Show Notes
Some estimate as many as 75 billion new devices will come online in the next
five years. This is where tech startup Latent AI enters the equation. Latent AI
is helping developers build AI models, specifically ones that can be used on
edge devices.
Latent AI co-founders Jags Kandasamy and Sek Chai joined Roland Siebelink on
this week’s episode of the Silicon Valley Momentum Podcast. In addition to
talking about Latent AI’s technology and market, they discussed all of the ways
they’ve tried to set themselves up for success.
- How to identify talent and find team members that fit the company’s needs.
- Why building a network within your given industry can be critical for new startups.
- The need to educate partners even before you can close any business deals.
- The role of advisors in Latent AI and how to find them.
- The importance of CEOs understanding “the four quadrants of me” and how to discover the “unknown unknowns.”
Transcript
Roland Siebelink:Hello and welcome to the Silicon Valley Momentum Podcast.
My name is Roland Siebelink and I'm the founder and CEO of the Midstage
Institute. We help funded startups maintain their momentum as well as their
sanity as they grow like crazy. And talk about crazy growing startups, we have
today with us Latent AI, the co-founders, Jags, who's also the CEO, and Sek, the
CTO. Hello guys. How are you doing?
Jags Kandasamy: Doing well, Roland. Thank you for having us here.
Roland Siebelink:Oh, I'm so glad to have you guys here. Both of you. It's a
huge honor for the Midstage Institute to have you guys here. Let's get started
right away. The first question every startup always gets is what problem do you
solve, for what target group, and what difference do you make in the world? Who
wants to get started with that? Jags, maybe it's you.
Jags Kandasamy:Absolutely. We are in the world of exploding devices. The
number of devices that's coming online is growing exponentially. The amount of
data that is getting created is also exploding. The only way we've been able to
solve this problem is applying AI. Today, AI is being applied in the cloud. We
thought, why not apply AI to where the source of the data is, where the data is
getting generated?
Our value prop that we bring to the table is we help developers build AI models
that can be deployed at small devices, small constrained devices, and along the
different pathways that the data travels. That's what we bring to the table.
Roland Siebelink:Okay. Your target group is really developers, right? Okay.
What kind of developers? Talk about the persona that you're targeting a little
bit. Are they just out of college? Are they very experienced? What kind of
companies do they work for? Let's make this as tangible as possible.
Jags Kandasamy:Sek, do you want to take that one?
Sek Chai: I think we want to cover the normal AI developer that has been
working on developing AI, but mostly for the cloud. When you go to the edge
devices, there's going to be a lot more constraint because these machines on the
edge are no longer plug-into-the-wall. They're mostly battery powered. There's
constraint with respect to size, weight, and those kinds of things.
And the question is then how do you build AI models specifically for those
subset of devices? We build tools that make that simplified. We can take the
hard work out of it. Because the tooling allows you then to create these models
for those specific targets. Then you don't have to worry about exactly what
target they are and they'll get done. They'll go through the process of
compression and optimization to run on those edge devices.
Roland Siebelink:Jags, what kind of use cases do you typically support, as
reference to a specific sub vertical or niche that you're honing in on where you
see this need being far more actual than in other industries?
Jags Kandasamy:A couple of sectors come to mind immediately. Of course, the
defense side has a lot of applications for edge AI. And then the other industry
side of vertical is more on the manufacturing floor.
Think about quality control. When you are producing hundreds and thousands of
items per minute, and it's coming through the conveyor belt, you got to make an
inspection to deem the quality of the material that you produced. It could be
bottled water or a bottled drink versus a casting part or a stamped part that's
coming out on the other end.
Most of these are visual inspections that people do. You have quality inspectors
standing there and they're sampling every fifth item or every 10th item,
visually checking it and putting it back. But it does not capture the error
rates that might occur there. In that case, you've got speed. The data is moving
very fast. The items are moving fast. You need to be able to inspect items with
a computer vision algorithm, an AI model.
We are mostly focused on the industrial sectors right now, on the quality
control applicant. Companies that have AI programs that are launching up and
they're looking at, "Okay, what do we need to make this happen?" They've already
collected the data. They have a certain pilot process in place. That's where the
users from a design exploration phase to figure out, okay, what do they need to
do to get this project running and also running in scale? That is a stage that
is absolutely perfect for us to enter in. And we help their engineer to look at
the problem and say "All right, this is solvable, not in nine months, but we can
get it done within a few days. Our tooling will help you get there faster."
Roland Siebelink:Let's then talk about that vision. Sek, how big is this
market that you're targeting with this edge AI? Do we think that it's just a few
of these sectors' edge cases or are we really thinking that this will be where
most of the AI ultimately will take place?
Sek Chai:It is a big market segment. The understanding that the sources of
data is plentiful. You have different sensors of different modalities in
different spatial locality. And right now, everything is streamed to a central
location on the cloud. And that simply doesn't scale. Then say "What is the
opportunity then for us to shrink some of these AI models and then move into
this spatial locality where it makes sense, closer to where the sources of data
might be." And then you say, "What are use cases there?" It can be a factory. It
can be in a home. It can be even mobile in the sense of having an autonomous
system, vehicles, wearables. A lot of things that can be made smarter.
I think there's a lot of opportunity there. And I think these devices can only
get smarter as we go over time. The AI models that we built will get smarter.
And we are able to shrink even the most complex models that have been running in
the cloud on these devices over time. And then I think the hardware itself would
get much more capable. Because now you have a specific target. We're actually
working with a lot of new semiconductor startups that are looking into building
new hardware accelerators. The resurgence in the hardware industry, in a sense,
sparks the momentum into this area. And we are in the mix of it, working on the
software stack that combines both of these areas together.
Roland Siebelink:Jags, how big are we thinking this total available market
is over time? What are you targeting in your big vision?
Jags Kandasamy:I think Gartner called this out. In the next five years, 75
billion devices are going to be coming online from an IOT standpoint. Right. And
what is today, I think 2% of these devices - whatever the devices that are
available today - are processing AI at the edge. That is going to turn to about
50% of them using AI at the edge of that 75 billion devices in five years.
Roland Siebelink:That's a big, hairy, audacious goal. And in just five
years, you said?
Jags Kandasamy:In just five years. Yes, exactly.
Roland Siebelink:That maybe brings us to how do you guys - having this
super powerful vision - how do you bring that down to goals that yourself and
your team can actually work on that that seem daunting enough but yet not
overwhelming? How do you do that? Jags, let's stay with you for a bit?
Jags Kandasamy:The secret is always in the execution. You can have a grand
vision, but unless, and until you're able to transform that vision into an
actionable plan, it doesn't work. Both of us have run larger organizations in
our past life. And we've done significant market MPs from that point of view,
launch products and release them.
The idea is how do you modularize it and have them contained within each one of
those units? And of course, hiring the right team members. And Sek is an expert
in identifying talent and bringing in people. And he's a rock star in industry
conferences. I was completely amazed when I went to CVPR a couple of years ago
in Long Beach. We were not able to walk more than 10 feet at a time. Somebody or
the other would recognize Sek and come and talk to him and thank him for all the
work that he's done for them or the guidance that he gave him.
I immediately realized that recruiting is not going to be a hard thing. Sek is
able to identify the talent and the right folk for our team. And I do more of
the cultural fit conversation. And we've brought in a good set of folks. They've
all been performing extremely well. That modularization and getting that plan
in place has been very helpful for us.
Roland Siebelink:Excellent. Sek, then we do have to talk with you about the
team, of course, Mr. Rockstar. How have you decided to build up your team at the
moment? The resources you have, how have you divided them up between different
teams or sub departments? What's your thinking around that?
Sek Chai:I think we're relatively small right now. But we do have a couple
of focuses that blends a couple of areas. As you see, we have tooling that
blends the machine learning side with the embedded side. You have to know both
areas. Because we are taking machine learning models, so you need to be an
expert in that area. And then you need to understand where it is targeted
toward. If it's hardware, all different flavors, you need to understand enough
of it. We bring a diverse set of teams that can have expertise in different
areas.
Once you put them in a room, magic happens. But then you still identify these
people, the team that we have, the staff that we want, that have the breadth and
the depth. Enough that they can make an impact and have that overlap where they
work. That's the key in terms of finding the right set of folks. The technical
breadth and technical depth.
And beyond that also, we look for motivation. A very well-motivated person
that's able to scale and learn because in the end, it's understanding how we all
grow, individually and as a team.
Roland Siebelink:You mentioned in the beginning, you're really using your
network to get leads and prospects. What's your vision about how to scale this
go-to market over time? What are some of the tricks that other entrepreneurs
could learn from? And how do you move beyond your network? How do you actually
start targeting a market at scale?
Jags Kandasamy:One, of course, you have to identify the industry that
you're going after and the relevant conferences. That's another way you build
your network and you make your name available. And then the other part, which is
very critical, is that you need to start nurturing partnerships, system
integrators, consulting organizations as part of your network. They may not be
valuable to you right now. They may not be able to sell or bring you value
today. But they are a part of the equation that helps you scale that one plus
one equals three, or one plus one equals 11 at a later stage value that they
bring about right.
You need to start educating your system integration partners. Don't spend too
much time on that right now. But spend enough time to get them educated, show
them the visibility of what the market space is, so that they can take your
technology - if you're building, of course, deep technology like ours - more
into a consultative sales process to take it to their existing clients. That
opens up a larger market for you as a startup.
Roland Siebelink:If I hear you correctly, Jags, you're really saying the
long-term vision is definitely to sell mostly through partners, system
integrators, and consulting companies. Of course, that's not going to work right
away, so you start educating them, start nurturing them. And in the short run,
you work through some of the contacts you already have or people you meet at
conferences to get your early-stage deals to build credibility. Is that a good
summary?
Jags Kandasamy:That is a very good summary. And your business is always not
going to be reliant on system integrators and partners. You will still have a
larger chunk of your business that is going to be direct. You always go direct
and then transfer that into a partnership because you will not have enough -
coming from experience, from the previous jobs - you will not have enough
resources to support a direct client for too long. You need to transfer that.
And you create an ecosystem where everybody wins. My famous belief is that a
rising tide raises all ships.
Roland Siebelink:You as leaders of this startup, and also maybe in your
previous lives, what resources, what people have you relied upon to break you
through some of these challenges? Do you read books? Do you talk to people? What
are your strategies to cope with some of the challenges you face when you're up
at night worrying about something? Sek?
Sek Chai:We were blessed with a number of technical advisors and business
advisors. These are friends, colleagues, and folks that we have worked with, all
situated in a different set of industry, CIO, former CTOs; those folks. And they
are helping us with respect to guiding us; Making sure we don't fall into the
same pit hole that they have walked through.
And those have been very valuable. I think it's very important to understand the
context in which we are in. A pandemic comes into your scope, what do you do?
And some of the things that we learn, at least I learned, is patience. Patience
is actually a strong virtue. You need to be able to understand how to use it and
to understand what needs time to mature products, business, especially for a
tech startup like us.
Roland Siebelink:That's that's very good. The people you mentioned, the
business advisors, technology advisors, is it typical that you have one or two
go-to people or do you like to treat them more like a portfolio and get opinions
from many, and then aggregate that into your own thinking? What's your strategy
there, Sek?
Sek Chai:I think we have both tech advisors that will give you insights
into the leading research and where the technology may be heading. You want to
make sure that you don't get blindsided with new things coming up. And we do
have business folks between myself and Jags that help us understand how to
scale. It's one thing to say, "I build like the coolest tech." But you need to
know how to scale and build premise products that get out there. Those are
advice for us and we reach out to folks as needed. And we do have one-on-ones
with them to walk through some of the things that we are facing. Those have been
very valuable for us.
Roland Siebelink:Very good. Jags. Who do you rely on? What do you rely on?
And how do you learn how to scale a business like this?
Jags Kandasamy:As Sek mentioned, we have a good array of advisory board
members that we have around for Latent. And this was a very deliberate thing
that both of us did, picking the people that would serve on our advisory board.
And there are no two individuals that overlap with the set of expertise that
they have. Each one is independent and very focused on their areas. We have an
engineering advisor on the team building.
We have a CTO that is more on the architecture side of the product. And then
there is a CTO that's more as a networker and a business development person.
Then I have a product person. I have a messaging person and a marketing person.
It's widespread and then the go-to-market sales leader as another adviser. We
have specific advisors for this. Then of course, I have one person that's like
my therapist that I go to. There are individuals that have specific roles that
we have brought into the boat.
Of course, I always believe there are things that you know about you that
everybody knows. The four quadrants of me. Things that I know about myself that
everybody else knows. And then there are things that I don't know that others
know about me. There are things that others know about me that I don't know
about myself. And then the fourth quadrant, which is a major black hole, things
that I don't know about myself that others don't know either. And that's where
books come into help. When you read books, it opens up those areas that help you
understand why you behave in a certain way, why you respond to certain things in
a certain way. And it helps you become a better individual and a better leader.
Roland Siebelink:Very, very good. I love that. I've talked to many CEOs
quoting this "leaders are readers" mantra, and I do think many of the best
leaders do read a lot. But I love how you pinpoint the books into that quadrant
of the unknown unknowns, if that's a way to label it. And how you use books to
lift things out of the subconscious into the consciousness. That's really good.
Again, I have one follow-up question about the therapist you mentioned. I just
wanted to hone in on that, not to discuss particular topics you discussed with
your therapist, of course. But more the general question, many CEOs I talk to
seem hesitant about opening up with an outsider that way, whether it's a coach
or it's a therapist. They say, I'm feeling a little bit vulnerable that way. If
you were to advise some younger CEOs that are just starting out on their
journey, can you maybe talk about what benefit this has given you over the years
to work with a therapist or a coach or somebody similar like that?
Jags Kandasamy:What I mean by a therapist is actually one of our advisors
who's actually a former CTO. And very good friends, we've known each other for
many years. None of my advisors are newbies that I've brought on, people that
we've trusted for many years. Both Sek and I have worked with these people in
independent capacities for many years before we bring them on as advisors.
From a therapist point of view, unless you trust somebody, you don't bring them
into your inner circle. I look at my relationships as concentric circles. The
family's, of course, in the core. And then, I have now second, the Latent
family, and then the next immediate circle with all our advisors in that.
Unless you're comfortable with somebody, you don't bring them into that circle.
When you have somebody in that circle, you should be willing to expose yourself.
You choose the person. And there should not be any hidden agenda in those
scenarios. But that person needs to understand they're there to help you. Most
individuals, you bury yourself, they are willing to take care of you, not take
advantage of you. But you've got to be very careful when you pick that person.
Once you pick the person and go expose yourself.
For example, being the CEO, it is never an easy job. Sometimes it feels very
lonely. And there is emotional turmoil that happens on a daily basis. There are
highs, there are lows. And you may hit both on the same day. And sometimes you
can have a peak for a week and then you'll have a valley for a week or two
weeks. You need to get over that.
I'm an extrovert. I need to talk to people and get things done. But I cannot
talk about my level of emotional turmoil that I'm going through with a team
member or with a board member or anybody. I need to pick somebody that's my
trusted advisor. I go to them and I will call them and say like, "I don't want
anything. I just want you to listen to me." And they will enable me to talk more
and more and more and more. And they say "All right. How do you feel now? Good.
Let's catch up later." hang up. That's it.
Roland Siebelink:Easy job, right?
So let's maybe close with some key learnings you would convey to other founders.
In your own journey and your own experience, what's the one thing you say you
have learned that you would like other founders to be aware of? Sek, let's start
with you.
Sek Chai:I talk to my team and I also reach out to new students and other
folks that have interest in the startup area, one of the things I always tell
them is networking is very important. It gets you to the type of bonding that we
talked about, having advisors at different capacities. For anybody coming up,
don't underestimate the value of that because it helps you open doors, it helps
you to reach out to folks and see new light. The kind of things that you may
have missed. I think that would be an advice. And I think that is correct for a
lot of folks in different contexts.
Roland Siebelink:Okay. Very good. And Jags, what about your closing words
for new entrepreneurs?
Jags Kandasamy:I could not say anything better than what Sek said.
Networking; no man is an Island. You build bridges whether you need them or not.
But when you need to cross them, you have them.
Roland Siebelink:Yes. I love that. That's very good.
Jags Kandasamy:Networking from your advisors earlier on in your life. Your
startup is not a one day event. It is a journey in process. You have come a long
way in a journey to get to that point. It's a milestone. And then you're on the
journey again. As your journey through, you build networks, you build your
relationships. You will continue to build them. You will utilize some of them
from the past. You utilize them, some of them that you are building currently.
Networking and building that relationship. Everything else, you'll figure it
out. You're smart. That's why you're starting a company. But that is a part most
people miss, the human connection.
Roland Siebelink:Okay. Very, very good. Okay. Last question for both of
you, if people want to learn more about Latent AI, where do they go? What should
they download and how can they be in touch with you?
Jags Kandasamy:LatentAI.com is our website. You can visit us there. Then
you can send us an email to [email protected] We will be happy to respond to it
if they are interested in testing our product. We do have a 30-day evaluation
program. Once you submit, we will be able to deal with the bits, the container
to test out the AI on the edge SDKs.
Roland Siebelink:Okay. And I understand you're also starting to work on a
new round. If any investors are interested, of course, I'm happy to provide an
introduction to Jags and Sek as well. Thank you so much, Sek Chai and Jags
Kandasamy, if I pronounced that correctly, the co-founder, CTO and co-founder,
CEO of Latent AI. It was an honor to have you on our podcast. Thank you so much.
Jags Kandasamy:Thank you, Roland. Appreciate the time.
Roland Siebelink talks all things tech startup and bring you interviews with tech cofounders
across the world.