Feb. 1, 2021

Where is AI getting us and how deep-tech investing works? By Amit Garg

Where is AI getting us and how deep-tech investing works? By Amit Garg

Amit Garg, Managing Partner and Co-Founder at Tau Ventures talks about deep tech investing and the AI field. We discussed how VCs make their decisions while evaluating complicated deep-tech startups. We also discussed how AI is used now, where it is going and what are the major misconceptions that people have about this field (hint: Terminators won't kill us any time soon).

Transcript

All right, and today's a guest speaker, we have a need Garg managing partner and Co, founder at towel Ventures.

And today we'll talk about deep tech investing and specifically what is going on in a field right now and what to expect from it in the near future.

And also, we'll talk about how founders in deep tech should test out their ideas really early on when they have little to no funding at all. So, I made a, let's kick it off by you giving us some background on yourself and on Tel Ventures.

Yeah, it's very much embedded into the company, so yeah, let me take a pause here and turn it back to you. Sounds good. Sounds good. So, 1st question actually can be about the robot that's making.

Smithers does just that just caught my attention and I would love to hear more about it. So, when this company came to you, they present you their idea. What were the major things that you looked at in that company before making the investment decision?

Well, I'm glad you asked because it was 1 of the finance diligence processes. I've ever done. 1st question. Can I try this movie?

That was the 1st question, which I did, and I said, well, I think I need to do more diligence going to have a little bit more smoothly, but I'm not so jokes aside.

This is a team that at the time we invested was about 15 people, and the 3 Co founders had been working together for a long time.

They were a very rare breed of entrepreneurs people who knew both software and hardware. Really? Well, and they had built a company that they sold into Barnes and Noble, that eventually became the NOC.

So, this is a, an experienced team and that's obviously a big factor into investing in the seat stage.

You want to see folks that bring something to the table either in terms of experience, in terms of vision in terms of just being very hard workers. Obviously not. Every team is going to have that.

We have taken multiple paths into 1st, time entrepreneurs but we want to see is that they have a deep passion and understanding of what they're doing. And this was plenty obvious to Vincent.

Jane, that's the name of the CIO and his team. The 2nd, part of it at the time of the investment was the actual product. So, this is a big robot.

That is meant not for a home, but for a commercial use so think about it in a university setting in a cafeteria, or in a corporate setting, and a cafeteria, maybe at airports, or any retail shop.

You have this robot that is making smoothies much more efficiently. Much more personalized.

When when you order through your phone, you may have an allergy to certain kinds of nuts, or you may want an extra bit of mangoes, or you may want to just exactly a smoothie. That's that's green.

And has these particular ingredients.

If you go today to any smoothie maker, you can't get that level of personalization not to mention that level of efficiency because a robot can program all those tasks and can get a lot more economies of efficiencies.

Right?

Like,
if they're making 40 smoothies,

they can make all 4 of them in the same batch and they can reuse the ingredients so that they reduce the amount of time that would take to pick up each ingredient from a shelf,

which is very hard for a human being to do a human being is much more of a serial thinker.

I'm going to make this then I'm going to make this then I'm going to make this, but for robots, they can shuffle the tax very quickly.

So, that was very exciting for us that here's a robot that has a mixture of hardware and software and the beauty here is in in the artificial intelligence of how you program the tasks. How you schedule the task, how you get them accomplished much more easily.

And then the 3rd thing here is the business model we, we saw that by making labor more efficient. You can make human beings more efficient.

So, in a in a slightly Pre pandemic world, at least, you might have had 4 people making smoothies.

And now you might have those same for people doing different jobs, which is much more about greeting the customer and, and obviously still handling payments. This robot is not doing that.

But up levels,

the tasks of human beings,

instead of doing the wrote task of cutting vegetables,

chopping mangoes and keeping everything running smoothly robot can do,

all those manual tasks,

much more efficiently mcminn shipping,

getting allow humans to do more higher value,

add a task along the lines of service,

so those were the 3 big reasons why we invested.

This is perfect description and I love the 1st, 2 parts of the due diligence when you were trying some of these stats. I think that seemed like the most important part of it.

But now, now, let's move on and talk a little bit more about the thing that you've mentioned briefly, which is the fact that Ventures is based in Silicon Valley and you've been there for, like, 20 years. What do you think will happen to that place after? They've been demak is over to you.

Are you 1 of those people who believe that everyone will return to the Silicon Valley or.

You do not think that well, Silicon Valley is 1 of many technology centers in the world it is it's the most developed in many metrics ecosystem.

So, roughly speaking about to use 1 metric about a quarter of the venture capital in the US is in Silicon Valley. It's specifically, maybe 3 or 4 parts of Silicon Valley and that is a metric, right?

It's not the only 1. but then you also have another metric, which is the number of folks that are involved in economic activity around tech.

Whether it's companies like Google and Facebook and sales force, all of which are headquartered here in Silicon Valley, or in startups.

And I do think that Silicon Valley is there's many reasons for why Silicon Valley is Silicon Valley. It's perhaps a different question.

Those reasons I'm not going away with or without pandemic and I don't see, I actually wrote an article about this just last week on my LinkedIn.

I don't see the fact that other parts of the world are developing and also becoming more mature startup ecosystems to be a bad thing. It's actually a good thing. It's a great thing. Doesn't mean the Silicon Valley is getting left behind.

It means that other parts are accelerating and also becoming developed.

I do think that Silicon Valley will continue being very much the biggest and most developed ecosystem for long time for any foreseeable future, but other ecosystem.

So we'll are, and we'll continue developing, perhaps at a faster magnitude.

In the US you have New York, you have Boston, you have Austin, you have L. A, where you are based yourself Constantine. So I think that's it. That's a great thing. We want a multi polar world.

We don't want a dependency on a single engine, driving the world economy, or the US economy, or any kind of economy for that matter. The tech economy. So your question, I'm going to modify a little bit.

People have never all been in Silicon Valley. So it's not like people will return to Silicon Valley. We, we have some exodus from Silicon Valley right now, but we also have some exit us from New York and from Boston, and from any major city in the world.

Because people want to live in more socially distance ways, and they want to live in suburbs and rural areas, and you can do that in a remote economy. Do we go back to the same economy were Pre pandemic?

No, we don't, but will we stay in this? Completely virtual economy. Absolutely not either. There are advantages and reasons for clustering their adventures and reasons for people to still want to be offices.

So, a lot of folks have written about the 3 tomorrow and I'm a believer that that will be a reality for a lot of tech, not for all industries. But the 3 to model basically means 3 days from home, 2 days, 3 days from work 2 days.

From home, I think that's a very feasible reality for a lot of us.

Right. And that's just that perfect answer. Absolutely. Love it. And I'll make sure to include the link to your article in the description this episode.

So, if anyone's curious to read more, and that's definitely check out the description of the episode and let's move on to next question is going to be about deep deck again.

So our preacher recall, he mentioned that sometimes you invest in something very complicated cancer research and.

Stuff like that, which is very tough. So, uh, on those.

On those investments, what are the major things that you look at? So, when you cannot see the product already done, so you cannot try this movie that is done by the company. But it has a very long term horizon.

So, what are the major things that you look at in those companies? Yeah,
yeah no,
that's a very meaningful question given the deep tech is,

in my view where a lot of great innovation happens,

and that's what gives us the Leafs of progress and society.

So that's 1 of the reasons we are deep tech investors. The other 1 is because we have done it in our careers. I'm a computer scientist and a biologist by training. That's what I did in my undergrad and I did my masters in healthy.

I, and I've been an operator entrepreneur primarily before becoming an investor, so my partner is a similar background.

He was a computer scientist who did a lot of work around healthcare also, and we can still both code and he worked at Microsoft. I work at Google.

So, we're bringing our own backgrounds to work here and that is very much part of the thesis is that we may not know all the answers, but we know what questions to ask.

And if we don't know how to validate those answers, we have a very strong set of folks around us. We have a strong set of advisors that can help us validate. We have our own investors. We have our own entrepreneurs.

So,

when I'm looking for a company is whatever they're doing there is some kind of differentiation with that's part of our thesis,

maybe they have a proprietary data set,

maybe they're simulating data through synthetic data maybe there access here to to analyze the data 10 times faster.

Right like, I need to see something here that's a technological edge. And what I'm trying to validate here is that this is not a research.

Project that this is a commercially viable company. So,
to your point,
if it's a cancer company,

and I'll give a very specific example,

computer vision for colon cancer,

the company that I mentioned called iterative scopes,

when we invested,

we saw that they had collected 20000 videos from 40 hospitals and they're already trained their algorithms on it,

so that when they were detecting polyps,

they could find almost a 99% accuracy,

whether somebody had a polyp.

That could be dangerous that could become into cancer. That is, by the way much better than any doctor out there. Even the best doctors are able to get to 75% detection. So.

We saw a proof point here. It wasn't approved by the day. At that point. In time they did not have paying customers at that point in time. But we saw that they had been able to collect the data. They had been able to train an algorithm.

They had a working prototype and I think that's the ultimate definition of a seed. It's not how much money you raise, but it's what you use your money for the framework over all that I have.

Is that a Pre seed is a, a seed is a prototype, a mature seed or a seed extension, or or whatever term you want to use for?

It is a pilot you want to see that they are customers that are starting to use it and then a series, a product market fit.

So, I play specially between the seed and the mature seed, and I want to see prototypes and ideally pilots with a line of sight towards the series. A, which is the product market fit.

Nice, I love the description of difference between those series of those rounds sounds very accurate. Actually. So before we move on and talk a little bit more about the, I wanted to ask 1 more question about.

Why you like the regulatory process so, in the beginning you mentioned that he liked to invest in um.

Health care, especially if there is hard tech or some sort of regulatory process. Why is that? I believe that. I don't think I've ever met investors who are like oh, yes, regulatory processes complicate stuff. I love that. I'm gonna I'm going to go in there.

So, what why do you like it?

Oh, absolutely, we are we pride ourselves actually on tackling hard things and look in healthcare. There's maybe a 10% universe of companies. That can go direct to consumers.

And by the way, I'm speaking more about the US, rather than about Europe, or other parts of the world.

But in the US, we have large insurance systems, and people by and large are honestly been overwhelmed in terms of dealing with healthcare.

We feel as a society that we're already paying into healthcare, that's our insurance to take care of it, rather than us paying out of pocket.

So, companies like, Telecom for me are to show how that is a very consumer driven, digital health company. That's an exception to the rule. Most companies and healthcare in the U. S are B to B to C.

so they will sell into payers and providers meaning doctors and into insurances and increasingly we're seeing them paint, selling to pharma.

The framework, incidentally, is the 4 piece of healthcare so, pairs providers, patients pharma, some people call it a 50, which is policymakers, but we want them to be selling into into payers and providers.

And when you're going to be selling into Paris and providers, they have requirements. They want to see that there's been clinical trials. They want to see that. This is real. This is legitimate and there will be an FDA approval process in many cases. Now, the FDA is not a, a.

A model is right it's there are easy regulatory process, and there's complicated regulatory processes for those of folks listening in, at a very high level.

There's class 1 class 2 class 3 class 1 is is much easier and there's fast track for those regulatory processes. There's 1 specific called fast and case, so we look for the low hanging fruit.

I'm not very unlikely to invest in a class 3 device, which is very invasive, but class 2 I'm open to it and class 1 for sure. And 510 case. Absolutely.

I want to see that you you put yourself through some rigor that you did get the approval. That you have something here. That's real.

And the scientific proof, and the scientific credibility will mean that you will go further with the medical community and give it gives me as a patient also much more conviction about using this technology. So.

You know, I say, bring it on company should embrace it if you're going to do something in healthcare, it's it's good news for all of us that there have been proof points around, around validating. We're dealing at the end of the day with people's lives. Right.

You can't get it wrong. True and I love that. I really like that something new something that I have not seen in the past. So that's awesome. Nice approached by Ventures here. So let's move on and talk more about.

And what started as discussion by talking about something fun and interesting, which is, what do you think is the best application that you have seen so far.

Oh, boy, that's like, asking me who's your favorite child right? There's so many applications of, uh, that that gets me really excited.

Models for cars to be able to take decisions by themselves. It's a problem that many other companies have tackled Tesla, Google and so on.

And 1 of the challenges is that a car needs to understand what's going on around the world, and you need to give it a lot of data for it to make sense. Now think about you.

And I, when you and I are driving a car, let's say, in a place, we don't know, it's not like we have mapped every single inch of this new location. Right?

Like, if your team, let's say, driving in Seattle, I don't know how familiar up in Seattle, it might be more lesser known place to you than than L. A.

And it's not like, you know, every single thing about Seattle, but you have certain rules in your head, you have certain experiences in your head, and you are have certain expectations of how the traffic will behave in Seattle and that's how you're able to navigate around.

Seattle, right that's what these guys did for autonomy in the car. They created models that allow the cars to make decisions required 10 times less data and that's all about.

It's about taking a lot of work that had been done in industrial robotics that had done in aeronautics and built on top of it to help the car make decisions requiring 10 times less data than other solutions

around it.

Now, that 10 times, the data is an incredibly powerful differentiation, because it means that you can go to market much easier and you can actually get these cars working much more effectively with much less data. Right?

So, that's at the heart of the investment that I made. I have a super cool. That is ultimately going to drive. No pun intended that change and self driving car.

I'm a big believer in it, because it fundamentally changes society right to think about where you and I spend a lot of time in our lives.

We spend at home, expanded office and we spent going places and the car or the boss or the vehicle. Right. Is essentially essentially what allows us to move around and society.

And it is,
if we fundamentally change that,
where we live,
where we work,
how we play,
where businesses are located,
where offices are located,
where real estate prices are like that,
that is a change in the same magnitude as what the pandemic has done right now.

So, I continue being a big believer in Tommy and its latest incarnation. It's not part of other companies. Self driving cars can do.

Nice. 1st of all great pond you've done there, even though it's unintended, I still loved it and it was sort of a guy that came up with it while talking. Nice.

And particularly, yes, it does sound really interesting and yes best of luck to and autonomy next question is going to be about how other people versus how really is. So, you know, I, I'm barely familiar with the field.

I seen a bunch of companies in that space. I've seen a few pictures, but never really worked in that space. So, for people like me, what do you feel is the major misconception about the field.

There's many misconceptions 1 being that will let me highlight too. If I'm a 1, is that I will solve all of our problems. It won't.

We're not going to have the Terminator coming back from the future. We will not have Skynet coming up.

We will not have the matrix all of this is science fiction the day, where it becomes truly, truly a general purpose.

Intelligence is really, really, really far away the eyes really, really good at very narrow specialized domains where we feed it lots of data.

And it makes sense out of, and there's differences between supervised learning and unsupervised learning at a high level, which we can talk about later.

But at a very, very high level is is good as long as there's human beings that are powering, it is a very powerful tool. It's a very powerful.

So it's nice and what can do much better than humans is computations. They can calculate things it can uncover mistakes.

It can show patterns that human beings wouldn't be, perhaps ever able to figure out, but it is not a creative system, like the human brain. So that's 1, big misconception that the general public has. I don't think that practitioner survey.

I have the conception, but what practitioner Sophia fall into perhaps is the 2nd track, which is that pessimism around has been around in this modern sense for about 50 years.

The wording I at least was coined in the 19 forties, and we have been through the promise and the pitfalls of the eye multiple times in fact, there's determine the industry called the winter where we say, okay, i's ready, ready, ready. Oh, not really. Okay.

i's ready, ready, ready. Oh, not really like, we have been through these cycles where people go, like, okay, I will solve everything and it doesn't.

And then we get disappointed about it and I think, I think we've got at least twice now, but this time it's fundamentally different. And the reason is fundamentally different is because we have a lot more data, and we have a lot more competition power.

The amount of data that's produced every year now is more than all the years combined. Like, we're talking about some data being produced by society. At this point. It's, we had the inflection point.

And that means, we have a lot more now to feed into these systems to help train them. And then the 2nd thing is, okay, just database doesn't mean that we can ingest it and train the systems.

We also have increased over computational power and an exponential rate. Like, we have Moore's law and we have exceeded more slide this point in some ways.

So the fact that there's more competition power in this little device that I hold in my hands, my cell phone, the computer, the goddess.

To the moon 50 years ago like, literally, how crazy is that right? Every single human being that has a cell phone is walking around with something more powerful than than a journey to the moon.

And that is unprecedented in in human history. We have never had access to so much power and so much fuel. So, that power.

So that means that is now possible, in a way, that was never possible before, and we're talking in order magnitude more than an order magnitude of a leap that we have taken literally the last 2 or 3 years.

It is part of the reason we have created Ventures is that we see that we've hit that inflection point and they succumbing a total explosion of possibilities of solutions of products. Now, that, that that will be created.

Nice that does sound very interesting and very inspiring. Actually I'm kind of caught up with high now when I read few articles on that. So, after that, I will sure as hell do that.

But before we wrap it up 2 more questions, 1 is going to be about early stage developments for deep tech solutions. So a lot of it requires a ton of research and development, which requires a lot of money.

So, how have you seen founders overcome this problem of? I have a cool idea, but it requires a lot of development, which requires a lot of money and money requires some kind of development already done to.

How have you seen founders overcome this chicken and egg problem early on in deep tech. Yeah, you ask great questions have constant Tina.

So we invest in fee verticals as I mentioned, and healthcare we talked about earlier. It's really hard to have a fully baked product early off. You can show proof points, right? You can show a prototype that's working algorithm.

That's working a pilot perhaps at pipeline enterprises at a different place. Enterprise. Typically, typically you have products that already work, even at the seed stage.

You may not have product market fit meaning you don't have deployments in many places, but you might have some initial deployments. So it tends to be a little further along and then automations all over the place.

They are companies like the self driving car company and described from my previous life where you're not going to have a self driving car available, commercially. It's not really available commercial yet, but you might have robots to make smoothies that might be operating. Right?

So, it's a little bit all over the place in general for us at the seed stage mature seed stage. What we look for is a pipeline once again, going back for the commercial uptake.

I don't need to see necessarily that you have pain customers, but I want to see a line of sight.

I want to see that you are having those conversations that you are, are being able to open the doors and walk through them. If it's a 9 to 18 month sales cycle, then, you know, you should have gotten

started on it. And how you're going about it matters to me, perhaps as much as.

What you have accomplished and this is a good lesson for me also as when I was an entrepreneur, right? Like, we were very mindful that you can build the greatest technology in the face of the planet.

But if you don't sell it, if you don't get to the right people, then it's a beautiful piece of technology sitting on a shelf. Right? So we will look for very much business oriented, deep technologists to me.

It's more interesting to find somebody who is applying.

The technology and and commercializing it, then the other way around, like, somebody who understands the technology very well, but can't commercialize it.

And what we really appreciate is partnerships,
like a lot of the teams you have invested in,
there's a technology technology minded typically,
but the business minded CIO that are working together,
it's not always those exact same roles but oftentimes we see that.

It's a partnership of 2 or 3 founders doing that in rare cases. It is 1 founder who is able to do both, but that's the Yang and the yang that will look for any deep tech company should really be doing in order to get to scale.

Right, right very accurate. And I loved that description. If you do have 9 to 18 months sales cycles yes. It's time to get started with those because it's like more than a year.

So on this pretty positive notes, we're moving onto the last question of the examples. Which is a call to action. So what is the 1 thing that you want to listen to do? As soon as episode is over?

Well, 1st of all be very successful. Whatever you're doing, but there's a startup or you're aspiring to build a startup. But if you are building something here, that is relevant to us at our Ventures.

Send us a note, send us the deck. We're easily find on LinkedIn and warm intros always helpful. That's true. For any venture capital fund.

vc's get basically flooded with emails and notes. I get about 2000 companies.

I'm looking at any given year and unfortunately we can't pay attention equal attention to all of them but we do look through everything. Somebody sends us and.

In the cases, where we do see a fit, we do follow up, it ends up being about 10% of the cases.

And the reason I was bringing up the warm intro is, because if we see something coming from somebody that we've already worked with that, we have a lot of credibility with then it helps filter a little bit better. Right?

It increases the odds of a company. An entrepreneur falling into that 10% that we, we go deeper into. We do our homework very well in whatever. We end up investing, but as many other VC funds, we have a large photo. Right.

Look at a lot and be very go very deep for the few. And for those companies that we end up investing in, we really, truly go off the bat we work as as much as we can to help them succeed. And that's a good early stage.

Investor should do is be a true partner.
Nice and that's very accurate. Very optimistic as well. On this optimistic note we're wrapping it up.

My call to action is can be 1st, of all listen to a call to action and be successful and also make sure you check out the description of this episode.

I leave, I mean, it's article in the description of this episode so if you're curious to hear what it thinks is going to happen to Silicon Valley after independently definitely. Check it out.

And also, I'll make sure to leave links to tell Ventures and also to, in the description of this episode.

So, if you're in deep tech, highly highly recommend, you checking out the description of his episode and going through those links and as usually have a good day.