[00:00:09.579]
Today, I'm joined by Suresh Venkatarayalu,
[00:00:12.310]
Senior Vice President and
[00:00:13.858]
Chief Technology Officer at Honeywell,
[00:00:16.350]
to discuss what it's like to be a leader
[00:00:18.859]
in this transformative time and how
[00:00:21.000]
Honeywell is using emerging technologies to advance innovation.
[00:00:25.760]
Thank for joining us today, Suresh.
[00:00:27.379]
Thanks for having me.
[00:00:29.079]
Yeah, welcome.
[00:00:30.228]
So let's start off, first question.
[00:00:32.658]
How will AI optimize the future of energy transition,
[00:00:38.020]
automation and aviation?
[00:00:41.310]
You know, I want to go back.
[00:00:44.310]
Honeywell is 135-year-old company, Laura.
[00:00:48.859]
Part of it is we
[00:00:51.029]
are a control systems company and if we really
[00:00:53.158]
go back and double click control system,
[00:00:55.010]
what do we do the best?
[00:00:56.539]
It's about sensing, connecting to the sensor,
[00:01:00.880]
having some of the control logic
[00:01:03.649]
that will close the loop.
[00:01:05.448]
I think that's probably a raw definition
[00:01:07.888]
of a control systems company.
[00:01:09.460]
And we do this very, very well across
[00:01:12.519]
various different sectors, from safety-critical
[00:01:14.990]
to mission-critical systems,
[00:01:16.870]
from buildings to industrial plant to an aircraft subsystems.
[00:01:22.668]
Now in the last 135 years,
[00:01:27.099]
how did we write a control system?
[00:01:29.308]
It's heuristic rule-based engine.
[00:01:32.540]
It's all embedded software.
[00:01:34.879]
In the last 20 years,
[00:01:37.040]
we have virtualized a control system.
[00:01:39.338]
So, the industry speaks about cloud
[00:01:42.409]
virtualizing the infrastructure,
[00:01:45.370]
IoT, we have done that with Forge.
[00:01:48.308]
But the last many, many years,
[00:01:50.870]
what I call a traditional AI,
[00:01:53.010]
we have been embedding
[00:01:54.638]
instead of a heuristic model,
[00:01:56.819]
we are actually embedding a regression model,
[00:01:59.189]
predictive model, mission vision system,
[00:02:02.269]
speech recognition system.
[00:02:05.290]
We're embedding the AI model so that these are a learning system,
[00:02:08.990]
a learning model that
[00:02:10.929]
will adapt to the dynamics of the building.
[00:02:14.220]
So, the building that
[00:02:15.538]
we are in right now, our corporate headquarters,
[00:02:18.399]
where we have millions of
[00:02:21.139]
data points, sensors that
[00:02:23.490]
feed into what we call a multi-modal, which
[00:02:27.588]
tends to adapt based on,
[00:02:29.758]
like in this room, both of us are right here.
[00:02:32.560]
How do we adapt and modulate
[00:02:34.300]
the temperature and then the light
[00:02:36.618]
and other controls, and then we have this algorithm.
[00:02:39.710]
So, AI is pretty much ingrained more and more
[00:02:43.258]
into our traditional products as we are pushing
[00:02:45.599]
the envelope toward
[00:02:47.038]
autonomous systems and autonomous control systems.
[00:02:50.368]
Generative AI is a very interesting shift
[00:02:54.189]
where we have a lot more with our install base.
[00:02:58.629]
And we are trying to say: How do we really take that data,
[00:03:01.058]
massive set of data sets, and how do we really
[00:03:03.219]
feed the large language model?
[00:03:05.538]
Do we build a knowledge repository
[00:03:08.250]
to assist workers,
[00:03:10.429]
assist field operators, assist pilots.
[00:03:13.990]
So Suresh,
[00:03:15.080]
can you give us an example then of
[00:03:17.740]
generative AI in our product set?
[00:03:20.790]
A couple of examples
[00:03:21.689]
that I can share at this point in time:
[00:03:23.919]
our plant operations
[00:03:26.599]
where we have the Experion system.
[00:03:28.889]
Right now, we are embedding large
[00:03:30.909]
language model to see - can we
[00:03:32.960]
build an operator-assisted technology?
[00:03:36.050]
We just shared that with our customers
[00:03:38.449]
in our annual User Group event a couple of months ago.
[00:03:42.679]
We had a lot of good positive
[00:03:45.179]
feedback, and I think we should take it down
[00:03:47.389]
to a deployment
[00:03:49.639]
or a launch with a few customers in the later part of this year.
[00:03:53.319]
Second, with pilot assist, more and more in the future,
[00:03:57.020]
it's going to be either single pilot or
[00:04:00.028]
less crew in the operation side.
[00:04:02.670]
Aerospace is really looking at it
[00:04:04.580]
to really embed the generative AI.
[00:04:06.020]
So, there are two broad themes, one
[00:04:09.129]
in the services side and
[00:04:11.240]
then also in the operational side.
[00:04:13.460]
How do we seamlessly bring in generative AI that can aid
[00:04:17.548]
and assist operators, workers
[00:04:19.730]
and field technicians to do
[00:04:22.798]
their jobs better and to impact their productivity.
[00:04:25.800]
Early days, but I think we should have a
[00:04:28.070]
very strong roadmap in the coming years.
[00:04:30.250]
What excites you most about your work and this era of innovation?
[00:04:35.500]
You know, I would want to say
[00:04:39.600]
creativity, coming up with new ideas.
[00:04:43.010]
But one thing that is exciting me the most now is
[00:04:45.939]
an era of co-innovation.
[00:04:48.379]
We are starting to spend more time with customers
[00:04:51.250]
and we're trying to realize the problems that they face
[00:04:54.678]
and how do we solve the problem
[00:04:56.928]
through their eyes and through their processes?
[00:05:00.319]
And I think that is probably going to be
[00:05:02.428]
the new innovation machine that I think
[00:05:04.528]
that we are changing. We're dramatically shifting
[00:05:07.019]
at this point in time.
[00:05:08.439]
The second area, we're also asking the question,
[00:05:10.798]
how do we co-innovate with our partners and suppliers?
[00:05:14.139]
Connecting the dots,
[00:05:15.699]
I think that's probably one of the areas
[00:05:17.970]
that I'm thrilled about - is if I'm
[00:05:19.988]
able to really connect the dots of customer issues with our roadmap
[00:05:24.588]
and with our supplier technology partners' roadmap.
[00:05:27.990]
And if we can do
[00:05:29.160]
that faster in a most innovative way,
[00:05:31.680]
I think that's going to be more
[00:05:33.548]
interesting for Honeywell and for
[00:05:35.699]
a role like I'm playing today.
[00:05:38.230]
It's pretty exciting to think about how
[00:05:40.309]
generative AI can help assist with that, right?
[00:05:43.110]
So, augmenting those interpersonal
[00:05:45.428]
interactions with customers
[00:05:47.548]
and seeing the process, with the data you can collect.
[00:05:50.990]
When you think about AI driven automation,
[00:05:55.580]
how does that impact,
[00:05:56.949]
you know, future engineering workforce
[00:05:59.488]
and given that impact,
[00:06:01.590]
how would you advise the next generation of
[00:06:03.949]
engineers to think about their future
[00:06:06.189]
and to navigate their focus?
[00:06:07.899]
It's interesting, when the first
[00:06:11.250]
major launch from Microsoft
[00:06:13.889]
GitHub Copilot was launched,
[00:06:16.660]
I picked up probably eight different pilot groups with
[00:06:20.889]
senior architects,
[00:06:23.600]
piloting and probably proving how the technology works.
[00:06:27.930]
I was very impressed about it.
[00:06:30.139]
It's a coding assist,
[00:06:32.358]
or I should call it software-coding robots or robots.
[00:06:36.778]
It had a tremendous
[00:06:39.579]
learning through that whole cycle from
[00:06:42.379]
code generation to code modernization,
[00:06:45.509]
to code testing, to documentation.
[00:06:49.100]
Things that were done traditionally by
[00:06:52.178]
developers and testers.
[00:06:54.480]
There are a number of things that are going to be automated.
[00:06:57.250]
I see the potential
[00:06:59.009]
to be close to 40% waste
[00:07:03.889]
that could be eliminated.
[00:07:05.358]
Now, what does it mean for
[00:07:08.769]
the software developers of the future?
[00:07:11.600]
I see them moving slowly
[00:07:14.778]
and steadily to be great designers and architects,
[00:07:17.830]
spending time with customers
[00:07:19.750]
and solving issues and building
[00:07:22.949]
the next solution, less about coding.
[00:07:26.750]
So, it's going to be an interesting shift,
[00:07:31.309]
but I think it's a needed shift.
[00:07:32.990]
You know, using this technology to really
[00:07:36.028]
enable our subject matter experts to be close to the
[00:07:40.199]
customer and to be more innovative and more strategic.
[00:07:43.350]
As you know, with this podcast, we
[00:07:47.004]
always like to end with a question to understand
[00:07:49.824]
a little bit more about you.
[00:07:50.860]
So, when you were young, what did you want
[00:07:53.875]
to do when you grew up?
[00:07:56.045]
I wanted to be an architect
[00:07:58.184]
for many reasons.
[00:08:00.733]
In the early days,
[00:08:02.653]
I was fascinated with the things that I would see,
[00:08:06.240]
more buildings,
[00:08:08.069]
house designs architected and then how
[00:08:13.298]
an architect would probably work
[00:08:15.420]
with people to bring
[00:08:18.358]
something from a drawing or a design
[00:08:21.199]
to a thing that you
[00:08:23.259]
can realize over a period of time.
[00:08:26.149]
And at a point in time, I was advised that -
[00:08:29.449]
what's the next best thing that you could do
[00:08:31.660]
so that you could find your job
[00:08:33.820]
and starting to learn and move forward?
[00:08:36.259]
So, I jumped into the software world.
[00:08:38.729]
I had a major in computer science, and I,
[00:08:42.100]
30 years ago I did my neural net. But today,
[00:08:45.229]
what I'm doing is pretty close to that, I think.
[00:08:47.990]
It's all about designing,
[00:08:49.940]
creating, solving for customers,
[00:08:55.410]
and creating something with people.
[00:08:59.489]
And something that we create solves
[00:09:01.989]
issues for the people in the world.
[00:09:04.040]
So, even though I had this view that I wanted
[00:09:06.769]
to be a great architect and designer,
[00:09:09.629]
to build buildings, but I'm
[00:09:11.649]
building something that is pretty closer to it, I guess,
[00:09:14.680]
which is what motivates me and
[00:09:16.918]
something that
[00:09:18.649]
I feel pretty good about it.
[00:09:20.519]
You have so much
[00:09:22.798]
opportunity driving this team and
[00:09:25.359]
helping to build the future with Honeywell.
[00:09:27.550]
Thank you for your time today. This has been a really great session.
[00:09:30.979]
Thank you so much for your time. Thank you.