacturing, you cannot sell it for 80,000 because you need to consider development times, the personnel, the employee costs, HR, certification costs, and everything.
25:09
Unidentified speaker: But these are things if if you never had touch points with that, and let's be honest, many technical universities and also scientific universities usually don't teach you entrepreneurship or you don't jump just into two classes of entrepreneurship.
25:22
Unidentified speaker: Yeah, yeah.
25:23
Unidentified speaker: So this is what exactly as you point out, people should be really open to discuss with these other kinds of experts and and not have this feeling of yeah, but I'm the true expert, and sales is just stupid because science is more important than sales, you know what I mean.
25:38
Unidentified speaker: So the barrier should go down a little bit.
25:41
Unidentified speaker: Exactly.
25:42
Unidentified speaker: Don't underestimate the go-to market and bringing that, I mean onto onto the market, the innovation.
25:48
Unidentified speaker: Yeah.
25:49
Unidentified speaker: Um now if we look uh in our magic classbook and looking into the future, if you could implement one structural change to get more mettech innovation to patients faster, what would it be?
26:01
Unidentified speaker: So, you know, I think many people who know me would hope that my answer would be less regulation.
26:07
Unidentified speaker: Let's let's have a fewer regulations and less regulations because everything is just complicated with the paperwork.
26:14
Unidentified speaker: But as I mentioned before, we are talking about patients here.
26:17
Unidentified speaker: Yeah, we're talking about patients' life.
26:19
Unidentified speaker: So even though many would hope or expect the answer is yes, the I don't know, European Commission needs to be less uh strict on these regulations and so on.
26:29
Unidentified speaker: We need some regulation because again, uh similar to cars, we don't want to die in car accidents.
26:34
Unidentified speaker: We don't want to die as patients if the medical pacemaker is not working and your heart is stopping.
26:39
Unidentified speaker: So regulation is good in that field.
26:41
Unidentified speaker: Yeah.
26:42
Unidentified speaker: So I won't give a monologue about how regulation is too complicated and so on.
26:46
Unidentified speaker: However, I see a little bit of an other approach and an angle, and I would like to take an analogy.
26:53
Unidentified speaker: Let's talk very briefly about financial literacy.
26:56
Unidentified speaker: 15 years ago, it it was a little bit hard to get knowledge about financial literacy.
27:01
Unidentified speaker: What's an ETF?
27:03
Unidentified speaker: How how much Nasdaq should you have?
27:05
Unidentified speaker: Um, should you pick individual stocks or as I said, uh general ETF and so on?
27:10
Unidentified speaker: Nowadays, if you go on YouTube, you get several good channels because the information has been completely democratized.
27:16
Unidentified speaker: And now it's just your individual decision if you consume that information properly and make the right decision.
27:22
Unidentified speaker: And I would really like to go structurally into a similar direction in the medical device industry.
27:28
Unidentified speaker: And this is also one reason why we created this book, saying, no, that shouldn't be a hidden knowledge that costs you 50k in special consultancy or something like that.
27:38
Unidentified speaker: There is a book.
27:39
Unidentified speaker: Will it contain all the, I don't know, all the steps for five years project learning?
27:45
Unidentified speaker: No, of course not.
27:46
Unidentified speaker: But it will give you everything why my first company, Tremetors, failed, so that you have a complete head start.
27:52
Unidentified speaker: And this is my aim to to really democratize this regulatory information and this knowledge, meaning it should be publicly available on LinkedIn.
28:01
Unidentified speaker: There should be YouTube channels for that one, there should be books about that.
28:04
Unidentified speaker: So the cost of acquiring this knowledge to really understand this whole iceberg in the med tech industry shouldn't cost you like, I don't know, 50,000 euros or something like that.
28:14
Unidentified speaker: And it shouldn't be a secret where only 0.01 of the uh of the population know about these magical secrets of how to get a medical device onto the market.
28:24
Unidentified speaker: Are you teaching the the the I would say the contents of your book uh online on YouTube or something like that?
28:30
Unidentified speaker: Um I'm posting on LinkedIn, for example.
28:32
Unidentified speaker: And this is the interesting thing.
28:34
Unidentified speaker: Um, I really try to post it in a simpler way.
28:37
Unidentified speaker: Because let's be honest, if somebody knows an industry for 15 years, I could use every vocabulary, jargon, and complex uh structure to explain something.
28:45
Unidentified speaker: But I think the nicest thing is if you can explain something in a simpler way.
28:50
Unidentified speaker: Yeah, this is why I really try to make these posts on LinkedIn really beginner-friendly.
28:54
Unidentified speaker: Of course, then some experts say, but Tibor, you missed this small part, and this paragraph also tells this when I say, Yes, but this is not a 200-page essay about this topic.
29:04
Unidentified speaker: This is a LinkedIn post for people who have never heard about this topic and would just simply like to know more about that.
29:10
Unidentified speaker: So, I mean, if you are considering starting a startup, or if you are in a med tech uh regulatory space, then you should definitely uh explore subtract to ship from from you.
29:23
Unidentified speaker: Uh one last uh question.
29:25
Unidentified speaker: Um, what are you most excited about in terms of what will actually reach patients in the next five years that doesn't exist today?
29:33
Unidentified speaker: Yeah, this is this is really a cool question.
29:35
Unidentified speaker: I enjoy that one.
29:37
Unidentified speaker: So, you know, um maybe five years is a little bit short, but if you would have asked me that five years ago, I would have gone with stem cells, cybernetics, meaning that if your natural heart is not working, we are replacing that by a better functioning artificial heart or a lung or something like that.
29:54
Unidentified speaker: But if you asked me today, right now, it will not be five years.
29:58
Unidentified speaker: If I'm super optimistic, it will be 15 years.
30:01
Unidentified speaker: If I'm realistic, probably 30, 40 years.
30:04
Unidentified speaker: But with the exponential progress we have made in medical science in the last few years, also with AI and so on, I really hope that soon we get to the point where we can slow down aging.
30:16
Unidentified speaker: Because let's be honest, dying means in many cases you're aging, your heart is not working properly anymore, your lungs are not working properly anymore.
30:23
Unidentified speaker: So, what I'm most excited about is really this point can we slow down aging?
30:29
Unidentified speaker: Is 80 years really the limit?
30:31
Unidentified speaker: Can we have more healthy years during these 80 or 90 or even 100 years?
30:36
Unidentified speaker: I know there are a lot of societal, moral, and ethical questions about what happens if people don't die anymore as soon as uh as nowadays.
30:44
Unidentified speaker: I know this is a complete different discussion.
30:47
Unidentified speaker: Yeah, but as I mentioned in the beginning, I'm so super excited about biology and technology, and that for me is this ultimate step of we died for three million years because that was the ultimate problem due to heart failure or infections or something like that.
31:02
Unidentified speaker: Can we slow that down?
31:04
Unidentified speaker: I I don't say we we need to be immortal or something like that, but can we have some more years and decades just enjoying life, not taking medications, not not having the problem of I don't know, not working limbs or or having dementia or or Parkinson's or some other diseases.
31:20
Unidentified speaker: This is what I'm most excited about.
31:21
Unidentified speaker: But again, I think five years is even for for for the super longevity uh fanatics, even uh very short term.
31:29
Unidentified speaker: Yeah, that could be very short term.
31:31
Unidentified speaker: But uh maybe we will have a conversation in the future and we can talk about exactly this slowing down.
31:39
Unidentified speaker: Uh it was really nice having you on the podcast.
31:42
Unidentified speaker: Thank you very much for joining us and sharing your insights.
31:45
Unidentified speaker: Thank you guys uh for uh tuning in again to Let's Talk About Tech, and we will see you guys in the next episode.
31:59
Unidentified speaker: Let's talk about it.
32:00
Unidentified speaker: Let's talk about tech baby.
32:03
Unidentified speaker: Let's talk about the app.
32:04
Unidentified speaker: Let's talk about it, baby.
32:07
Unidentified speaker: Let's talk about
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3 and interpreted by Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights