Connixt Presents: iBot – The Future of AI-Driven Maintenance

0:00
[Music]
0:15
[Music]
1:01
[Music]
1:26
all right well uh hi provu hi hi and
1:30
welcome to everyone that is uh showing
1:32
up for this LinkedIn live also being
1:34
broadcast on YouTube live we Sati we are
1:39
excited that we are
1:40
live live that’s great so uh so here’s a
1:44
quick background so I’m Satish I’m the
1:47
CEO and the co-founder at Conex with me
1:50
is pru yeah pru co-founder and C at
1:53
Conex nice meeting you all so just a
1:56
quick background to those of us that
1:59
don’t know us you know you know at least
2:01
KX has been in business uh for about 10
2:04
years now or a young company still um in
2:08
primarily focused on mobile applications
2:10
for
2:11
maintenance um Asset Management uh Work
2:14
Management our life a product lives uh
2:17
out on the shop floor uh in the fields
2:20
and basically out of the open so that’s
2:22
what we do we integrate the range of
2:24
backend Enterprise systems we can also
2:26
operate
2:27
Standalone um and uh we’re right now
2:30
here to talk
2:32
about uh one of the newest bolon we’ve
2:36
introduced called iBot which works with
2:38
our Flagship product the KX imar Mobile
2:41
App
2:41
Suite and uh there’s just essentially
2:44
going to be a conversation between CL
2:46
and me talking a little bit about the
2:47
background to ibod what we what it does
2:49
today what the plans are for the future
2:51
and of course please pop in your
2:53
questions comments and so on as we go
2:56
forward so uh pru so one of the first
2:59
things right
3:59
it it generates Revenue additional
4:01
revenue for the customers and end of the
4:03
day it basically allows the mechanics
4:06
and technicians uh to do the work that
4:09
they need to do um rather than spending
4:11
time and researching stuff so with all
4:14
these things in mind we took enough time
4:17
uh we talked to our customers we
4:18
validated our use cases and then we are
4:21
introducing this so that basically leads
4:24
to this uh point right that the focus is
4:28
first on the use cases
4:30
yes and uh and the technology is kind of
4:32
the easier part I’m not suggesting it is
4:35
it’s it’s non-trivial but but it is easy
4:38
yeah it is easier than this stuff right
4:40
and and essentially that leads leads us
4:42
to as a business um timing has to be
4:45
right yeah and we don’t believe in build
4:48
it and they will come uh the basic
4:50
question is really rather than figure
4:53
out will someone buy this new the
4:56
question is are even our current
4:58
customers ready for it you know folks
5:00
that are already using imar for example
5:02
are they willing to try something like
5:03
this you know that’s really the big
5:05
question so um one thing I I I feel is
5:10
kind of significant for us is we’ve had
5:12
a number of conversations with our
5:14
current customers and let’s say soon to
5:16
be customers right and I’m really glad I
5:19
mean this is something we can be proud
5:20
of uh that a couple of even newer
5:24
customers have already signed up to
5:26
implement and try out iBot even at this
5:29
pre-launch stage you know simply based
5:31
on the progress we’ve made and on the
5:32
use cases we showed so maybe you can
5:35
talk a bit about thatu on the use cases
5:37
sure before even getting into the use
5:39
case itself why we say technology is
5:41
easy right uh the reason for that is
5:44
imar itself as a product it’s a
5:46
configurable product it’s it’s a 99%
5:49
configurable product and couple years
5:52
ago we made enough changes to our
5:54
framework built in enough hooks so that
5:57
we can extend to these other areas and
5:59
that’s the reason why when we wanted to
6:01
implement the technology part was easy
6:03
for us but we spent a lot of time in the
6:06
use case like we we identified like five
6:09
broad areas uh in which this can play a
6:11
role Diagnostics work instructions
6:14
automation language conversion patterns
6:18
recognition and based on that doing some
6:20
actions and then sending alerts uh now
6:23
sending alerts everybody does but we
6:26
want to send meaningful alerts right
6:28
where AA plays a role so these are all
6:30
the five areas you know we identified
6:32
and we also talk to our customers out of
6:35
those five areas two things we have
6:38
already implemented in our product the
6:39
rest of it is in the product road map we
6:41
are introducing as we speak um so to to
6:44
get a little bit deeper into I’m not
6:46
going to go too deep uh because it’s
6:49
just a 30 minute conversation so uh
6:52
let’s let’s just have a birdday view of
6:53
the use cases itself right Diagnostics
6:56
what is that we are going to do
6:58
mechanics and technicians can can ask
7:00
questions to our app in our app and then
7:04
it can come back with responses like for
7:06
example U my battery is draining right
7:09
overnight what could be the possible
7:11
cause I’m taking a very easy
7:12
hypothetical case like this any
7:15
questions can be asked to the user so
7:17
that to the to the app so that the app
7:19
will respond saying that hey this could
7:21
be the possible causes for this or they
7:24
can even ask for uh instructions on
7:27
changing an oil or changing a battery
7:29
battery maintenance things like that
7:31
again there will be suggestions made by
7:33
the app the final decision will be done
7:35
by the customer itself or the technician
7:37
itself but there will be suggestions uh
7:40
from the outside world that is from the
7:42
Diagnostics perspective work
7:44
instructions automation right it
7:46
was it was an interesting U topic when
7:49
we brought this up to our customers they
7:51
were excited because they have so much
7:54
instructions out there and for us to go
7:58
ahead and SK through those work
8:00
instructions on a manual basis it was St
8:03
so we are planning to automate that uh
8:05
using AI engines saying that a user asks
8:07
a question we kep we pick up the right
8:10
work instruction and then show it to the
8:12
users again this work instructions can
8:14
be from the outside world for customers
8:16
who do not have any standard procedures
8:20
that needs to be followed for customers
8:22
who have standard procedures in place
8:24
it’ll pick up from their set of
8:26
documents and then it’ll bring it up so
8:28
because you know we deal with a lot of
8:29
of Transit organizations which are um
8:32
kind of regulated in nature so we cannot
8:34
just like that go and you know open it
8:36
up and ask the techncian to do whatever
8:39
uh that the app suggests right so we
8:41
have standardized it that way that those
8:42
are all our plans now next is language
8:45
right now I remember I don’t want to
8:48
name the customer but uh you know I
8:50
remember one of the uh VPS was telling
8:53
hey Brau you know what or Satish you
8:56
know what I spend most of the time
9:00
in understanding the comments from a
9:02
user so this is a a real life situation
9:06
where uh you know when a person wants to
9:10
enter commments in the app right in our
9:13
configurable mobile app they will feel
9:15
comfortable in talking in their own
9:17
language rather than talking in English
9:20
sometimes right so we support that we we
9:23
use AI engine um to translate from any
9:27
language to any language so so that end
9:30
of the day we document in their own
9:32
language and also in translated English
9:34
language and then it gets stored in the
9:37
server so that when a manager or a
9:39
supervisor goes through the um commands
9:43
it can make sense right it’s again
9:45
quality of work quality of data
9:47
enriching data it goes back to all those
9:49
uh getting collecting accurate data from
9:52
the technicians and mechanics right so
9:54
so any language to any language
9:56
multilingual support so that we do and
9:58
it is again again when we demonstrated
10:00
that to one of our customers existing
10:03
customers they said hey I want to
10:05
implement this what it would take and we
10:07
are in process of implementing that so
10:10
uh the those are all some of the use
10:12
cases are pretty straight for forward it
10:14
hits hard and it has a real life uh
10:18
implementation right a real life use
10:19
case so prau before we move on to the
10:21
couple of others I know there pattern
10:23
the alert so I think language is
10:24
particularly interesting because um
10:27
especially when you look at uh buiness
10:29
businesses that are you know globally
10:31
dispersed U and the the workforce is
10:34
also Global right yeah it is a real
10:37
question as to you know if you have um
10:41
you know uh technicians or crew on the
10:44
ground in Asia in a particular region
10:46
there encountering a certain issue and
10:49
they have a question about how do we fix
10:52
this or what is our Standard Process
10:54
yeah can they tap into a documentation
10:58
or a similar experience that been
11:00
documented you know halfway across the
11:02
world perhaps in Europe yes yes so this
11:05
is truly uh where I ask questions in my
11:09
language the system Taps into the
11:12
company’s own database or Goos and goes
11:14
and looks at the oem’s documentation and
11:17
comes back and gives me that information
11:20
in the language I’m most comfortable in
11:23
I can register comments which now get
11:25
embedded in the system which now
11:28
somebody else can access from halfway
11:29
across the world I mean this is truly
11:31
powerful stuff and again we are so used
11:34
to this in our day-to-day lives right we
11:36
talk in one language and there’s
11:37
translators and all of these things I
11:39
mean this is what we bringing to work
11:41
correct you remember that our visit to
11:43
one of our customers they asked question
11:46
in their own language um again I don’t
11:48
want to mention the language but still
11:50
uh it’s like uh when we showed it to the
11:53
customer not just grabbing documents
11:56
right Diagnostics they can ask questions
11:58
in their own language y our app will
12:01
respond in their own language telling
12:03
them hey this could be the possible
12:05
causes so all those combinations are
12:07
possible and those combinations are
12:09
possible because we have implemented the
12:12
AA engine AA framework as part of our
12:14
product yep now going back to the next
12:17
use case Sati is there anything else
12:18
that you want to add no no no go ahead
12:19
please okay now the next use case is
12:22
pattern recognition right now uh we
12:24
collect so much data um through our app
12:27
50k transactions runs through our app on
12:30
any given month that means that much
12:32
data is available to us right so we are
12:34
implementing models in such a way that
12:37
it recognizes the anomalies right
12:39
recognize the pattern on which like for
12:42
example um I I’ll again give you a
12:44
simple use case oil temperature engine
12:46
oil temperature right it suddenly shoots
12:49
up or radiator temperature suddenly
12:50
shoots up there is some issue with that
12:53
asset right now it follows a pattern
12:55
then we we we tap that we create try to
12:58
create uh you know actionable steps
13:00
where um breakdown work order or a
13:03
maintenance work order and things like
13:05
that right again it’s a hypothetical
13:06
case so we are implementing models so
13:09
that you know we can recognize patterns
13:10
and based on that take actions and then
13:13
the last but not the least alerts and
13:15
campaigns right uh one one thing that
13:18
you know we are planning to do is or we
13:20
are already implementing some portion of
13:22
it is where when we are sending alerts
13:24
right for ordering parts hey your parts
13:27
consumption is going down it has not hit
13:30
the reorder point But at the rate at
13:33
which it goes down in a week you may
13:36
have to place a order right things like
13:39
that making it more intelligent
13:40
previously yes we were automating
13:42
purchase orders yes we were automating
13:44
orders based on when the reorder point
13:48
is being hit but now we are extending it
13:51
in such a way that hey based on that
13:53
consumption pattern it’s not just today
13:56
but in a week you will be placing the
13:57
order so things like that is made
14:00
possible because of this AI framework
14:02
Camp right recall campaigns if there is
14:05
a recall by an oem we are planning to
14:07
you know tap into that information and
14:09
then bring it as an alert now in all
14:12
these things saes like we discussed
14:14
earlier in all these things except for
14:17
translation translation is um you know
14:19
one to one right it’s Apples to Apples
14:21
so that’s straightforward whereas all
14:23
other diagnostic information or anything
14:25
else of that nature we give suggestions
14:28
we don’t desire side so that’s that’s
14:30
also a key aspect where we give guidance
14:32
to the user end of the day it makes a
14:35
mechanic and a technician life easier
14:37
and all these use cases are we run run
14:40
it through to through our customers and
14:42
then made sure it makes sense and we are
14:45
open enough where some use cases may get
14:48
added some you use cases may get dropped
14:51
if there is no use for a customer so we
14:54
we believe in that we don’t do again
14:56
just for the sake of Technology but
14:58
unless there is a value at we don’t
15:00
introduce that and you know that sa so
15:02
this is a great couple of points so in
15:04
fact one without going too deep into the
15:07
mathematics or engineering gravit hole
15:09
one thing I can say is why is pattern
15:12
recognition such a big deal um we could
15:15
view this as simply automating manual
15:18
task but there are some things that are
15:19
humanly impossible to do right can you
15:22
go through massive amounts of data and
15:24
come out with a pattern to say that not
15:26
only are you hitting the reorder point
15:28
which is all programmed and in the
15:30
system but each successive order you’re
15:34
hitting it faster than before yeah or uh
15:38
it is perhaps the other way around maybe
15:40
you’re hitting the reorder points later
15:42
than normal than than you would then
15:45
maybe you have to go back and look at
15:46
the lead time maybe you have to look at
15:48
other factors in there you know I need
15:50
break pads more you know during winter
15:53
yeah just like I need more umbrellas
15:55
during you know the rainy season so this
15:57
this is interesting about pattern
16:00
one other point I just want to add is
16:02
see if there are 10 10 parts 20 parts
16:04
spare parts or 30 spare parts it may not
16:07
add that much value but we are talking
16:10
customers who are having thousands of
16:12
parts right uh thousands 10,000 20,000
16:15
30,000 parts now for a human to go and
16:18
skim through and take a decision on this
16:21
it’s impossible so that is where you
16:24
know these kinds of solutions come in
16:26
and it goes back to the use case and
16:27
validity and value and and all those
16:29
things I I think it’s a as as you can
16:33
probably say and as when people see this
16:35
they know we’re pretty excited about the
16:37
possibilities that still presents so
16:39
let’s uh switch gears just for a moment
16:41
right we’ve talked about the things that
16:42
we are planning and that we can do I
16:44
want to talk for a second about all the
16:46
things that AI broadly and iBot
16:49
specifically cannot do yeah so one you
16:51
already pointed to this obviously it
16:53
can’t actually do the work of fixing
16:55
repairing or inspecting yeah on the
16:58
field I mean and that’s or in the
16:59
workshop that is still dependent on
17:01
human beings and their Ingenuity in
17:04
interpreting data that comes that is
17:06
presented to them or interpreting
17:08
suggested action that is presented to
17:10
them yeah and even more obviously much
17:13
like a
17:13
GPS it cannot substitute for you as the
17:17
driver seeing that the bridge in front
17:20
of you is
17:21
incomplete uh or the the road uh that
17:24
the GPS is asking you to take actually
17:27
leads into a lake mhm so do make sure I
17:30
mean this is a standard disclaimer I
17:32
think everybody would do well to heat
17:35
that uh these are suggestions you can
17:38
use this as guidance and the guidance is
17:41
more accurate than you know based on the
17:44
data that is presented right so that’s
17:47
one angle to just keep in mind yeah and
17:50
again just to reemphasize right we don’t
17:52
decide we suggest we give directions we
17:55
don’t decide we don’t want to decide end
17:57
of the day thech and the mechanic and
17:59
the customer has to decide number one
18:01
number two is we basically don’t want to
18:05
replace a human that’s not our goal but
18:07
we help them or we make their life
18:11
easier that’s that’s the reason why we
18:13
do all these boltons where hey um in
18:16
fact we have barcode scanning as part of
18:18
our app nothing to do with AI but
18:20
barcode scanning why we do it we make
18:22
their life easier right so that we
18:24
always emphasize on I should I should
18:27
bring this up Sati when we met
18:29
uh one of our customers you know we came
18:32
we went and told hey we can also
18:34
automate work instructions he told hey
18:36
don’t touch my work instructions because
18:39
it is regulated I cannot ask my uh
18:42
technicians to follow your work
18:44
instructions they have to follow our
18:45
work instructions right that was an
18:47
excellent point he brought up we also
18:50
told him that hey you can do diagnostics
18:53
with this then he said yes that is the
18:55
area that I can use but not on the work
18:57
instruction side because you have to use
18:59
my own work instructions that time U we
19:03
we not we we mainly pointed that our
19:05
product is configurable so we can tap in
19:08
into his work instructions and then show
19:11
show to the mechanic again this goes
19:13
back to the use case and this also we
19:16
want to reemphasize this again we are
19:18
not going to decide we are going to
19:20
suggest our app is going to suggest to
19:22
make the life easier for the mechanics
19:24
and
19:25
technicians so I I I think this is a
19:28
really important important Point uh we
19:30
come here we come to right um this is
19:33
especially useful for you what you might
19:35
call I mean there is some things you
19:36
know you’re trying to automate um when
19:39
you put in something like I bought you
19:40
know that are every day right do I have
19:43
to go and pick up manuals and read them
19:45
or do I have to skin through pages those
19:46
things are kind of uh you know pretty
19:48
standard within uh the product but it is
19:53
especially useful when you start looking
19:54
at the outlier cases right I’ve not seen
19:56
this problem before who else has
19:58
encountered this problem either in my
20:00
own company halfway across the world
20:02
yeah or or out is the OEM telling
20:05
something about this that can be uh uh
20:08
that can be useful to me I mean these
20:10
are interesting things but sort of leads
20:12
me you know as we go towards uh sort of
20:14
towards the end of this uh we’re
20:17
starting to get some questions so I do
20:19
want to be mindful of that yeah before
20:21
that just a very quick thing saes um
20:24
with one of our prospects right this
20:26
happened this is not a customer it’s a
20:27
prospect to to emphasize on um in fact
20:31
that is is it hard to use kind of
20:33
question is also coming but again no it
20:35
is not hard to use we we can demonstrate
20:38
that but having said that see we we were
20:41
doing this demo to one of our prospects
20:43
right they told uh they came back and
20:45
told hey by the way um you guys are us
20:48
showing with um some of the AA platforms
20:51
out there we are developing our own GPT
20:53
engine can you guys integrate the
20:55
immediate answer was yes you know why
20:58
because we made our configurable prod we
21:00
we made our product to be configurable
21:03
enough so that as long as there is uh
21:05
API to access we can configure our
21:08
product to utilize the GPT engine of
21:11
that customer itself which is again goes
21:13
back to what Satish was mentioning right
21:15
which is tapping in into their own
21:17
information their own resources and then
21:19
coming up with suggestions go ahead s
21:21
sorry so no this actually is a great
21:24
segue to the last couple of points I
21:26
want to make um before we wrap up right
21:29
so do we know all of the use cases that
21:31
we are going to implement uh the answer
21:33
is no clearly no I wish we could say
21:36
that the path is very clear these are
21:37
the five these are the only five things
21:39
we’ll do and these are the only ways in
21:40
which those five things will be done uh
21:43
because uh what we what we have done is
21:46
we have built the hooks that can help
21:47
with whatever the future may throw up so
21:50
that’s the flexibility of our core
21:52
product you know imar as well as what we
21:54
have also built into iBot um and I think
21:57
Pro as you pointed out the hooks can be
21:59
for example to different as yet unknown
22:01
AI engines yeah that may come up in the
22:04
future yeah including the company’s own
22:07
AI engine yes yes so so so on a broad
22:11
basis either go search literally the
22:13
whole worldwide web yeah or their own
22:15
company’s database or the oem’s
22:17
announcements I mean depending on how
22:18
narrow what Hooks you want to say Yes um
22:22
this Le
22:23
me tomorrow fright liner comes up with
22:26
their own GPT engine we will tap into
22:28
that GPT engine because they have even
22:30
more accurate data right so just as an
22:33
new flyer electric bus they come with
22:35
their own GPT engine we will tap it to
22:37
that and our product will become is
22:40
configurable U so we can tap it into all
22:43
these different engines so the the last
22:46
thing before I go to my last Point
22:48
somewhat related to this right as newer
22:52
asset categories come into play for
22:54
example we’ve talked about uh EV
22:57
vehicles uh you know mass transer or
22:59
self-driving vehicles or new
23:02
technology I think this becomes even
23:05
more interesting to see what are the new
23:07
kinds of issues because we’re after all
23:08
replacing technology that’s over 120
23:11
years old you know when we move away
23:12
from internal combustion engines to
23:14
something else so that’s 120 years worth
23:17
of data that we cannot tap into anymore
23:20
right so I think this becomes especially
23:22
interesting right how much how quickly
23:24
can we grab data from around the world
23:27
and sense out of it right grab
23:29
meaningful data grabb meaningful data
23:31
meaningful data and convert to
23:33
actionable intelligence yes yes yes so
23:35
the very last question and I this is
23:37
just a sort of
23:38
a quick kind of a summary of of many of
23:41
these things right the biggest question
23:44
you encounter when you go and talk to uh
23:46
some of our clients especially within
23:48
the SE sues or SE suets is uh what the
23:53
basic question of Roi right and you know
23:56
which sort of leads to the basic
23:57
question of why
23:59
why should anyone be interested in this
24:01
and to be candid uh prau you and I hear
24:05
this very frequently when we go to
24:06
customers right hey you know we have to
24:08
walk before we can run so don’t don’t
24:11
come and talk to me about Ai and all of
24:13
these things now we’re just starting to
24:15
crawl and you know let’s give us some
24:17
time and you know then we can go to the
24:18
next
24:19
stage I think we need to bring it back
24:22
to the basics right increase Revenue
24:24
reduce cost mitigate risk I mean these
24:26
are three core areas that we have Focus
24:28
on yeah it goes back to how do we use
24:31
technology today in our personal lives
24:33
is it making life easier for us and what
24:36
does that translate to and for
24:38
businesses the impact is high because we
24:40
reduce time spent on non- value tasks as
24:43
you were saying prau ease of
24:45
communication across geographies and
24:47
languages
24:49
onboarding um and training and retaining
24:52
employees how easily can you transfer
24:55
knowledge to new employees because from
24:57
a business stand point we are capturing
25:00
a a ton of information from the field as
25:04
the crew sees it and experiences it and
25:07
in the language in the mode they are
25:09
most comfortable with whether it’s
25:11
photograph type text uh spoken you know
25:14
speech to text or simply recording audio
25:17
whatever it is and that’s basically
25:19
enriching data ensuring accuracy and all
25:22
in real time yeah in my mind that
25:26
basically drives value and I think
25:28
that’s sort of the goal we are driving
25:30
towards so um you it’s been fun while
25:33
this is lasted Pro and we should do this
25:34
more often so to everybody listening out
25:37
there um you know we do have a beta
25:40
signup
25:42
URL uh you know at the bottom of the
25:44
screen uh go ahead give it a shot try
25:47
this or scan the QR code um the one
25:50
thing we can say is that uh this is
25:53
working today it is not Theory it’s not
25:56
PowerPoint where to put in mindly so
25:59
this is stuff that is working so happy
26:01
to address any questions you know pop
26:03
those questions out to us either here or
26:06
or on LinkedIn later or directly reach
26:09
out to us but happy to address those but
26:11
otherwise any parting words or thoughts
26:13
PR no I’m good I think we we pretty much
26:17
um discussed on all the areas that we
26:19
need to cover saes one thing I would
26:21
like to is it hard to you I just want to
26:23
reemphasize this go goes back to again
26:25
our philosophy where you just talk to
26:28
the app
26:29
the rest will be done by the app even
26:31
talking to the app with two clicks you
26:34
will be able to get the answer whether
26:35
it is translation or diagnostic
26:38
information whatever you want within two
26:40
clicks you will be able to it is that
26:42
easy please reach out to us so that we
26:44
can demonstrate it to you well with that
26:47
uh I’m going to close this uh this
26:50
LinkedIn live session this has been a
26:52
super interesting 30 minutes and we’re
26:54
glad that uh you some of you chose to
26:57
listen in on
26:58
one of pru and my conversations here
27:01
thanks everybody for joining in and uh
27:04
you know as reach out to us if you have
27:06
questions or if you’d like to give I got
27:09
a try yeah take care see you thanks
27:14
[Music]
27:27
bye
27:36
[Music]