PromptPod

#31. Deeptech investing and AI

Episode Summary

In this episode I speak with Mikhail Taver about investing in deeptech and AI.

Episode Notes

Topics:

Mikhail’s background in investing and interests in AI

Deeptech/AI and why he loves those investments

GenAI, why it is a bit hyped, how it might change the investing landscape

Fields he’s most exciting investing in and how they will benefit from AI

Links:

Fund's website: https://www.taver.capital/

Mikhail's Linkedin: https://www.linkedin.com/in/mtaver/

Mikhail's Twitter: https://twitter.com/TaverMikhail
Fund's Linkedin: https://www.linkedin.com/company/taver-capital-partners/

Fund's Twitter: https://twitter.com/TaverCapital

Episode Transcription

Welcome everyone to another episode of PromptPod, an open-ended 
exploration into the field of AI and my quest to document conversations with 
bright minds in the space. I'm your host, Danny Kirk, and today I'm joined by 
Mikhail Taver. Mikhail is a seasoned career investor with over 20 years of 
experience in venture investments.
Mergers and acquisitions and strategic consulting. As the founder and managing
partner at Taver Capital, an international venture capital fund based in 
Delaware, he focuses on investing in artificial intelligent companies worldwide, 
with a diverse portfolio of over 20 companies spanning various verticals 
including cybersecurity, mineral mining, tech.
Health tech, agritech, legal tech and sports tech. Taylor [00:01:00] Capital has 
achieved four successful exits throughout his executive roles in major financial 
groups in industrial companies. Mikhail has successfully closed over 250 m and a
and private equity deals. Mihail, welcome to the show. Hi, Danny. Thanks for 
having me.
Excellent. Well, would you tell, um, our audience a little bit about your 
background and how you got into investing and also how you got first got 
interested in ai? Right. Well, I'm, uh, I am what you might call a career investor.
I've always been interested in doing this sorts of thing. Uh, I was born and grew 
up, uh, at the very.
North of Russia, sort of polar zone. So there was really nothing to do there, 
especially during the long winter nights. I had this, uh, Huge, uh, sort of wall of 
bookshelves in my room, and I just, well, literally, there's nothing to do but 
read, uh, yes. I, I, I got a computer when I was about, 10 or 11, [00:02:00] but 
before that, I was reading that when I was about six or seven years old.
I read a book, uh, by Theodore, uh, a series of books. There's a three film, uh, 
uh, I did not understand most of it as I see from now, but I got interested in 
securities and. Investments and markets and, you know, those, those, uh,

 

difficult to understand pieces of paper, they apparently managed to, you know, 
uh, Drive the world in a way.
Uh, and this is how I got interested in investments in general. And since then 
I've been, uh, teaching myself to do that. Uh, I went to college to do that. I went 
to the university to do that. And then, uh, I became an investor. I've been 
investing ever since. And, uh, hope I'll be doing that for some time. So that's 
how I got into investments.
Very interesting. Did you know other people, um, in the town you came from 
that were interested in investing or was it until you got to uh, university? Uh, 
well, I mean, no university I went to, [00:03:00] uh, my, I did my gcss in the uk 
and that's when I, uh, started hearing the word investments from other people. 
So, no, no, no, no.
Before that it was only me. Some, I dunno. Excel tables and it's, and a 2 86 pc. 
Amazing. And how about ai? When did you first come across that, uh, Uh, this 
is interesting cuz I mean, the first time I actually, uh, came across that and the, 
the first time I've, I've learned something called a neural net, uh, was back in 
1999, I think, when I was, uh, at the university.
And I, I, I actually tried to create a, a neural net, but uh, it was just, uh, with a 
pen and paper, so I was just studying the concept. Uh, I understood how that 
works. It was interesting. But other than that, you know, I mean these were the 
times that, you know, things like data digging. Were not fashionable and were 
almost a swear word cuz I mean, you know, the more you look at data, the more
patterns you [00:04:00] see and that's not something you should be doing with 
numbers.
That's what they used to say at the time. As as we're all now, now, you know, 
this is the bread and butter of uh, ai. Starting. So this is when I first tasted that, 
uh, after that, mm, I don't know. I suppose the, uh, closest sort of real AI for me 
to touch was Siri as, uh, with the majority of people, I would say, this is when I 
got interested in that.
but as an investor, well, uh, our fund's first deal was something called 
masquerade. Uh, that's the company that you know. You might remember all of 
a sudden everyone started, you know, wearing those, uh, dynamic masks. Um, 
it, it was a quick, pretty quick deal. We sold it to Facebook, uh, three months 
after, after it all sort of started.

 

they weren't strictly speaking, they weren't using ai. There was something called
ris, but it doesn't matter. It was, it was still something AI ish. So this got me 
thinking and when, when we started planning, uh, a fund,[00:05:00] the real 
reason to be honest was that I'm not that good at delegating, so I knew that the 
actual, uh, team is not gonna be that large.
So we needed something with good potential, but manageable. Something 
niche. this is where we made a mistake of thinking that this is gonna be ai. 
apparently it wasn't that niche as we all now know, but yes, that's how I started 
investing. And yes, this is how the current portfolio came to be. Our, uh, second
investment was a company made Prisma that you all might know with the 
avatar craziness.
Uh, last year was December and so on and so forth. Now we're proud. I'm a 
proud owner of, uh, portfolio of, uh, almost 20 companies. in all sorts of AI 
applications, just like you said, from cybersec to health sector, to legal tech to 
sports, you name it. Did you know masquerade was gonna be such a hit, or was 
most of that just timing that Facebook was interested in buying them three 
months later?
you know, when you invest in something, [00:06:00] if you're a venture 
capitalist, uh, you always sort of try to look for something that is, looking in 
interesting and having the potential. I mean, obviously, uh, there's lots of luck, 
but, you know, luck favors those who do something. So I suppose it was, you 
know, a great team, a great application, uh, very novel at that time.
It was a good timing with Facebook. it was a perfect storm and I'm, I was lucky 
enough to be in the middle of it. Yeah, that's a great storm to be in the middle 
of. let's talk about, uh, deep tech and ai. I know that you love investing in deep 
Tech. Could you tell the audience, um, what deep tech, your definition of it and 
why it's of interest to you?
I would say that the best definition of deep tech I would, I've seen is that, I 
mean, it's something very boring. it's something that takes a long time and it's 
something that does not have, uh, market risks off the bat. You get lots and lots 
of tech risks, technological risks for, I don't know, five years, seven year, eight, 
[00:07:00] eight years.
You, you, you're doing something that does not exist. And, you know, even 
before you start trying to find the product market feed, it is something very 
complicated. uh, it involves lots of, bright minds coming together in a perfect

 

storm. so I would say it's one of the, well, I wouldn't say, yeah, it is a pinnacle 
of venture investing.
You know, it's, it's not the. prone to fashion and trans and, uh, human attitudes, 
B2C stuff. It's something like, say, I know, take prescriptive, prescriptive 
maintenance, for example. That's, that's, uh, something where, you know, you 
diagnose, uh, machinery and try to, you know, see. If it's gonna fail before it 
fails to, to make sure, you know, you do it in time in order to do that, you know,
you, you, you, you need to take a multi-discipline approach.
You need, you know, a specialist in, in vbro diagnostics, in mechanics, in 
modeling, in signal processing, and in machine learning. You just can't do that 
in, you know, in a few months. [00:08:00] And, you know, that's, that, that, 
that's what I find interesting about this, uh, as an investor. I love that. You know
it. You know, given the amount of time and effort and money often that, you 
know, uh, goes into such a project, it it does create, uh, some difficulty for 
someone else to get and, and compete with you.
So it is a bit of a protective, uh, environment as well. If you manage to find the 
right companies, it, it's very hard to replicate. So an investor, that's where I love 
plus, uh, I mean, from my sort of private equity, uh, background, I've always 
loved, you know, uh, varying industrial stuff that other people seem to find to 
find not interesting.
And, you know, uh, not many people know about this, but they use the, uh, 
products of that. Indeed. Yeah. It's so, it's so true. And on, on the deep tech side 
of things, do you have, you know, it seems like, um, in a lot of venture capital, 
it's at least [00:09:00] eight years till liquidity in deep tech. Is it even longer 
than that?
Do you have a kind of lifespan on your funds as far as like your expected return 
date? Well, uh, my current fund, uh, as you can see for the portfolio is a bit 
different. I'm sort of get trying to focus my second fund around the, uh, deep 
tech industrial ai, but yeah, from what I've seen for a portfolio, yes, this is 
correct, it's about seven to eight years to, you know, till you see the company 
properly operate.
But I mean, I, I think it's not just the dip tech companies, you. Any good 
company takes at least eight, 10 years to, you know, to become something. 
Other than that, it's all luck and, you know, circumstances on the tech side of 
things, what, what particular types of tech are you interested in from an AI 
perspective?

 

Either ones that you're currently invested in or problems yet to be solved that 
you would love to see. Um, companies being built for. Well, uh, from both 
current portfolio, [00:10:00] uh, I loved my, uh, mineral mining company. The, 
the Austral. They're bunch of Australian guys that use AI to, uh, find minerals. 
What love about them?
I mean, yes, it is ai. Uh, yes it is, you know, orders of market magnitude faster 
and cheaper to find those. It does a lot less damage to the environment. And 
what sort of got me interested in that is that they, they can, you know, they are. 
fuel. The ultimate goal was, uh, operating outside the planet Earth. And that's an
ambition I see, I see in some startups.
And you know, it, it's a long shot, but you know, well the future is tomorrow, so
why not bad on it today. Uh, so that's the, that's one thing I'm interested. I mean,
I'm quite used to mineral mining and, That's probably a bit of a competitive 
advantage on my side, cuz I mean the, in general, the general sort of SARS fans 
often don't understand or don't speak the same language as those guys, uh, do.
This is just a different sort of, uh, beast for them.[00:11:00] So that's where I got
an H. I also love food tech. food tech is great food tech. you know, as they say, 
software is eating the world. So, you know, why, why not feed the world? Uh, 
I've got some, some food and agritech in my portfolio as well. It, apparently, it 
is, a deep tech year sector as well.
It takes a long time to develop something that, you know, uh, works, is tasty and
is economically viable for the potential clients to purchase. And I assume that 
you do like deep, deep tech, just becau as of that defensibility, if you're gonna 
take time to invest in something, you might as well make it hard for others to 
copy it or do something competitive.
Is that correct? Exactly. Yeah. I mean it is just, it's not something you do, it's 
something, uh, it, it is an inherent feature of a deep tech project. It's hard to rep 
replicate cuz it involves, it involves a law of, uh, knowledge and work. And 
that's fun. I mean, it is probably the same fun as, you know, watching the paint 
dry or the grass grow [00:12:00] as, uh, I think Warren Buffett's put it.
But I mean, you know, like I said, I'm a career investor. I, I'm used to it. I love 
it. That's why I like to see. Yeah. And also it seems like anything that is 
connected to the internet, uh, is easy for an AI to help, um, compete against or 
to replicate. But for a lot of this deep tech, it's offline, it's in the fields, it's in the 
mines.

 

So that's also defensible from ai, even if it's using it. Is that right? Yep. Plus, 
you know, uh, well, if it's not connected to internet, not that many people know 
about this. Not, not until the acquisition happens. Exactly. Even then, they 
wouldn't know what, what's happened. You know, they'll probably read it, you 
know, just flip the page with the news and you know, go, go to something like 
masquerade.
With all due respect to masquerade, I love you.
Let's, let's talk about generative AI a bit. You seem very uninterested in 
[00:13:00] generative ai. Tell me, tell me why. Oh, no, not at all. I'm not 
uninterested. Uh, I, I am interested in, in it. I'm, uh, I'm even grateful to it in 
some respects, cuz, I mean, I, I like, uh, happy things. They take the, uh, Noise 
of my markets.
it, it is a, it is an interesting thing. Uh, it is fun to play with. It is. Hyped at the 
moment, I would say, I mean, it, it, it's easy to understand why it's, uh, sort of, 
uh, immediately understandable by people, you know. It is something they 
understand.
Unlike detect it's, it looks a alive, uh, it, you know, the very name, uh, artificial 
intelligence. It, it suggests something from this sci-fi. Uh, excessive and people 
think, oh, whoa, at last, you know, we have some, an AI we can talk to. You 
know, AI and AGI in their minds are something similar to me. I'd say it looks a 
bit more like a Chinese room, if you will.
You know [00:14:00] what that is, right? It's, it's a experiment. No. What is the 
thought experiment? you got a room. You got a person sitting in the inside bed. 
He doesn't speak Chinese, but he's got lots of instructions that, correspond, say 
English words to Chinese words. So get, see, he gets an set of instructions.
He takes the necessary say, I don't know, uh, book from the shelf, composes a 
phrase, and, you know, returns it back. Does he speak Chinese? No. Does he do 
the job of translating? Yes. I'd say that at the moment Genia is, uh, a bit of a 
Chinese room, so, mm, I wouldn't put, put that much trust in it. It's fun to play 
with, T that's a different, uh, thing that, that's not chat.
D p t it, it, it does demonstrate some, uh, more interesting features, but it is still, 
uh, an automated agent. It is still a Chinese room, but you know, on steroids if 
you will. Yeah, I force getting very wide. Yeah, [00:15:00] and I mean, you 
know, you, you know, the, the, the common uses of, uh, gen Geni at the

 

moment are co-pilots, which are helpful, but, you know, knowing how people 
operate, uh, chances are these are gonna be used as scapegoats.
You know, you, you kind of, Transfer responsibility for, for your decisions to 
them? You know, it's the geni that, that told you that it's not my responsibility. 
Yes. I took a sort of destructive action and, you know, lost all my funds on the 
stock market cuz they told me to do that. And yes, I probably prompted it 
wrong.
Speaking of prompting, I mean, uh, I keep thinking whether prompting is more 
like cooking or like baking. Cuz uh, you know, like with cooking, you, you 
know, you, you learn the basics, the sort of, you know how to cut the five 
mother sauces and so on and so forth. And then you're on the old, you can 
generate, you can synthesize and create new things with baking, you can't really
do that.
You have to take a precise recipe and follow it, otherwise you get a mess. I 
think [00:16:00] at the moment prompting is more like baking, which means 
that, you know, everybody is gonna learn it and then, you know, you are gonna 
start losing competitive advantage. How do you, um, what rules do you have for
investing around hype bubbles and, um, investing for the long term?
I, I don't, I try not to invest in high bubbles. Uh, sometimes I manage to get in 
them after I've invested, which is fun to watch. But, you know, uh, Fellow, 
pretty much the same sort of Gartner cycle partner, if you will. You know, they 
peak, then they, you know, uh, plateau a bit and that's when you start to realize, 
you know, what you've got yourselves into.
I've seen that with Prisma twice now. So, I mean, no, no, I, I don't invest in 
hypes. Uh, I, I am grateful to hypes cuz like, like, like I said, they, they, they 
take the noise off my markets. Uh, like it, it was the same with the, you know, 
the first AI was, uh, starting to pick up was around 20 16, [00:17:00] 17. 
Everyone started put putting AI in their, uh, uh, investment decks.
Everyone was starting, you know, saying that they use it. You, you, you poke 
them a bit and you know, is that necessary or not AI at all. But then crypto came
about. And everyone became a crypto guy. That was good. That was a fantastic 
time. Cause you had, you know, proper, pure companies. Now the hype is 
offered it, but Yeah, I mean, everyone's AI again, but they're gen ai so, you 
know, they, they, they're different kind.

 

So, you know, my, my startups are now clear of noise again, which is good. So 
when I'm not, I'm not saying I'm, I'm, I'm, I'm, I'm, I'm not saying investing in 
hypes is a bad strategy. It's just not something I tend to do. I prefer, you know, 
the fundamental things that are always going to be there. The, uh, Jeff Bezos 
quote, invest in things that don't change isn't, isn't that what he said?[00:18:00] 
I don't know the quote, but I completely agree with him. Yeah, I, I think, I think 
he quoted that around, um, investing in things that don't change and that was 
why he spent 20 years investing in two day shipping for Amazon. Cuz he is 
like, people will never want slower shipping, you know? So I invest in the faster
shipping possible and they, you know, exactly.
Spot on. So, I mean, you know, people are always gonna eat. So I am investing 
in, you know, agritech and food tech. People are always gonna require some 
sort of materials, which is why it's good to invest in mineral mining. Yeah. You 
might change from, I don't know, metals to composite materials, but that's still, 
you know, the same thing basically around, around mineral mining and mining 
in general.
How do you, um, how are you looking at investing in mining as it is relevant to,
the kind of changeover to green energy and sustainability? Well, I, I don't see a 
contradiction here. Like I said, one of, well, I mean the, the, [00:19:00] the 
portfolio company I was referring to is doing the same job, but you know, it's 
much less damage to the environment.
And if we move to space, yeah, we're still gonna be polluting, but it's gonna be 
some, not the earth.
It's, it's certainly true. What, what other, types of investments or, industries or 
fields that are using ai, are you excited to invest in these days? Metals, I mean 
like steel furnaces, things like that cuz, they pollute environment, but you know,
they're gonna be polluting less if they used effectively.
You know, you, you put some advanced AI in sales, say, and it's still fair 
fairness to. Change the, uh, proportion of fer composites. That's the stuff you 
use in iron to make steel, you know, while keeping the quality, you use less of 
them, bam. You got immediate returns, you got, you know, growth and 
capitalization and you, you, you do less damage to the environment cuz you use 
less stuff.
Are you, um, seeing any deals around people using AI for optimizing direct air 
[00:20:00] capture or anything like that? I'm not competent on that topic. I'm

 

afraid. I, I, I, I don't think I know that deals. No. Yeah. No, no worries at all. I 
was just curious. Yeah. Um, any, any other tech and as far as, um, kind of AI 
goes, anything kind of in the agritech space, what's, what's exciting about that 
these days?
Oh, well, in agri, uh, yeah. Well, again, I, I mean, I, I don't, I don't. Enjoy 
selling myself on a podcast, but I mean, alright. Right. Let's look at my portfolio
again. We got a robot that collects, uh, tomatoes, harvest tomatoes at the 
moment and they got, you know, plans for cucumbers and bell peppers and so 
on and so forth.
Uh, it's fantastic cuz I mean, you know, 2020 showed everyone that, you know 
how Frazier. Supply chains are how fragile humans in journal are. And you 
know, if you got your workers ill, that's it. You've got a problem. If [00:21:00] 
you got a robot, you know, it just works 24 7. It doesn't get covid. Uh, I don't 
know, it doesn't.
Drink. It doesn't get sick, it doesn't form trade unions.
So that's one example. Another one is, uh, another one where portfolio 
companies that uses AI to, uh, grow plants. So, I mean, you know, there's an, 
there's an inside joke that when we need a third startup, you know, got one that 
bro stuff, they got one that have a sit. We're not, we need a third one that eats it.
I'm sure AI could figure that out, I'm sure like that too. Yeah. Um, any people or
books or blogs that you enjoy in the space that you would recommend to others?
That's a bit of a problem cuz I'm, I don't really follow any particular authority. 
Uh, I've always tried to do things by my own, [00:22:00] uh, by my own mind, 
cuz I mean, following someone's mind is not the right thing to do, I think.
Cuz you never know what made them do a particular thing, even though they 
translated by a book which you paid them to, to read. Even you, if you get that 
thatk, you don't really know why they acted the way they acted. So, following 
someone is, uh, Not the best idea. It's like using chart G B T. Yeah. What, how 
do you, how do you digest information and then come to your own conclusions?
What's your model for that? Well, you just, you know, read and read and read a 
lot, read a lot of news, um, read a lot of, uh, books in general and just, you 
know, live longer than, I don't know, 30, 40 years. And then you start to. Kind 
of see how things operate and then you develop some kind of, uh, I dunno, uh, 
intuitive feeling of, you know, where you feel comfortable investing, where you

 

want to invest and where it [00:23:00] might, you know, uh, work that's your 
kind of strategic asset allocation, if you will.
You, you, you figure out the sectors that have the, this potential, then you drill 
down that sector and stop, you know, actively. Just talking about it to people, 
you know, searching for things. And that's where you find the particular 
companies in that sector. That's your, uh, tactical asset allocation. And then you,
you know, you start negotiating.
Any, any other, it's easy. Six to six. Yeah. Yeah. As long as you do that. Any, 
any other keys to success for, uh, long-term investing in venture capital? do 
your thing, you know, do what's right and you know, eventually if you do the 
right thing long enough, you're going to get it right. Yes. Don't, there are may 
there, there may be some short term bumps along the way, but, you know, just 
make sure you have enough, uh, dry powder to go through these t tranches and 
then.
[00:24:00] Bang, you'll be there. Always. Our final question, what interaction 
with AI have you laughed the hardest at? I've asked Chad, g p t who ta is
fact. Oh, no, no, no. It was, it was such a weird mix of wrong facts, but it was, 
uh, I laughter not what the facts, it was the confidence was that it was, I mean, if
I didn't know myself, I would've trusted it. That's the problem with chat gpt. 
Was it, uh, was it impressive? I mean, was it better than the truth or what, what 
Uh, it, it was very impressive.
Yes. That's, that's the problem. It knows how to, you know, uh, make sure you, 
you'll, well, I mean here humans, you know, fall for that. That's Nate here and it
bloody knows how to use that. Oh, that's amazing. Well, Miha, thanks very 
much for coming on our [00:25:00] show. If listeners are interested in learning 
more about what you're doing, where can people find you online?
Uh, they can, you know, check out our website that's, that's, uh, taylor.capital. 
They can check out my LinkedIn profile. Uh, or they can just, you know, if 
there's something they want to, they want me to know, they want to 
communicate with me, just. Send me an email. It's Mt. Tabor Capital. Very 
simple, and I always read my emails.
I don't reply to all of them. It's too sign consuming. So sorry in advance, but 
you're welcome to read. Perfect. I will mention all of those in the show notes. 
Well, michaeli, Hey, thanks again for coming on the show. Thanks a lot, den.

 

Thank you and thanks to you, my dear prompter for tuning in, and I hope you 
enjoyed this conversation as much as I have.
If you enjoy the show, please consider subscribing and leaving a good review. 
Take care and always be prompting.