Rise of AI Virtual Chat 2020 · Berlin

How to Start an AI-First Company

Erik Pfannmöller · CEO & Founder, Solvemate

At a Rise of AI Virtual Chat in 2020, Erik Pfannmöller, CEO and founder of Solvemate, discusses how to start an AI-first company with host Fabian Westerheide. Pfannmöller describes moving from a background as a competitive athlete and founder of an e-commerce company to spending a full year in 2014-15 self-teaching machine learning, from reading Peter Norvig's textbook to attending meetups, before an idea emerged from helping his father with an email problem. He explains Solvemate's path from an Excel-based concept to a viral Facebook Messenger animal-guessing game that proved the underlying algorithm worked, calling this step "algorithmic fit," which he argues should precede product-market fit. He describes it taking one and a half years to launch a first product and two years to reach a first paying customer, and advises raising three times the funding founders think they need. He argues the label "AI-first" is largely irrelevant compared with whether a company solves a real business problem, and describes Solvemate's own approach of combining rule-based dynamic decision trees with limited natural language processing.

Key takeaways

  • Pfannmöller spent about a year self-teaching machine learning, including reading Peter Norvig's textbook, before founding Solvemate.
  • He calls proving an algorithm works before building a product "algorithmic fit," distinct from product-market fit.
  • It took Solvemate one and a half years to launch a first product and two years to land a first paying customer.
  • He advises founders to raise three times the funding they initially think an AI product will require.
  • He estimates AI-driven customer service automation had penetrated only about 0.5 percent of its potential market.

Detailed notes

From competitive sports and e-commerce to AI

Pfannmöller describes a background as a twelve-time German and world champion athlete before studying business and building an e-commerce shopping-club company that reached over 20 million euros in funding, 70 employees and 20 million in revenue before he sold it. He describes his path through what he calls "quantum leaps," from athlete to entrepreneur, driven by an underlying interest in mathematics, statistics and IT. After selling the e-commerce company, he decided any new venture had to combine IT, mathematics and business, which led him to consider either Bitcoin or machine learning around 2014, ultimately choosing machine learning.

A year of self-taught research before founding Solvemate

Pfannmöller says he spent about a year deep-diving into machine learning before starting a company, reading a thousand-page math textbook by Peter Norvig, attending every relevant meetup, and meeting people for lunch to understand which data feeds which algorithms, from NLP to self-driving cars. He argues this research phase was necessary to build enough confidence to found an AI company, comparing it to needing to understand animals before running a zoo. The idea for Solvemate emerged when his father called saying he could not send emails; while diagnosing the problem, Pfannmöller recognized he was applying decision-tree logic he had just read about in a scientific paper.

Algorithmic fit before product-market fit

Pfannmöller describes a concept he calls "algorithmic fit," arguing that before a SaaS company can pursue product-market fit, it needs proof the underlying algorithm works, which Solvemate first modeled in Excel. In 2016, when Facebook opened its Messenger platform to chatbots, Solvemate built a viral guessing-animals game to test their intent-recognition technology, reaching over 85 percent accuracy and letting users train new animals into the system. He calls this the moment they had proven their algorithm before building a product on top of it.

What it actually takes to start an AI company

Pfannmöller says he immediately brought on a technical co-founder as CTO rather than attempting to be both CEO and CTO, and argues founders need at least a fair understanding of the underlying technology even if not coding it themselves. He says it took one and a half years from starting research to launching a first product, and two years total to land a first paying customer, warning there was no existing framework like TensorFlow to shortcut the process for their specific use case. His central funding advice is to raise three times the amount founders initially estimate they will need, since AI development always takes longer and costs more than expected.

Is "AI-first" a meaningful label?

Pfannmöller argues that whether a company is called "AI-first" is largely irrelevant, comparing it to asking in 2003 whether a company "uses the internet"; what matters is whether removing the underlying mathematical models breaks the product. He describes Solvemate's approach as combining rule-based dynamic decision trees with natural language processing added back only in small parts, a pragmatic choice he says let them launch faster than competitors betting fully on NLP, which he calls "still tricky." On competition, he estimates AI-driven customer service automation has penetrated only about 0.5 percent of the market, so the biggest competitive threat is companies not acting at all, not other AI vendors.

Q&A highlights

Asked whether a DeepMind-style AI research company could be built in Germany, Pfannmöller says research companies are effectively nonprofit cost centers funded by grants or state and corporate sponsorship rather than standalone businesses. Asked which industries benefit from the Corona crisis, he cites e-commerce, SaaS and automation, and predicts virtual conferences will keep thriving after the crisis. Asked about his biggest challenges, he cites generating enough personal energy as a founder and wishing Solvemate had moved faster with more strategic foresight. He advises technically trained founders to study business, fundraising, HR and company strategy, and describes Solvemate's "marble rule," where late arrivals drop a red marble and on-time attendees a green one at team meetings, as an example of using simple processes to change team behavior.

Quotes

You need to have more money than you think it takes.

Erik Pfannmöller

It's not about the AI, it's about… a technology that enables companies to thrive.

Erik Pfannmöller

The biggest fear is not the biggest competitor, it's people are not acting.

Erik Pfannmöller

Questions answered in this talk

What does Erik Pfannmöller mean by "algorithmic fit"?

According to Erik Pfannmöller at a Rise of AI Virtual Chat in 2020, algorithmic fit means proving an underlying algorithm actually works, for example through a simple prototype or viral test, before pursuing product-market fit for an AI product.

How much funding does an AI startup need compared to initial estimates?

Erik Pfannmöller advises raising about three times the amount founders initially believe they will need, since AI development consistently takes longer and costs more than expected.

Can a DeepMind-style AI research company be built in Germany?

Erik Pfannmöller argues such research organizations are effectively nonprofit cost centers funded by state grants, universities or corporate sponsorship rather than standalone profitable businesses, making the chances of building one as a company very low.

Mentioned

a"": ""According to Erik Pfannmöller at a Rise of AI Virtual Chat in 2020

Keep exploring

More from Rise of AI Virtual Chat 2020

More on Enterprise AI & Adoption

Transcript

Auto-generated transcript, not edited. · 11,281 words

Show the full transcript

welcome to the rise of our virtual chat series - Herman Vesta Hyde I'm an intrapreneur an investor with a passion for artificial intelligence and as every week I'm inviting one interesting personality from the global air industry to my show and in this week my guest was Erich von Miller CEO of the AI company soulmate and he is a business engineer by background and we talked about the topic how to start any I company so if you're interested into the Entrepreneurship or artificial intelligence you want to start your own company stay online watch this video and enjoy good morning we are life and I say hello to everyone out there this is our second edition of the virtual chats format and if you have time you can try to read this it's one of my favorite t-shirts and each week I have a new shirt for you always in the context of a I um our guest today is the elegant I will introduce in him a couple of seconds um he's in an in the

green room he will join us live when we are ready so first I would say please introduce yourself you see Chad to send and you link and contact your website don't do a lot of advertising don't was four or five times everyone sees it you know when you do it once a please use it at your Twitter link and whatever you know we have a handful of people here live currently and everyone else is getting the chat and will log in later this will be live or this will be recorded and this will be able to be watched over and over again we will also try to cut this to a videocast and maybe make a podcast out of this and distribute so the content can be accessible for everyone well um yesterday was not the best day for us I would say you might know or not we organize every year at the Rise of AI conference and recipe I started as a community I remember really well at 2014 started a meetup group

and this reminds me by the way a lot of this for me today it started as a community event very small ten people came with a great evening of a conversation then with the first official meet up with 50 people than 150 people and then we started the rest of my contrast we had 350 people 600 people 800 people and this year it was planned for almost 2,000 people the Germany I week plenty of things and we all know why we're sitting here we're sitting here because Quranic aims something we didn't expect live events at that everything physical is dead we all switch to virtual I think the truth is great I think virtual has a lot of potential and we will extend this and we'll learn each time so that's why I have a new screen for you now right and just it just was a painful moment yesterday because yesterday we official cancelled rise of AI we got the the the the legal paperwork from the government that can't do it and that

was needed and necessary and yesterday we cancelled but that means we are now here and today my guess is Alec and with Eric I will speak about the topic of how to build an AI first company so I would like now to introduce Eric before his coming life and first I invited a link because he's not open or he has a best of experience he's a sports man and he's a friend he's such a friend that he was part of our wedding three years ago or don't ask me at three years ago yes my wife is gonna check to you so three years ago he was there with his wife I know his kids and as a safe and no kids from someone this is a benchmark for how close a relationship is currently so to start about the Eric he was a sports man in his youth and nerd like me has a website with his old pictures and I was allowed to take one of those and he describes his progress really well

so he started very young his family stirring so I think his brother is doing canoeing tomb and he has all those pictures in all his careers and he's describing each year how his season was and he's like with 18 you brought all I entered the the male competition this year to seniors and was tough and it was a struggle but I'm getting better and I will improve and this reminded me a lot about intrapreneurship it's the same you start small you improve you get feedback you fight I think this is really interesting I will ask him later about this what he learned from being a sports man an athlete to being entrepreneur and by the way he was successful with this I mean he was twelve times German champion um he was world champion I assume in his class in age group but he was a world champion and his family they even have Olympic medals so successful family he was successful that means he did it right he did it serious he did it

proper and this is always for me also from investor perspective an indication when someone was really passionate about something as a kid he might be as a person too and then the usual he studied later business the high ed in like six and strategic management he started and I call it equal Merce company for sports so the topic was there because it was in the 2010 2012 it was shopping club as e-commerce he grew it he had his apples outs you know you can read in the news about firing people hiring people down loans at Browns at the end he had over 20 million in funding over 70 employees official members from crime face 20 million in revenues so he commerce 20 million revenues no fail no exercise except proper thing and he sold it became a venture partner in the fund of his brother called venture stories and a couple of other wealthy people from the Munich area had a small break and started the next thing and the next thing was I think

it was 2015 he started soul mate and then we got to know in person 2015 because I don't know he caught it I think he contacted me and I believe we knew each other from this sense we just met because in 2015 very few people were thinking about the I and even less people were investing in the I I'd like I didn't ask her and less people were starting a company so he started early before there was a trend he saw it early I will ask him why and also just to mention Eric is hosting the a I unplugged podcast also podcast in German um you can download and listen to I was a guest in a show so today he's a guest in my show and that's how we welcome him now let me try to bring him online first the video and then the audio I'm going to see you um lovely background Eric tell us show us more please yeah first of all I'm not in the office I'm in a home

office we are somewhere in the in the forest that we have a small cottage which we bought 2012 for the family because the kids shouldn't grow up and like a big metropolis of Berlin so I'm 70 kilometres south of Berlin in a small cottage and it's middle in the middle of the forest and I'm very lucky that last year I put like very fast internet here it's I would say please in action so first full of people life there ask questions if you have you can comment anything second we talked about this before that this is even a business model you know that you have cutters in the woods and one day when we have the money and the time will maybe run this full time you know as an intrapreneur next thing I love it being outside of the green so um what changed in your daily routine since you don't go to the office anymore and you are in your home office um so basically like the pool company sold made we are

fully remote since I think six weeks or so as we are a b2b SAS company and we like work on slack and and GG sweeeeet not a lot changed in my daily work routine so I'm not meeting people but I'm having a lot of zoom meetings and it works pretty well for us and my daily routine changed a lot I think the first two weeks like we took the whole family to like a super small cottage everything needs to be reorganized because kids are not in school and I wanted to work I built my own like a desk in like a super small home got a second screen set everything up but kids are running around here and I'm very happy that my wife is supporting us and can take care of our three kids they are 11 7 and 5 years and we also make our first steps in digital schooling the haircutter you know 60 millimeters that's what I could do myself and my wife is supporting me but you know here cutting

shop so I close this place so did you try it yourself I tried it myself and it's pretty easy to just take 16 millimeters it works but while my wife is supporting me I'm happy that she does I thought remove mark I tried it myself and on the back it's terrible because I don't see it yes and it's interesting some skills we have to learn even checking out YouTube tutorials how to make you hear and background I quickly wanted to say how we met each other back in 2015 because I was about to found soulmate and we look for the first seed rod we're gonna talk about this later and you're one of the very few people in the Berlin ecosystem investing into AI topics with your escort we see and we got to talking them that's how we met because I wrote you a message on LinkedIn hey I'm about to fund in our company you look to be interested in this topic let's Emma Channel and to say honestly if you did this

in 2015 you got the attention if you do this today you're one of hundreds of people so the whole market exploded and we both out there early in and I think we are destined to talk about running any economy that's what we focus today entrepreneurship and AI first company um so you did equals before the jump from e-commerce to AI is similar like my John I did wedding dresses before I was selling wedding dresses and then I did any I a couple of AI things afterwards so it's a pretty harsh job and not logical how did this come about this emerge for you now um I think there is like this concept of quantum leads and I think my first quantum leap was from being a sportsman to being like an entrepreneur in the first place in the first place so that was the first one and then kind of I strive for my quantum leaps when changing when changing circumstances and it looks like I was a sportsman that turns to be a businessman

and a founder but there is an underlying theme and the underlying theme is I always scored best in statistics and math I was about to screw my first PC together a th of 12 which was like this now with the spinning disks back then and I'm a very very mathematical like an IT nerd I really like iti would love to be a program ever wouldn't be a business person so um after after handing over this e-commerce company to my management which was technical already I figured out that if I'm ever gonna found a company again it needs to like really fit my DNA and my DNA is processes as IT is mathematics combined it with business skills and you kind of figure out that this is either Bitcoin or machine learning back in 2015 or 2014 did you buy some because no I didn't I did not but I really deep dive did with VidCon technology back then in 2014 because for me it was very clear to to kind of you know you need

to have an underlying technology that did like disrupt something and this underlying engineering technology combined with business skills opens startup or like business ideas and so I really deep dive into Bitcoin but and I decided against Bitcoin back then and I never bought some because I kind of tend to be a black-and-white person and I'm just fully focused on my company's ultimate so with crypto or blockchain technology and AI you had both translator and I said until today they are the most valid trends as you could say it's fuzzy and DNA modification but we both don't have any biological backgrounds that makes a lot of sense and will come to this later but your company is successful it is growing so yes survived and you have the merch and you're running the typical scaling phases from product and cetera what were what were your sentiment when you start 2015 that's seeing where you today five years like what were your sumption sandwich of assumptions you've got right and which did you say too early too

late or you miss mmm-hmm I would start one year earlier so mid of 2014 am I needed to decide whether I want to be a founder again and I kind of first decided there needs to be a SAS company and it should leverage some kind of bit kind of machine learning technology but I didn't know anything about this engineering skill so I took me a year to research everything I could about machine learning literally from reading a thousand page math book from Peter Norvig and people interested in this know this book I'm I attended every meet up I met a lot of people for lunch and I really wanted to understand which data are crunched by which algorithm to a to a trained model and then you you insert real time data and make predictions that's that's machine learning that's AI it's always the same but I wanted to understand which kind of data feed NLP algorithms which kind of later feed self-driving cars how does the weather car forecasts work and once you've understood

this data model and this algorithm crunching you kind of start to understand the state of the world at what is possible without rhythms and what is not and like then like there was this moment after like nine months of research and I had a lot of ideas and I always push the business ideas to the back because I need you to research first there was this moment when you go to a medium and you think well I've heard about recurring convolutional neural networks I know what you guys are talking about again I cannot program this but I understood how it conceptually works and there was this moment when you thought okay I have understood what is possible right now and then then it's about taking back your business ideas and then as typical there is this sparking moment when I got a call from my father saying son I can't send emails anymore and I was his service agent and I was about to kind of you know asking these diagnosis of questions can you

send him a yesterday which device are we talking about how big is the email we online yesterday and these things and I figured out this issue because I had expert knowledge and and then there was this moment while I was reading a scientific paper about decision tree kind like logic and then I figured out well that was exactly what I was doing I was following a certain logic to find out his root cause and that was the moment where the idea of solve med was born and five years later we have this product and we saw millions of customer service requests automatically via chat pods so I sum it up you had the technical interest you can dig into the market you met people you did what we do you know you go to meetups conferences you're you read a lot of scientific stuff until you come to the point where you have heard it all before and it's starting that you switch from learner to using you need to have a certain level of

confidence in AI to found an AI company and that was like I think it's a central theme of this of this talk you need to understand the basics like for example I could never run a zoo without knowing anything about animals right so I need to study like I need to know how a zoo runs and I need to know how to run a suit from the business side so I had no idea about this and if you ever want to found in AI company you need to [ __ ] know what you're talking about it like this this Neal a year research really helps me even now to understand more you need to know what you're talking about so it is it is industry experience that you need to have to find a company for me I had a I had a company in e-commerce we had a call center we had customer service we had our all I know what customer service is about you need to know the underlying technology like this engineering skill

we if you manufacture things you need to know how to manufacture metal or so I need to know how to manufacture was one develop engineer algorithms and then because we had an e-commerce company we knew about tech and how to build like a good web content in these things and lately as a father you need to have the business skills and understand like these which also had the experience with my prior company going to business school and all of this together makes kind of an AI company but you cannot found an air company as a business person saying I want to apply AI because you need to understand what you want to do because the product is deeply ingrained and math and algorithms and I recall it right you started writing the first code in Excel oh yeah we did pen MVP so it's about MVPs and I think in an AI company or every SAS company it has like this product market fit and then you go to market I think there is something

more that is called an algorithmic fit so before you have a product that you can sell you need to have an algorithm and for us even before that algorithmic fit there was okay I have a concept and you need to visualize the concept and you need to kind of try this out and yes we sell in Excel because we could model is really easy and then we very fast transition to like building this into a product and trying these things out so to sum it to sum it up if you want to start an ad company you need an MVP algorithm so you don't need a product to show you need an algorithm to work second you should know the market which I agree and you should have a fair understanding of the technology because the main question is as a business person by education can you start in our company you are proof you can and I know other founders in my portfolio have a business degree to formally and started successfully at Collins

yeah first of all I agree that that there is two things first of all I am NOT a one-man show so there is a co-founders and the first thing I did is like at a CTO because like I am I have the business skills and we have a CTO and he's a genius building the product at the architecture and managing the development so I think you always need a technical counterpart because I'm not a like being a being a CEO and a CTO that's very special so of course you need a co-founder and my co-founder young is great in doing so so that was one of the first things like we raised money and we started the company with like five people but of course you need like as a as a as a business founder because every company it's a business part in a tech part it's in a high company you need to understand the technical basics and so let's say if that year you you invested your time in digging into the

market when you started the company was it strapped legally or did you already have cash then like in the center um it was kind of wood scrap because I was in a small sue turnaround office somewhere in Berlin doing the research and then I met the first colleague on a meet-up and he said well I'm a I'm a student I want to go into an AI company program something and I said well I'm about to find a company so why don't you just work for me for the first month and I buy your for this so that that's kind of how it literally started and so then we were like three to four month like hiring people we were a few people and then my co-founder joined super early on and then we raced out for a seed round and would like this mini MVP this mini algorithmic thing and from then on we supplemented an office like just five people and like this this family term were like your business people and five develop

and you tried to build this algorithm and try to understand how this works what you conceptualized before and that's I think that was 2016 was the year which we call algorithmic fit it was back then when Mark Zuckerberg said in April 2016 you want a message a business the way you want a message a friend and they opened Facebook for a chatbots in what's called or yeah yet there was two things first they opened the platform on Facebook and then they launched their own chat but M would stay by the way shut down because it's abhorrent to do I think the technology that they purchased with today I and an NLP platform where you can train intends and like it's it's a basic underlying platform for NLP processing and NLP intent recognition and on top of that you can build you can build BOTS it's like the first wave of bots and and and these things so Facebook was open and we built a fun game that is called guessing guessing guessing animals you might

know the marking game where you put something on a for it you need to guess the person in who you are that's kind of what we do kind of we find out the intent customer service it's about you have one out of Hondas then it requests and hundreds of thousands of people are calling age here so it's about finding your request and letting to execute so the core of our product finds the customer intent super fast and then lets the cust lets the end consumer execute it in real time so we were building this super fund but not about guessing persons about guessing animals we launched a chat bot like a week after because we've been working on this before we worked on chats words before they came to Facebook so we launched it invent viral and tens of thousands of people played the guessing animals game like they answer questions how much legs is it a mammal does it effort as a black does it have a trunk is gray is it big yes

it's the elephant our algorithms solve that problem and we could even handle wrong answers of you and users and we could train new animals into the system so in the end in 2016 we had like hundreds of animals trained into our system and our system gathered worked with more than 85% and that's what we said well we managed to not only have the the experience of finding the animal we could also train the system to add new animals into the system like adding new intense adding new customer service request types and so we said we have nailed this and that was the moment of end of 2016 where the next stage it was about because then it was about building a product on top so you started your technology testing it for free with in-game I would say we didn't plan that to happen because you don't like we didn't know that Facebook was opening that we just had fun we launched a Facebook Messenger but we were because we wanted to test the technology

and it went viral like because it was fun to play you haven't I think it's not life anymore because they took it off but like it's a combination of we needed to test and we wanted to test it for ourselves because we want to have a public messenger bot and then it went viral so we had a lot of fun doing that in that time I still remember the days where all the numbers are growing the numbers are growing and so yes that was our algorithmic proof that we could manage to find animals okay okay so in between I will get the first user question I am and this is from Sebastian and Sebastian asked would you think it's possible to build not only in a first company but in AI research company like deep mind in Germany so something product focus and research focused what do you think about this let's say something like deep mind here in Germany hmm first we need to understand the business model of a research company and how

this works because founding a company always needs to hide there needs to be a business behind deep-lined when I recall right is a subsidiary of Google and they are funded by this the point is are they across center or are they have an actual revenue model I don't know this because I'm not an expert but research companies you do yes it's a huge constant haha I think the last numbers goes like 500 million cost a few - million in revenue consulting is huge huge cost I would assume so - so the question is is it a company or is it just a project or something like a strategic research initiative and founding a research company needs to funding because on research that's like when we do AI algorithm research we don't know the outcome that means we don't know what's gonna happen it's it's like like pharmaceuticals you need to invest to find out something but you don't know the business result first so is it a business I don't know what research is a

business because even if you research like vaccinations for over 19 you need to know that you're selling these vaccinations later so research is not a company it's kind of a to do and the question is how is it funded it typically is funded on the state on universities on state grants on European government grants so of course you can do so but who's gonna fund 500 million or like a hundred million or so a month I fully agree so there are business models working it's the Deaf cutting the German artificial intelligence Research Institute we have some who find Max Planck we're doing AI research - you have a lot of universities all of them are partially funded by governments of substantial funded and then they get grants they have to write proposals or they get a plan they get paid by corporations to research something but all of this is mostly nonprofit and they don't have a profit so it's not a business model so the answer for you is the best young the chances

are very low it's unless you have as he knows I'm going to give them like a lot of money to start this like you get the first money get the personal in your high are super smart people and the next step you apply for the next grant with your research results and then you maybe get a get a get a sponsor from the economy so you can build a research company but it's more a nonprofit and I think it needs to start with a big research grant because this comes to the topic where say you can't do strip AI because as I said AI is a lot of experimenting you need data you need process you need resources infrastructure and you don't know the outcome right away you need to tweak it quite often so I would say this is a challenge clearly still because there's so much abundance of open source data and open source tools your perspective on this I mean you you adequacy round and if you if you build a b2c

app for the iPhone you can't book this alone in days or weeks and you can test the MVP and you can iterate in research and for us it took around one and a half years to launch our first product to get the first unpaid better customer but not only half years Wow let me repeat that it was one and a half years from like the first people started to we got the seed round we got the money we had this algorithmic fit we launched this thing and then kind of we figured out what the product would look like because after having a doubt rhythmic fit we started thinking about who would buy this thing and what would be our go-to-market strategy and let's build a sauce product on top that people can build their own chat pods to automate their own customer service to feed their service teams so the point is you need to fund that upfront because there was no revenue even when we launched and then it took another six months until

we had the first paying customer and so two years down the road for us it was two years and of course this like we did something completely from scratch and so and there is no Google tensorflow thing that can help us because that is not a standardized case so we need to invest and if you do a b2b ai company even if you see a company you can have an idea of a product but you don't know if the algorithm is gonna deliver in the end you just don't know that and if you don't know that and we come to this later and one of my key learnings you asked that a few minutes ago is you need to have more money than you think it takes and so I asked like I had this concept in mind and identification investors yeah yes I asked a few people how much like person will develop a month do you think it's going to take to develop this kind of product that we can make some revenue

because that's what you need to have the race and X funding and people said X I thought okay I'm gonna have two weeks because it takes longer and it always takes longer always costs more money if I should do it again I would raise 3x of the amount that I believe it will take because it's research in the beginning so I have a handful of AI companies in the portfolio all of them are AI b2b not all of them are SAS honestly half not know they it's very hard to build SAS again and it's project based and implementation features and then gives us in Pinterest I would rank them let's say degree of a Imus your company's significant lower than others III know it but what you did right is you solve the problem which was very painful and you made it easy accessible without over-engineering without selling them this is the best day of the world this is solving a problem and describing it for my world what solve meet us it's not only

check bother I I learned it the way one of their clients they're selling Bluetooth speakers I think there was one of your examples of Bluetooth speakers and then what two hundred thousand calls a year because people weren't able to connect a bluetooth speaker to the Wi-Fi or something else and with your solution you were able to reduce this what about 50 60 70 percent like reducing the call center and the emails to a self self providing intelligent chat bot which is giving me the problem in eighty or ninety percent confidence within a couple of minutes and being a user myself if if and Bob can solve it within the one or two minutes wonderful I don't have to pick up the phone no one likes to pick up the phone we live in an economy like in an instant economy and of instant gratification we have spotify netflix and everything on demand so you don't want to wait in line to get someone on the phone you don't want to send an email away two

days for an answer you want to solve your customer service requests in real time today we are solving millions of customer service requests from various industries from e-commerce to tech companies to SAS companies to banks and banks can configure the check but in in a SAS environment take it live within days and then like with the chance of more than 80 sent people find what they want and do not need to call which means happy and consumer but also happy agent because I'll talk to service agents they don't want to answer the 50th time a day the same stupid question like hey can you send me a copy of my invoice this is what a bot can do I want to change my dress yes of course we can do that we can even fire scripts to execute processes inside a company so it's ultimately ultimate is H so yes something which is human has to do you're doing virtual just 10x faster and cheaper if you think about a chat BOTS like for customer

service or for like marketing more like sales but it's two things first find out what the customer wants as fast as possible that's what we owe and typically 23 seconds and we found out what they wanted then let them execute their desire on first touch that means given data legitimate the user find out what they want execute scripts or send over data sometimes it's only information that they look for but give it to them on the first touch in seconds on the go on mobile wherever they are that's what we don't care second question I want to come back I want to comment on one point what you said about the degree of a Ines I mean I think it's it's not right to call a company in AI first company or so it's just a tool that you use so and think about in 1995 you would have said to a company well your degree of using the internet like something like we are made in 2003 now with AI compared to the internet

time so you would never ask a company well are you using the internet anymore right now or like even 2010 it's like are you using AI and to what extent yes we use the internet yes we use algorithms and I think there is AI first companies that means if you turn off mathematic calculations with the product work so it's a rule-based or hard-coded and we are err first that means if we turn off nothing is going to work and the second question is what is the degree on technological advancement so are we using this most super fancy deep kind of additional recurring whatever networks and use buildings of data points like Tesla is using for self-driving cars no we don't and do we use fancy algorithms to do something super great with it yes yes so I believe it's totally irrelevant like you can do like deep vision video algorithms or you can do easy algorithms the point is you need to solve the business problems with the technology and the internet was a technology

and AI is a technology Bitcoin is a technology you just leverage a technology to build a product and that's what we always need to keep in mind it's not about the AI it's about the I that is a technology that enables companies to thrive to sum this up you AI is not depending how complicated to this it is if you use mathematical models and if you turn them off it doesn't work or it's still confused with them that it wasn't the essential second is and I see to solve a problem you don't need to sell complicated AI you can you know it's like under the hood it's like a fancy car and it's very good engine below but no one has seen this yeah truly just to keep on that one thing that we learned is b2b buyers they don't want to deal with AI they want to have a solution and we talked about like we change in the past two years our communication from we talked about like when I I was really

near people what are you doing which kind of algorithms people don't ask these questions anymore they ask about business results and if you are in any business department in any company or any person I don't care how things are working for example in my iPhone item care how ever produces these photo forecasts for me I just want to get the results out of that and hey I must work under the hood and that's important and then he's wonderful the question for Michael who asked he says AI companies are very resource-intensive to build and the real customer benefits of not visible right away it takes data in time to train AI system he would like to know from you if you have some advice how to convince clients or investor to jump on on board early even the technologies of fully at potential um I would say there was two parts on that question and the first one and I would like to answer ai is state of the art it's hard to vote that that's

why it takes resources time and and super smart people again let me compare to the internet 20 years ago and what's the same thing today you can build a web shop and a website in five minutes 20 years ago you couldn't it was recent it took time and nobody believed in what you're gonna do it they thought you were crazy the same thing applies here and I would like I'm using this analogy a lot to make people understand and yes it is resource intensive and it will be easier over time we have frameworks now in place we have like standardized libraries that we use we have like Python evil developed in the last year's field to be like a good language so things develop it will get easier it depends on the degree of complexity of your AI how much research-intensive it is which has set a few minutes before to answer the question about investors and there is nothing special in AI to to to convince investors you need to have like a rock

so I think a rock-solid team and background of the founders is still important and then the business idea that you want to solve needs to fit the market timing market timing is something really important are there people that believe in the product that you want to build in the near future so if you are too early in the market you're not going to get customers because you have not cross the chasm if you're too late in the market it's going to go to competition and a lot of people want to do the same market timing is important and the third one is I think a compelling like as I said founder team timing and then the compelling story of why are you going to solve that problem and to have this hypothesis of these data or crunch like this they're going to make this this model and this model will make good predictions that's like the if I would invest into AI in addition to team and timing I would ask for the algorithmic hypothesis

and to have proof around this that it's gonna happen and it's gonna work out the research so to add my perspective it's always team you need a great team then it's about a business model approach at the end you never know how you earn money it's just figuring out because you don't know what the client needs that the answer so believing in the team because and abilities and its market as I said it's a question of time as we've said Facebook and there were friends stirs and other networks before only one survived in emerge there were search engines for Google and then was there such as after Google still Google was in the right spot at the right time and sometimes it's a question of luck and timing or patience and for this you always need someone in your team who stays with you for a burnouts and second investors who are patients and even can give you more money that's tricky but if you today go and pitch AI to an investor you got

the attention and it's every investor out there is considering it that was not 15 yeah there was this moment of you know the internet 20 years ago that was this mobile first moment where everyone was about to launch apps and there is a timing point and I think the first hype is over for II I investments it's if you put AI to a pitch tech that's that's like not the best anymore it's about I want to solve a problem and I'll leverage a certain technology and it just makes sense it's I can you try and use every new company it should be data-driven and use kind of prediction to something great because the products themselves I wouldn't say they are built but the next level after like Internet and SAS and super cool products is like doing some great predictions and using data so let's talk about competition for a second because I did a study in 2017 mapping globally I ecosystem everywhere and the largest bulk of AI companies was in this sector of

communication mainly chat boards customer support everything I was invested in a company called Paulo mind who did problem-solving via email something like this first what I learned is NL piece not easy understanding language is something still tricky and in fast hoomans to humans interactions look so it's not that everything is finished out there second a lot of companies out there my thesis is because of a lot of especially brilliant entrepreneurs who weren't ecommerce or marketplaces alliteration for they knew the problem and they wanted to solve something they know themselves there are few people in Berlin or Silicon Valley working in engineering departments or at at an automotive company in new automotive problem or industry problems that's my assumptions so you you verified it a little bit how do you see competition like when you emerged you feel competition do you think there's competition is the market way too large for this um there was a lot of questions into one question I tried I'd have to break it down and the first one is yes

chatbots is a big topic it's a multi building not even trollee and think that's gonna change the society how we communicate um that because everyone knows that because it's so obvious that you want to have everything at your fingertips instant and virtual assistants that's how we communicate as humans and what Mark Zuckerberg said you want to message a business the way you want a message to friend that is just [ __ ] right just pull stop there was a ton of hype ton of companies emerged a lot of them already changed their business model or didn't exist anymore and there was an over promise there was a hype and chatbots did not deliver to that because of what you just said an app is just so hard and don't get me wrong and everything is great I am dictating my whatsapp messages I love any LP it's gonna help me Google is suggesting me when I type in gmail email and epi is great but it is not not so obvious to have a human-like conversation in

a multi-step thing and to hold context over time because we humans we learn with millions of words natural language my kids learn irony at the age of five and six and are really deep tucked into the NLP topic and we had solve mate we have a very different approach a very pragmatic one so we we do what we call dynamic decision trees and now after a few years we are adding back NLP on really small partial parts to our product so we came from a different part and that's why we can go live so easy so fast with our clients have so much regretted results and then basically have the best of both worlds in our chat pods and that's why they are working so great we had a [ __ ] like back then five years ago I found a contrarian truth and the contrarian tools you just set NLP is hard pull stuff so why don't we have a different approach to chat bots and again there is rule-based chat pods and there is like

deep learning free texture what's and we combine the thing because we have multiple choice and we are dynamic so it's a very special approach that works pretty well that's why it also took so much time to both in the beginning okay I need to answer that the second part I think the market is so early the biggest competition is not acting for companies because they just don't have they just haven't understood that like customer service automation is something that is is a pain for them so of course there's competitors there is people that just say they do good chat boards and people that have a lot of buzz around but the market is penetrated at maybe 0.5% so the biggest fear is not the biggest competitor is people are not acting like business miles and this is something good I always say hey I'm happy to search you to have only one percent penetration yes if you look at SMP and other companies nothing's running AI it is and this is really good too the

question from cloudian and he asked at the end he says he has a strong background in Al as a machine learning in AI and would like to start in a our company what does he need to know what business knowledge does he need to know to PC or CTO so for the technical people out there not coming from a business side from the technical side what is your advice and you can give them basically do the opposite of what I did I spent nearly your researching about machine learning spend a year and research everything you can about how to fund startups at around startups how to do funding rounds how to hire people when everything about HR law will learn about company strategy you don't need to attend a full Business School like I did but learn about this stuff because either if you're a CEO you need to have a ton of business skills because it's about strategy and funding and finances and HR and all these topics if you are a CTO you

need to have the same knowledge but just on a lower penetration you also need to know what company's strategy product strategy you also need to hire people into manage people so get your ass inside a room and watch a ton of Coursera courses like like I did about machine learning is the other way around we read for some of my block wlwt's referring to me I mean ten years about exactly this how do you hire people how do you find a business idea how do you find a team member a co-founder I have plenty of positive and negative experiences I had a corner was really let's say intense now I have my wife as a business partner which is a different level complication but what's really well what I learned is assemble a team first find people who share the vision and the idea and who are complimenting you so no don't put two business guys in a room and say we starting a company look for business people out there by the way there

are tons of smart business guys out there have no clue about technology and would love to have a smart CTO who can build them a product yeah I think this is a wonderful combination it's way better as a CTO like person to find a good sales people person as the other way around it's way harder to find technical people on there so then you clouded you have a wonderful start for this looking at the time we still have a couple of minutes so let's come to the learnings five years running any a company five years learning ecommerce company before what changed like what different skills did you need and what did you learn differences and with just summary like five years a I [ __ ] ecommerce ok that's interesting so that's a very tough question the first thing a degree of execution and I start again ecommerce was very execution driven it was something a we need to build something to grow fast we need to like a little central to marketing campaigns get more products

and like a lot of people are needed they need to be orchestrated that they are working well in an AI company need less people but this work needs to be really streamlined so we are really in project management into documentation into into I would say we are an extremely extremely organized company we like we have meeting serious we have confluence pages every meeting has notes we have we have we call a standard operating procedures about how to use like how to run meetings how to use calendar and writes you need to be extremely extremely digitally organized in running an AI company compared to and like a normal company of chips physical good city to be physically organized or the execution or the excellence is in the organization of your team and where you a need to have this mind set off whenever there is a recurrent problem there is a software for that so I'm scheduling appointments with the software like we we have a documentation we have a software for virtual on boardings all

the developers have like tons of tools to deliver to have continuous integration and it's about excellence on a virtual level I think that's what changed and I think it's mentally more challenging because you solution like that's the first thing the second argument in e-commerce the solution sometimes obvious so hey okay we're gonna do this yeah please solve it like this just like by this logistic Center get this done by this by these new goods and just build it the solutions in in SAS or in AI sauce of very harder to find so you need to you need to discover what customers want it's more deeply before you start even programming so customers don't know what they want and it's so hard to find out how the product should be that's like we spend much more time in planning our features and doing strategic market research I love this ecommerce physical process orientation may I company virtual course orientation at the end it's about processes just as an example when you visit the office of self

mate and you go to the large meeting room you have huge balls of marbles and what I like about he was reprogramming his team so what he did is incentive structure if you come too late you have to drop a dark marble in a box and if you have on a condo meeting a red one you get a clean one and visible one and if there is a marble in it it's not that your name in red one there's a punishment it's a collective punishment for everything and what I learned is I visit him let's say three years ago I saw the marbles this deck and the good ones this deck when it came two years later it was almost to zero so within two years he was reducing significantly the track little that people didn't come today anymore it was reprogramming them that they all showed up on time everything is about processes and it's like I really it really annoyed me that people were five minutes too late to meetings just as a

quick side note and I really looked for the best solution as I said like it's it's not AI related but you need to look for the best solution to a certain problem and this we call it the marble rule it's like coming late to a meeting you put a red one coming on time they put a green one we don't count individual scores but we count the number if they are filled up we count them we put it to our confluence and we document it and you see that people are just so on time and I really love this because it drives efficiency in a virtual company and this is one of maybe a hundred processes that need to run in a company so after the QA aided by the we'll stay in that shed for a couple of minutes longer so you'd have more questions I have one more but if you have more questions you can direct him to him he would be there along 12-ish so time for this to one question

is coming from Android and he asked for industries positive effective back corner in terms of they are I think this is a really nice questions we can discuss now as it's almost a final one where is Khurana having a positive effect um we just recently did like some analysis and some are obvious like travelers down restaurants are down hotels are down and it's clear but some are thriving for example ecommerce companies are thriving on all types of e-commerce there's some product don't thrive but most of them do in communication I think every SAS company that does some automation like soulmate I think we will be going out of this like recession stronger because there is a more need for automation at SAS and to answer that question I think first of all it's the need for solutions so how do you need more AI these days I think if your business thrives which means you're positively affected you need to analyze more you have more data and that means you can do more III don't

have it I don't have a cure for this which company leverages AI but I have a hypothesis which industries in general thrive so I agree everything which is by the book ritual will thrive everything physical this is punished clearly I mean its physical contact to online events will thrive software will thrive automation will thrive I mean zoom is skyrocketing all this this will be when the mess is over and people having a clear head again there will be so many business opportunities some very small people said the best time to start a company is during a crisis if you get the funding and the funding by the way is still there all the funds have plenty of funding that central banks are pouring buildings into the markets and money will not be the bottom yeah and why is that because of my crisis and now you see problems the garage and there isn't like there is some external fact that I changed something and Fabiana me yesterday I talked and I said hey the future

is virtual conferences because it's like I attended a virtual conference it was awesome and I don't like to travel a lot so but I would like to have great content and I don't need to miss a visit people physically so I see virtual conference is thriving if you now are about to start a virtual conference series do it because you're maybe gonna do it better than like the old physical 10 year old trade fairs because there is a mega trend for that and like Robbie and us you know but he's a digital later so by the way if you were to start an online business contact us because with the team the skills and we're doing this for a while so we're happy to help to assist to figure this out one more good question which I'd like to ask now is what was the biggest challenge and how did you overcome it in what in life and privacy let your life okay and second in your company that's a good point and I think

the biggest challenge in life to overcome is to please like as a founder to produce more energy than you think you can because you you're a you're not a lightbulb that means you're consuming energy you're creating energy like a power station and the only like the core thing that you need to have is energy because as a founder your problem-solving machine at the one side and the same side you need to have visionary visionary energy so it's but creating energy and I think it's a lifelong learning for me to create more energy to to solve bigger problems to wait to do that so it's a question of self motivation and that is what needs to be overcome to found a company in the first place um also at solve made I think I would like to be faster so like there was a few strategic learners that we have made because of course we make mistakes and I would like to be faster and that means having more strategic foresight because in the end as

I found a you need to like have an iPod with this and you know kind of the goal and then you need to find your way and I would like to take one or two shortcuts maybe setting happy and development time in the beginning maybe iterating our go-to-market strategy understanding our customers faster there is nothing I could mention specifically but I think each step like having a more professional foresight I like the part of the energy and I think I will give it to the editor maybe make a small video of this like to answer what is the biggest challenge how do we energize the last question how do we energize by the way I love the birds in the background it makes me feel great you know I'm a very very digital person I spent my whole day in front of the PC and like knowing all keyboard shortcuts and being a digital native but I love the exact opposite on the weekends like we have a wood-fired oven here and I'm chopping chopping

wood with an axe kids are keeping you healthy like we go on a trampoline here and we go for a walk I Drive five kilometres to the next petrol station to get some ice cream for the kids but we also do this in non corporate 19 times so I'm spending a lot of time out there in nature it's rejuvenating for me and to kind of have this exact opposite I'm also doing a lot of sports I like run a marathon a few years ago I will cycle over the Alps maybe not this year I plan to cycle over the offices here so probably that's not gonna happen but doing sports doing physical activities being alone in this super clear nature rejuvenates me wonderful great last words everyone else stay please online and thank you for your time thank you for your your sharing please say that said there are two more questions which way a little bit more detailed you might can answer them thank you for being here thank you it's a better talk

to you again bye-bye so unless you do something live I have to tell you it is not I need to focus for a second so thank you for being here I saw there was a constant flow of people I saw the numbers it grew like BAM up and then it stayed at a higher level some people joined those who haven't greeted a person yet who came in later hello welcome to the resume our virtual pets and each week I will invite one interesting personality to talk about AI it can be demo it can be an interview I will not have he knows I think they are not the right form it so you can suggest tweet it blog it sent me here comment send me an email hard when you want to about speakers you would like to have like I should invite I have a list here behind me of the top ten AI people I follow and I like and I will ping each of them until they have accepted and yesterday started

already has some really interesting people and who will join us one reason why I conference will not happen so I invited them over the patrol chat so will be second as topics I would like to figure a little bit out what you're interested in do you want to speak about AI business and ethics and regulation AI technical or do you like to have the mixture of everything you know everything around and each week it's different topic and you join when you like the topic or you don't run when you don't like the topic as again this will be also podcast maybe videocast so this content will be out there for a while you want to generate not only its activity here right away but also on something long-lasting I like the activity as user asked questions this question pod works really well a nice tool here where I can click answer question now so everyone later can just see this snippet of the answer there and next talk next week is with barrel and baby

L she is a researcher and she has a very multi ethnic background I think she's from Turkey and studying and teaching in Amsterdam or in the Netherlands and now I'm Karen Tyne in Sweden so she is like all over the place truly European and her research field was coming from image recognition and today it is truck worth the air and oh she says she will talk about with me about opening the black box Ave I honestly have no clue so I will be very curious I would go into the talk with her and asking her what is trustworthy I what do we how do we find what criterion is first time how can a code moral and ethics into a machines to do we still need this problem so it will be open conversation it will be interesting it's also important that we are female e.i researchers on stage honestly very hard in the market to find someone who's working this field so I'm very happy that we have all these factors coming together so

next week it will be thirst again but I think it's 3 p.m. so register W I will share the link here so you can sign up for this too and this will be my next call and then I am almost in time 2 minutes left I don't have to use them I face a thank you again for being here it was wonderful information and we are doing this for you from me I mean I enjoy to I have a team everywhere in other rooms or in the office where there - I think my mother even tried to lock in it she couldn't do it so I check box yourself this next time I think my father failed to but everyone else is here so kudos for this Alex if you're too late you can we watch the video after I have pressed on on end this broadcast else we have plenty lined up when you will get all an invitation for the rising as some addition we are planning something we will have in November

to summit story i where we do hybrid events it will be free or may be free for everyone to watch Patrol but if you want to be physical there meet the speakers physical want to do business physical what do you have to pay because we have costs for this or there will be some it's for a Iong strategy and then there will be someone to apply the ISO opportunity to if you're interested to speak or advertise or be the content part or something like this King asked King Dominick are happy to do this so for now enjoy this thirst thing stay home stay safe and we'll hopefully meet again next week have a great day bye

Rise of AI 2027 · Berlin · May 2027

Be in the room for the next one.

Ten years of talks started as conversations in one curated room in Berlin. The next edition brings 300 selected leaders together in May 2027.