EP 200 - MSI - Roland & Doug

Transcript

Doug (00:00)
Hey, Roland, it's great to be on the podcast again. And it's a little weird as one of the biggest fans of the company because I'm the co-founder of it. But I'd love to ask you seriously,why are we launching this podcast now? What's changing in Silicon Valley? And what do we want founders to get out of it?

Roland Siebelink (00:08)
Well, Doug, you are my favorite guest on the podcast, so it's always fun to just do the internal version, talking between co-founders of MidStage Accelerator. We see so many changes in Silicon Valley at this stage. We see new companies being founded, of course, completely focused on AI, but also organizing themselves around AI.

And I think every founder in the world is just super curious about how people are leveraging AI, how they are achieving breakthroughs with AI and trying to learn from that. And we thought that it would be a good opportunity now that we're at episode 200 to really focus our podcast for a while on the breakthroughs that all founders can achieve with AI. This is not our...

opportunity to sell AI or anything. We're not even an AI agency. It's just more, we talk to so many founders in Silicon Valley and beyond that I think there's many lessons to be shared and our typical audience of founders, investors, and advisors can really benefit from.

Doug (01:27)
I love that. And for those of you that aren't familiar with the Midstage Accelerator - and I won't pitch here - we count our success on breakthroughs with businesses we work with. One of the things that resonates with me deeply is we're leading this with breakthroughs because it's not about trying AI, it's not about using AI, it's about breaking through with AI in your operations and with operators. Talk to me a little bit about why that name and what resonates with you.

Roland Siebelink (01:56)
I definitely also feel like it's not about AI. AI is just a means to an end. But it's also a super powerful technology that will allow many people to get to a breakthrough much faster than before. I'll give a very simple example. We often run workshops with clients that help identify the key breakthrough and then often the conclusion as well.

We weren't aware of this blind spot, so let's hire someone who has experience. As an example, they may not be doing enough in demand generation, and so their conclusion is now let's hire somebody with a marketing background. Makes total sense. However, from a momentum perspective, that is a disaster because they take the decision to start hiring someone, it'll easily take at least three months before they even have a good candidate,

let alone the person being onboarded. It's six months, nine months until that demand generation actually gets started. Now with AI, we cannot hope that AI will actually completely replicate a person like that, but you can get started. You can start gathering data. You can start running experiments, so that when that person finally comes in, they already have, in a way, a lab of marketing assistants that has been doing experiments and they can just

start guiding and deciding there and hit the ground running from the outset. And more importantly, the team will have the feeling, we decide today from next week, our change is already happening. And that is the key change that we start seeing that also allows us toaccelerate startups faster and bring breakthroughs to the table much faster than before.

Doug (03:38)
That resonates a lot. I know a lot of clients we've worked with have had strengths in inbound, strengths in outbound, and wanted to build the other capability. And you're right, today with AI, to at least start building those strengths, you can do it immediately while you get the right talent pool around that.

Roland Siebelink (03:53)
And it'snot just about demand generation. We're not even a marketing agency or anything. Other examples are where a founder is completely dominating their team meetings and all we do is set up an agent that checks the transcript at the end of every team meeting and gives the founder some personal coaching. This was much better than last week. This is the part where you should still pay bit of attention. Assuming, of course, that they want to change themselves. That's where it always starts, right?

Doug (04:22)
Yes, of course. Let me put myself in the role of a founder of another organization. What's going to be different about this podcast and why am I going to want to tune in and spend 30 to 40 minutes of my drive or my commute or my time on the elliptical trainer to listen to this one versus the other great podcasts out there.

Roland Siebelink (04:41)
a few things. One, I think the large majority of AI-focused podcasts out there invite people who are vendors in the AI space. They have a vested interest in making it feel like AI can do everything and are not talking about the weak points or the extra drag that you may create with AI as opposed to just focusing on the breakthrough.

For us, I mentioned, AI is just a means to an end and only one means to an end. There's many other means too. But since it's so prevalent at this stage in Silicon Valley, we do feel it warrants a certain focus for a while. Andin the end, the real goal here is to get more breakthroughs for your startup to move much faster and to reach that next funding round or that unicorn status much faster than you could do before.

Doug (05:30)
Do you think this podcast is only for founders and key executives at startups?

Roland Siebelink (05:35)
No, it's not. We've always had with our previous podcasts a sizable audience of investors and all kinds of startup employees, wantrepreneurs, as I call them, people who want to be an entrepreneur and are not yet or have not taken the jump yet. And the entire startup ecosystem, both in Silicon Valley and outside, is part of our audience. And I think everyone has that same interest.

There are even some companies out there that might not refer to themselves as a startup or not as a startup anymore. Companies that arein a more traditional business such as financial services or even something hardware related like air conditioning installation. But we used to joke, Doug, we only serve software businesses. But then again, software is eating the world, so every business should be a software business.

And I think the same now goes for AI. Every single business should be at least aware of the changes that are happening. What are the companies that are most at the vanguard of this new evolution? Does that mean every company will get all the way there? No. But you should at least be aware of what are the changes that you can hope to achieve as a founder and where you cannot fall behind.

Doug (06:48)
I love the fact that you brought upair conditioning companies and those types of things because those will get disrupted, just like they got disrupted with the Internet. They couldn't advertise in the yellow pages going back 20 years. Now they have to be on Google. You really see this for technical and non-technical folks, founders and non-founders, but people who are worried about AI disrupting their business.

Roland Siebelink (07:11)
Yeah, and I think it's often translating what AI can do into an actual pain point that is the crux. To stick a little bit with those hardware service companies, whether it's air conditioning or they area contractor or something like that, I think we all know that the key for them to get more business is to be super responsive to when people call them. Now, of course, they're always on the job at the same time.

I think only those that have reached a little bit of scale have figured out to havemaybe their wife oran assistantpick up the phone and tell them to call back. But even there it's been a problem. Well, the specialist will need to call you back becauseI cannot tell you anything about what you can do or what we cannot do.When we look at the statistics from a platform like Thumbtack,

it's amazing how the companies that are the fastest at responding actually get the bulk of the business, and are even able to charge two or three times more just because they figured out how to be responsive. Now, is being responsive on your phone a core competence for a contractor? I don't think that's what they learn in trade school. But it's actually what drives the business the most. And so in this case,

those that have been experimenting with just having an AI agent pick up the phone and be able towalk the customer through some basicguidance and then make an appointment for a checkupare those that areat the beginning of the curve and are alreadyseeing a lot of growth in their business. Is AI the goal? No, it's not, but it helps to resolve a real pain point.

The other thing I was going to say is that many of the other podcasts about AI is engineers talking to other engineers. And so they go very deep into the use cases, particularlyin the underlying technology platforms, like where do I get the most efficient inference and all that stuff. Very interesting. I had a guest on my other podcast the other day that is completely specialized in that and just raised $107 million just on that.

But still, as a founder, you're probably more on the business side. You're not necessarily getting the most leverage out of knowing the full details of the technology. What you need to know is what are the benefits, the opportunities, but also the risks.

Doug (09:32)
When you think about the difference between a founder who's using AI or building something to be an AI product versus a breakthrough AI operator, how do yousee those as different things? Where do they overlap? What's that Venn diagram look like?

Roland Siebelink (09:48)
That's a very good question. I'm not sure that we can already make a very clear separation between both because even those founders I talk to in very traditional businesses, when I'm asking them, what are you working on, then their answer these days will often be, I'm completely revamping the entire organization around agents and I'm going to be doing a giant layoff and all that stuff - the bad news around AI sometimes.

What they actually get there is a very different story. But the vision is often super compelling to them. But I think the companies that have made the biggest mark in AI, the fastest growth inAI, ARR, if you will, companies such as Cursor and Claude and others are basically those that have been building the tools that a lot of people can use, especially in coding and related areas.

They, of course, use AI for a very core competence of the business and then also try to expand that core competence into every other area such as HR and marketing and all the other functions of a business. But I also think that many other businesses will still need to figure out where does AI fit? How do I balance the company between what humans do and what AI agents do? And how do I still

serve my customers the same way or even better without getting a risk of them being disappointed by some automation that went wrong?

Doug (11:20)
Can you give me an example of someone real,a company that I can actually picture what you're talking about?

Roland Siebelink (11:25)
The company I just talked to the other day is a financial services loan provider and imagine that their internal systems have been essentially manually driven where they would get leads from a broker and then some internal loan officer person would look at what are they asking for, can we provide it, what's the risk, all of that in Excel spreadsheets.

And then they come back with the proposal six weeks later or even longer because it takes so much time to process that.And after that, there's a lot of back and forth negotiation and seeing whether they're actually in the lead or not. Then afterwards, there's a whole servicing component, which is also super manual; making sure that the payments get back on time and adapting the interest rate based on the risk profile as they learn more about the customer. All of that,

I think I have always seen things that a lot of it could be automated, but the crux was in how do we actually translate that knowledge into deterministic code? And I think that's where a lot of people have been stumbling because they were not able to abstract the knowledge of those loan officers into code that is deterministic and reliable. Now with AI agents, you can just start playing with it and saying, I have this proposal, here are some input, what do you think? And in the beginning, the results will be very bad, just like with ChatGPT or any of the other consumer-grade tools that we use. But over time, if you're disciplined about saving the context and keeping the documents in place and giving the AI agents access to that context, then you can very quickly get to a point where the average performance is going to be better than the average loan officer performance, especially when you compare it to a junior person that is still learning.

Doug (13:26)
Right, and then it also enables people to learn from that, to review and correct mistakes, but can do a significant amount of the work for them. They're more strategically thinking about what's missing, not how do I generate this from scratch.

Roland Siebelink (13:40)
Exactly. And many people have to learn to actually see the value of their job in the more strategic, in the more exceptional, in the more guiding aspects of their job, as opposed to just doing the work, because that's the part that's quickly going away.Many of the more junior knowledge workers just gathering lots of resources and putting them into a report. That's the stuff that AI is super competitive in. That's not going to be a long-term job prospect for most people. But making sure that it's not just at that intern level and that there's actually some thinking behind it and that there's some critical assessment of a first draft into well, why are you writing this? Does that not seem inconsistent with each other? That is actually where human thinking comes in. And this is also why the subtitle of our podcast is the 50% startup. Because we don't believe that everything is going to be just AI agents. We feel that there's always going to be room for people, thinking, pattern recognition, the things that AI agents are not that great at, but where people can actually bring in their experience and their human judgment in a much better way.

Valuation is also a great use case. And I think investment banks are already applying a lot of AI as well,because just plugging numbers into a spreadsheet is something that not just junior analysts can do. This is actually the kind of routine job where people used to be taught very procedural approaches. Here's your standard spreadsheet.

Here's the company information. Now you dig up which numbers you have to get out of where and put them in the spreadsheet and produce the valuation. That kind of non-structured work is something that AI is actually quite good at because it doesn't actually require much judgment. It only requires being able to interpret synonyms and different semantics, which only humans could do in the past, but now AI agents are perfect for that.

Doug (15:40)
I think about the AI company that we took public more than 10 years ago, and so much of the time was spent structuring the data properly so the AI of that time could consume it and use it reliably. And that has just completely changed over the last 24 months.

Roland Siebelink (15:45)
Yes, absolutely. That's where the revolution of large language models has really made a difference. I also remember 10, 15 years ago that the very act of placing advertising in the right place at the right time was not actually that difficult for even that generation of AI agents to accomplish.

Doug (16:21)
Let's go back to the listeners of the podcast.I know that we're going to have a variety of people on. I will not give any previews of the guest list because it's very exciting. But what do you want people to walk away from this, where they're not going to get that from an AI newsletter or an AI substack?

Roland Siebelink (16:33)
Yeah. I think the balance that it's actually something that can help your company move forward, but it's also a challenge and it's something that you have to manage and that you have to lead. One thing that we see in a lot of startups is where the founder is street lengths ahead of their team in the application of AI tools. They may have subscription to five different AI tools from ChatGPT and Claude and Code and...the Gemini and then they'll have Perplexity on top and then one or two graphical. We can keep mentioning brand names, of course, but then they are frustrated. They feel so much more productive, but their team doesn't seem to be taking up any of their suggestions, isn't actually using AI tools all that much, or at least they don't see that they do. And so we see a lot of change management actually being necessary in...

one, the founder realizing that just making themselves more productive with AI is not necessarily going to move the needle if the whole point is that they're sometimes too overbearing or basically the chief chaos officer in their company. And then second, thatit requires trust and to some degree also freedom for people to figure out for themselves how to use AI for their jobs as opposed to

the founder wanting to prescribe to people in the company, this is how I want you to use AI. That's taking away the entire benefit of AI if it comes to them as a system that they can use just like that standard spreadsheet they would get at Goldman Sachs before.

Doug (18:20)
Yeah, it's interesting because a lot of companies we work with and companies that we know of and have friends at, they're doing a very bottoms-up and top-down approach. They're solving big things from the top down, but then encouraging curiosity and innovation at the lowest levels. By the way, I realized that a lot of people tuning in for this podcast may not know who you are. How the heck did you get here? And why are you on this podcast speaking out of whatever device that people are tuning in from?

Roland Siebelink (18:28)
Okay, well, let me start with my name is Roland Siebelink, and I have a funny accent because I was born in the Netherlands and then lived in Belgium and Switzerland and now I've been in California for 16 years. I'm a nomad and I love traveling and I also have had the benefit of experience of helping to scale three unicorns in three different countries in the course of my career. Every time, we grew a company from roughly 10 to roughly 1,000 people in three years time, typically went to an IPO or something similar. And I think you and I are probably some of the few people in the whole world, Doug, that have actually seen that journey from the inside several times.

Every single founder trying to scale their company, trying to accelerate their startup, is typically going through that for the very first time. And so they are not aware of the caveats, of all the traps that are out there, and also, what are going to be the actual success factors over time. I found myself getting more and more demand of people asking, can you help with that because you've been through it three times?

And so, about 10 years ago, I started MidStage Accelerator, which is an accelerator, but it's really focused on that mid-stage of a startup. Not the very early stage like other accelerators, but we're really focused on that area of you've reached some product-market-fit, but basically everything is still a mess inside and it's not very organized. And now how do we get to a company that doesn't have just product-market-fit, but actually reaches product-market-dominance?

Doug (20:29)
Very well said. And this is not a competition, but Roland's had three, I've had four. Doesn't mean anything. It's not a point of contention whatsoever between us ever or all the time. Yes, exactly. That's a good response. We'll move forward. After your last unicorn, why didn't you stay in the corporate world? What compelled you to help founders?

Roland Siebelink (20:39)
I'll just smile at that. For two reasons. One, I just got a lot of demand from people saying, "Hey, can you come do a workshop with us? Can you do an offsite?" And at my last Unicorn, I was already doing internal offsites and saw the benefit of teams actually learning those best practices for scaling a business unit in that case, much more rapidly. It really felt like a body of best practices that wasn't very well known yet and where there was a lot of demand. But I also found that trying to do this inside one company, I was sometimes maybe a little bit too impatient. I wanted to see so much change at one go, and of course, companies have only a limited capacity to absorb change. It was for my psychological health and mental health a lot better to have 10 clients where I was trying to push the needle and see some needle being pushed a little bit every week.

Doug (21:49)
I find this, and I want to find out from you to what extent this is true, I also find when we're working with clients, being a little arms distance away from the day-to-day internal, every minutia of operation helps us see the forest through the trees a little bit more clearly. Do you want to talk a little bit about thatline of sight?

Roland Siebelink (22:08)
Yeah, I think that's a really good point. We have been in startups where basically we were fixers. We would be doing anything it takes or anything it took to basically bring the startup through whatever problem they were facing. We were producing breakthroughs on a day-by-day basis, most likely. But in retrospect, I think we also have to admit when you're in...

the thick of things that way that many of the things that feel like breakthroughs at the moment don't actually move the needle all that much over time. And so, I think the value of having somebody with a bit more external perspective, one that is a little bit more distant from the tactical problems that you're facing or the fight you just had in your leadership team meeting, but on the other hand, also that has seen many, many more companies from the inside than any of the founders or leadership team members could have hoped for, helps to set that perspective. And we always tend to say with the clients we work with, startups we work with,you guys have all the answers - guys and girls, of course - you have all the answers.What you may not have is the right question to be asked. And that's what we see all the time, that people are very focused on solving what they think is the actual problem. But if you just scratch the surface a little bit, you find out actually there's a much deeper problem underneath.

And as external accelerators, we can often see that faster than people who are in the thick of things. And that's, I think why still there's lots and lots of people out there that suggest they can do startup coaching and facilitation and all that stuff. And many of them are doing a great job. But I would also say the biggest competitor we still have or the biggest substitute we still have is just teams trying to do everything on the inside and everything doing yourself. And they spent so many cycles, Doug, I'm sure you've seen it as well,months, even years sometimes trying to resolve certain problems that they think are holding them back, not reaching that breakthrough. And the moment they just ask experienced people to come in and help them facilitate a workshop, one, the founder is no longer the person who has to push everything. They can just sit back, observe their team, see the real consensus emerge. And second, very often, it's that external experience that frames it in a slightly different way, still often around that same problem area that the founder was worried about. But in such a framing that suddenly it's much easier to reach consensus with the team and not have the founder to be the one trying to push everything and then struggling with that inertia that every team builds up over time.

Doug (24:58)
I want you to think back to when you were a child and what was one of the first things you did as a kid that you think resonates strongly with what you're doing today.

Roland Siebelink (25:01)
I think I was always trying to fix things. There was something wrong in the school, for example, and I would say, "Why don't we organize it this way?" I think I was always, even as a little kid, I was seeing things that were just structurally not very well organized,even putting a desk in a different place. I happened to be reading quite early. As a four-year-old, I was able to read some of the books already.

While the other kids in school were still learning to read,the teacher just told me,you can just take whatever book from the shelf and just read whichever you like, because it's a little bit before - the others still have to catch up, of course.And before you know it, I had reorganized the entire shelf and I think color coded the books or something like that to make it easier to find them. And I think that urge to get it better organized, get it better structured is still something I see in myself every day.

Doug (26:15)
OK, so you've been doing this forever. That's fair, and I will tell you 100%, that resonates as 100 % true. Makes a lot more sense now. I guess the last question: Someone's listening and they're in and they want to get more. What should they do?

Roland Siebelink (26:17)
Well, the best way to learn more is to book a Breakthrough Workshop with us. We have a very simple offer. It'sone you can reserve your date right away online. It's at breakthrough.midstage.ac and it explains a little bit what we do. We do the full interviews with your team before. We understand, of course, what the founder wants to achieve, but we also make sure that the founder can sit back in that workshop

and actually guarantee that that breakthrough is there at the end of the workshop. And if you like what we do and you like that workshop, then typically customers want more and more breakthroughs, of course, and they strive for what we call our quarterly breakthrough system.And typically, startups stay with us for two or three years, but there's no commitment, no obligation to do so. Basically, just start with bringing us in and do a simple breakthrough kickoff workshop and see how you like it.

Doug (27:29)
I love that. That's great. The other thing I would just say, if you're not quite ready to do that, I have gotten some feedback from some listeners that because sometimes our guests break into demos, so sometimes the videos are much better than the audio. If you're an audio podcaster, please subscribe and stay there. But we do find that the YouTube subscriptions are a little bit better if we have some good visuals and you actually see who's talking there. Please all check that out.

Roland, this was really fun. Let's not wait until the 300th episode to do this again.

Roland Siebelink (28:03)
Well, who knows by the time we have episode 300 what the new trend is that we'll have to jump on, of course. But I'm pretty sure that AI will still be a major lever even at that time. Thank you for interviewing me on this. This has been amazing and I'm really looking forward to having all the new breakthrough AI operators on our podcast.

Doug (28:23)
Awesome.

‍

The AI Breakthrough Stories Most Founders Are Not Telling | EP 200

Show Notes

Transcript