How AI Will Improve Business Efficiency and Productivity In the Workplace – With Ben Plummer

Channel: Alex Kantrowitz

Published at: 2024-09-05

YouTube video id: PLtMuqvO58A

Source: https://www.youtube.com/watch?v=PLtMuqvO58A

let's talk to a CEO with firsthand
knowledge of how AI is working
throughout business not just chatbots to
improve efficiency and productivity and
where these gains might be hidden we're
joined today by Ben plumber he's the CEO
of invisible Technologies which is a
company that combines humans and AI to
improve processes and make businesses
more efficient this conversation is
sponsored by invisible technology so Ben
thank you so much for being here great
to see you again and great to speak to
you about where the AI field is heading
thanks for having me Alex it's great to
be here so first question for you is
where's the ROI okay because there's
been this meme recently that companies
have said we've spent or really
investors have said there's been so much
spending going into Ai and nobody's
making money from it I doubt that's
actually true and I think you're
probably closer to the return and the
actual implementation of this technology
close to where the businesses find Value
uh from it you're closer to that than I
think anybody else so where do you think
the returners today and where do you
think this is heading yeah look there's
sort of two ways to look at this
question I think one as you said is on
the sort of foundational model providers
or the technology companies that are
investing huge amounts of monies in
building uh and training these models
and I expect that's only going to
continue that the amount of money that
goes into training these and building
these foundational models is going to be
in the billions and if that is only
going to increase and so there's a huge
question there around the sort of
long-term viability of that the business
models but these Technologies are so
transformative that the ability to
actually crack open AGI or something
that even resembles something close to
the power of that is so transformative
for the economy that this is a real long
game that they're playing here that this
is not something they would expect to
see in Roi over the next 2 or 3 years
and so squinting at their revenue this
year or next year I think is sort of
missing the broader potential of these
Tech Technologies then when you go to
the Enterprise side the sort of
application of these Technologies we're
just very early in the adoption of this
you know most companies got caught
flat-footed with the release of chat GPT
and the development of some of these
Technologies and it really just sort of
spinning up their own teams and
allocating resources to it and so you're
starting to see the first Frontier of
companies that have been able to
drastically transform the cost of
service delivery or their operations you
know think there's cases of companies
being able to streamline 6 700 people
from their operations and deploy those
to more customer facing activities or
other areas and I think this is just the
beginning of of that sort of broader
transformation and what to avoid here is
these sort of science project AI poc's
we've seen a lot of those over the last
couple of years which is companies just
Building Technology for technology sake
it's really important to focus on what
actual business problem you're trying to
solve solve to work backwards from there
so your view is that the value really is
going to be unlocked in Enterprise
before consumer am I reading that right
yeah look I think consumers are already
benefiting from these Technologies I
mean I don't I can't even go to the
fruit shop without someone talking to me
about how they're using chat GPT to do
something here and there um I think the
business model around that is sort of
much more unclear in the same way that
you know sort of advertising really
drove the the sort of search business
but I think consumers are already
benefiting from these Technologies and
we're probably the first to adopt and
that enterprises are actually lagging a
little bit in terms of how do they
actually rewire the fabric of the
organization using these Technologies
we're very early in the adoption cycle
of that um and companies just don't have
the capabilities inh house to go after a
really ambitious transformation
agenda right and what's coming through
here is that you know I think it's so
interesting in the popular imagination
people say Ai and everybody thinks chat
gbt right it's a chatbot moment but
what's coming through and what you're
saying is yes chatbots are unlocking
value for people and they're fun to use
but it's sort of going back to that
original discussion around chat GPT
which is that's a demo that sort of
shows the
capabilities and the rest of the um
industry and you know in Ai and Beyond
is starting to find ways to apply that
technology in intriguing ways so can you
talk a little bit about how AI is
starting to spread today beyond that
chatbot use case yeah look I think it's
interesting that that the release of
chat GPT sort of helped in some ways and
hurt in others I think the way it helped
is really spurring a huge amount of
imagination around what might be
possible for these Technologies and
really getting boards and executive
teams thinking about what might be
possible
the way it hurt is I think it sort of
over represented how mature these
Technologies are and how ready they are
for Enterprise applications and as you
said really shaped people's perception
around what AI really is which is only
just sort of one form factor in
imagination and so I to me you sort of
got to think about it much more
holistically and I think we're starting
to see that with some of these
multimodal um interfaces where you can
use video to sort of film the
surrounding around you and start to
interact and ask questions and you're
engaging in multiple modalities
simultaneously which is much more
natural for hum the same way you and I
are going back and communicating and and
seeing each other and interacting with
the world around each other so I think
these sort of multimodalities I think
the next Frontier of that is going to be
AI agents agents that rather than just
talk to you and go back and forth and
and chat can actually go do work on your
behalf can take down a complex problem
plan out the work that it wants to go do
and execute beyond that and ultimately
to me and this is a bit of a sort of
kick back to our name invisible
Technologies it's really about making
the technology invisible making it
seamless and integrating it within our
life I don't know about you but you know
the proliferation of all these different
apps and Technologies over the last few
years has only made things more
complicated um when the whole intent was
to streamline things and so I think
these technologies have the the ability
to actually deliver on that promise and
create seamless experiences for us
across our personal and corporate
life okay so two things first on the
agent front how close are we to seeing
agents in practice because I keep
hearing about agents and every time
someone cites an example of an agent to
me I go and check it out and it's just
like it's a it's a chat GPT rapper but
we're we're talking about real agents
here that actually are helpful how far
away are you from that
look I think this is going to be
something which is a gradual Evolution
that you're we're sort of pretty close
to having agents that can do discrete
things sort of one step twep sort of
planning exercises and that you're going
to need to build into that in terms of
how much trust people build into these
how sophisticated they are in terms of
planning multi-step um things that have
a sort of overall Arc and then a number
of sub threads underneath them that will
take time I think we're going to see
that sort of develop over the next
couple of years even if you just think
about something relatively simple like
planning a trip or a holiday or
something like that sort of starts with
a very conceptual Arc of where do you
want to go the type of experience you
want to have and then drills down into
you know booking flights booking Ubers
booking hotels planning activities and
all of those you know individually a
discrete actions that you could
potentially delegate to an agent what we
haven't really seen is the ability to
sort of hand that meta problem and have
the AI go sort of oversee and execute
that work handle changes in the in the
face of disruptions those types of
things I think were probably a couple of
years away but you'll start to see I
think a couple of these sort of killer
use cases that handle a more discrete
part of that problem really really well
in the next few years and is that
happening already I mean I know that
you're working with companies to put AI
into their operations right so is that
agent-based it look more like the Apple
intelligence example where it sort of is
aware of what's going on and gives Smart
Suggestions I mean talk a little bit
about how companies are integrating this
into their operations today and what
you've seen so far yeah I'd say most of
the use cases that end up people start
with is much more on the efficiency side
of things it's like okay can we take um
something and reduce the cost of that by
50% or can we take an employee and make
them 50% more productive and you're
seeing that with co-pilots and all sorts
of Technologies
I don't think you sort of broadly
speaking you're seeing companies
fundamentally rethink how their business
might exist in a sort of AI first world
and I think we're going to start seeing
the first round of of startups that are
sort of AI native that are really going
after that from from day one and you'll
see a real Divergence in outcomes in
sort of more traditional Enterprises
those that actually cross this Chasm and
and make that transformation and that
those that sit on the sidelines I think
really going to be left
behind right and so can you talk about a
few let's say concrete examples that
you've worked on uh that you can share
exactly how this works for a company
that's looking to integrate it yeah um
one of the the sort of really
interesting use cases that we've been
going through recently is working with
um one of the sort of top tier
investment manages where they obviously
um investing billions of dollars of
capital around the world they have very
sort of high Talent expensive cost base
and they're looking at saying how can we
improve the throughput of this sort of
very profitable Enterprise and one of
the ways we're looking at that is saying
look they've got this huge body of
knowledge that they've built up over
Decades of investing all the Investments
they passed on how do they take that
knowledge and actually build a sort of
superb brain that consumes all that
knowledge over Decades of work and
empowers the first year analyst to be
able to exit access all that um
knowledge and and Analysis and apply it
to a new deal or apply it to a new
opportunity and so you're not talking
just here about efficiency gains you're
talking about a fairly transformative
experience where you can basically take
a inexperienced firste analyst and give
them the power of someone with Decades
of experience and hundreds of deals
under their belt um and scale that
globally and so these are one of this is
a really interesting exciting use case
that sort of been in progress over the
last few months that's really
transformative this is not just 5 10%
better here this is like a fundamentally
different way of trying to build a
business like this wait so do they are
they like speaking with let's say a
knowledge bot that will sort of explain
to them how to view the deal or how does
this look like in the actual interface
yeah there's a couple of different
modalities of this where they their um
technology is capable of actually doing
research of analyzing a company in
coming back with a point of view around
what are the exposures what are the sort
of positives within that company and
then yeah i' I there's one example where
before going to an investment committee
each member of that investment committee
actually has to engage with a Bard and
ask its questions and go through the
verification process as as part of their
overall investment committee
decision-making process and this this is
a pretty fundamental shift in how
companies are thinking about the
application of these Technologies
what about a more boring example I mean
are you guys working on like automation
within back office because let me tell
you there's this quote that I was
looking at today from bendi devans I'm
just going to pull it up he goes there's
an interesting difference between people
outside Tech nearing at generative AI as
chat Bots that get things wrong and make
crappy stolen images and people inside
Tech who are mostly working on using it
to automate huge numbers of boring back
office processes inside Giant
corporations for billions of dollars so
both Benedict and I we're not using
boring here as a pejorative we're like
actually using it as and I spoke with
Matt Wood about this at AWS it's a way
to um this technology ends up taking
over some of that sort of root work and
making room for more inventive work so
I'm curious if that's something that
you're you're seeing as well yeah I
think this is honestly the most exciting
um opportunity here not just at a
personal level how do we shift people's
um work allocation from 20% sort of
creativity new ideas and 80% execution
the other way around how can we spend
80% of our time on more creative thought
processes and have ai sort of deliver
the execution side of that I think
there's very similar parallels on the
sort of Enterprise and corporate side I
think one good example of a sort of more
boring use case if that's what you want
to call it is we worked with the um one
of the largest um e-commerce retailers
has over 100 million products in their
catalog and were suffering from really
poor data quality around those items how
they were categorized how they were
tagged the descriptions that was really
impacting their their Topline revenue
and so we went through a process of both
using generative models to help classify
and recategorize that information but
also importantly humans in the loop to
QA and oversee that and cut what would
have taken a team of 100 people 10 years
to actually work their way through that
backlog um cut that down to a number of
months and so that's why you're seeing
these sort of
transformative um exponential changes in
the curve of what it takes to go do
these things that it just weren't even
practical you know it's by the time you
got to the end of the 10 years all of
those items were probably Obsolete and
being replaced by other ones and so
you're really seeing these things that
just were not possible even a couple
years
ago and is that a multimodal project
does like the model input uh the images
and then of use text to describe them
and things like that yeah that's right
we sort of the initial versions of it
were only text based but as you sort of
start to think about that more broadly
there's you know C color categorization
there's size there's all sorts of things
you can do as you start thinking about
the multiple modalities of information
that's
available yeah so we've kind of talked
around this we have this very powerful
technology and we have people and there
is a challenge right in integrating
human capital and Technology
because you know it always sounds great
in a PowerPoint but the change
management side is hard and U there's
also fear among people about okay so I'm
going to hand over you know maybe this
this project to to AI you know maybe
that 10 it's I mean it sort of sort of
explains some of the problems in the
workplace today but maybe that 10year
project was my paycheck for 10 years so
talk a little bit about this integration
between human and Ai and how you see
that playing out
yeah look I think the the best place to
start is to go to the Future right go 10
years from now and imagine where we're
likely to be I think if you do that it's
pretty obvious that AI is coming in a
pretty meaningful way and is not going
to slow down and wait for anyone and so
the the option to sort of sit on the
sidelines and see this out I think is a
bad overarching strategy and so then the
question becomes how best to sort of
integrate this and work through it and
what I'm seeing is I think the best
organizations are really starting to
think about this in a much more
integrated way I love sort of coming up
against organizations where the head of
technology and the head of sort of
people and talent are coming together
and they're starting to think about
these are just capabilities and some of
these capabilities come in artificial
form and some of these come in human
form how do we think about them
holistically and bring them together and
so there's sort of like an
organizational structure and philosophy
around capability acquisition versus
sort of talent and technology and then
there's the platforms that enable you to
sort of shift work between those and
that's an area that we have invested
very heavily in an invisible around how
do we promote a world which really does
think about these two capabilities in a
much more hybrid integated way and so we
build a platform from the ground up that
allows us to harness the best in Ai and
automation but also supports you know
over 5,000 experts in every topic and
domain that you could imagine so you can
match those capabilities very
fluently yeah so how did those experts
then integrate with the platform yeah so
we um we sort of manag the end to-end
life cycle of the of finding those
people breaking down their skills and
capabilities think about them sort of
like Badges and certifications of all
the different things that you could go
do and then have you know a matching
algorithm that's not too dissimilar to
an Uber or something like that that's
sort of saying okay we need this process
or this piece of work needs these type
of skill sets who's available and people
can basically opt in and do that piece
of work they're trained and certified on
on that piece of work we manag that end
to-end process with quality control and
oversight all on our
platform oh that's cool could you would
I'm curious could you see AI taking over
like the role of an expert in the future
yeah I mean the way I sort of let me say
taking on right not replacing but saying
hey like if we had these 5,000 experts
maybe now we can have 20,000 if we hire
quote unquote 15,000 AI yeah look you
you're already start I think we're early
in it but you're already starting to see
this where it's a I think about it like
a Glide path right where in the in the
early days you might have the humans
doing 100% of the work and then over
time you start to have the model maybe
suggest what the person might want to
pick or or come up with ideas in a sort
of co-pilot fashion
and so you've sort of got the human in
the driver's seat the AI being there is
sort of an assess and then you move to
the next phase where you've probably put
the AI in the driver's seat and there's
some still some level of human oversight
that's checking that that's quality
controlling they're handling the more
sort of extreme circumstances and then
at some point you get to a point where
the you're confident enough in the AI
and you're able to move the human
completely out and move them on to other
more complicated or more messy
activities and that Glide path is just
going to sort of basically work its way
through every process and there's things
that are easier to move through that
process and then others that will be um
take longer to fall but I think
ultimately everything is going to follow
some version of that
trajectory this reminds me a lot of what
I heard within Amazon because they've
been on the automation path for more
than a decade right and it's that uh
people who used to be keying in
information now they become Auditors and
people that used to be you know moving
something from column A to column b
instead of doing that they become taste
makers and people who eventually you
know in a company that's functional and
wants to compete who eventually have ai
take their work uh they go on to the
next inventive process and are able to
help the company reinvent and attack new
markets and the companies that do that
well it seems like they're the ones that
are going to have the upper hand in the
economy and the companies that don't do
it well will maybe be profitable more
profitable for a while
uh but then ultimately you know succumb
to competitive
pressure yeah I think that's 100% right
like I don't I think this is much less
about replacing humans and just sort of
repurposing our effort towards things
that we're uniquely capable of doing
this sort of invention and creative
process and actually coming up with new
ideas or novel solutions to to problems
and and stop us being sort of held down
by Boring sort of monotonous rot work um
which honestly machines are much better
at doing than we
are now Ben let me ask you this we've I
mean I talked about Amazon in the
previous question right they've been
implementing automation since 2015 by
the way they've grown their Workforce by
hundreds of thousands since so uh in
that case at least doesn't look like AI
is taking jobs but anyway um so we've
we've been having years and years of AI
implementation automation implementation
the most impressive technology that we
can invent putting it into the workplace
yet if you think about the lift on
productivity that we've seen when from
economists studying this stuff they say
it's been relatively muted I still can't
fully wrap my head around the fact that
productivity Remains Not static but you
know slowly increasing while this
technology that blows our minds is
available to us for free or low cost and
then being implemented in the workplace
what's your thought on that from
yeah look I think um there's lots of
different thesis around what's going on
here I think a big part of this is the
complexity and and switching cost that I
mentioned earlier the the proliferation
of all these different systems you know
it's the SAS tool for this the SAS tool
for this the SAS tool for that um each
one in theory is sort of solving some
specific problem and meant to streamline
processes whether it's for reviews or
payroll or Salesforce or or whatever
system that each person needs to use but
ultimately the work of sort of
navigating between those and shifting
between them ends up falling back on the
Enterprise and so I think a lot of the
proliferation of these Technologies has
added this huge tax of complexity and
switching class that the technology is
not yet streamlined that we haven't
actually seen a shift from increasing
complexity down into actual streamlining
when you look at what a person 's
day-to-day work looks like and that
switching cost ends up being a huge tax
when you look at what people spend their
time doing and I think the next wave of
AI transformation really needs to cut
that out to give you sort of a single
point of interface to really sort of
think about how does it wrap around you
as an individual or you as a
professional to supercharge you and
actually make your life easier and what
might the interface look like if we were
to get there like what does that
actually look like like in practice
because you still have these like like
you're talking about very thorny change
management problems human capital issues
so what if you like got a chance like to
dream a little bit and we'll allow you
the space here what does that look like
yeah look I think the most compelling
Visions for sort of modality are these
sort of
omnipresent um sort of over your
shoulder AI assistance that are always
there whether that's sort of through
vision and some of the goggles and some
of these ideas of sort of AR and VR or
just a co-pilot that's sort of sitting
um on your computer the whole time it's
it's interesting to sort of go back to
clippy and some of these early visions
of what AI might have been um that's
constantly watching and is there right
when you need it I think that the sort
of switching cost now is to pause
whatever you're doing and go type
something into GPT or type something
into Claude and that inherently is is
still a piece of friction versus or
something that's sitting over your
shoulder and saying ah I know what
you're about to do I've been watching
you know you're about to write an email
you most likely going to be writing
about this thing here's some suggestions
is that right and it's sort of ever
present there and just gently suggesting
nudging taking work off you coming back
and saying hey I thought you might be
interested in this and has enough
contextual awareness that that is sort
of seamless and delivered just at the
moment that you need it it's you know
it's oh it's been a couple hours you
probably about to look for a coffee shop
here's one around the corner and that
there's no sort of friction in the
process of actually requesting and
collecting that information um I think
it looks like that I think that it is
sort of always on always there building
up a better understanding of you and the
context around you and then delivering
really relevant just in time sort of
information or doing work on your behalf
before you even ask for it yeah and Ben
there's a important variable that I've
left out the whole conversation which is
the state of the labor market right I
mean in the US here we're almost at full
employment and unemployment is
exceptionally low and across the globe
it seems like the access to jobs not
everywhere but in many places is about
as good as it's ever been and I think
that means companies probably you know
can benefit tremendously from Gaining
technology that will increase the
average employees productivity and and
adaptedness and and um
top subject matter expertise in a way
that you know might be difficult to hire
for at this moment am I reading that
right or is that completely off no I
think that's right there's sort of two
things going on simultaneously one is I
think we've seen huge shifts in the
Global Talent markets both sort of
skills and and high quality Talent has
become much more Global than I think
companies have historically thought
about it there's a lot more people that
are moving into sort of fractional or
freelance work and that companies are
still thinking about you know hiring
full-time employees within 20 mil of
their office or whatever is missing out
on an increasingly huge percentage of
the overall talent pool and so the
companies that are rethinking the talent
strategy and how they access Talent are
getting um really significant advantages
there and then the second is is being
able to leverage these Tech AI
Technologies to really
democratize knowledge and expertise the
ability you know even if you go back to
that investment case the ability to
bring in a relatively Junior untrained
staff member and wrap them in an Iron
Man suit that gives them the expertise
and knowledge of you know a career-long
professional is really extraordinary and
you're seeing that in sort of high
Talent Industries like this but also
dealing with the sort of Aging knowledge
transfer of call centers all sorts of um
e expertise that are sort of moving out
of the workforce that companies don't
know how to replace these technology
provide finally a solution to be able to
harness and deploy that knowledge um
throughout their
Workforce okay last question for you we
focused on big technology on this
channel a lot but what about smaller
companies I mean it seems like if done
right this technology can help a small
company spin up pretty quickly and
reduce some of the weight that's needed
effectively to to get going and start
competing against bigger guys what do
you think about that this is probably my
most um exciting and compelling idea
around the real power of these
technologies that you know I think
would've heard Sam Alman talk about the
the one person unicorn um and so while
that might be you know a little
farfetched I do think directionally this
is very right that these Technologies
and companies like invisible will enable
you to build really powerful global
companies that are able to compete with
others that have you know thousands
hundreds of thousands of employees very
very quickly to be more agile to be more
Nimble and to operate with just a very
high density core set of talent that's
sort of overseeing and steering the
organization but is enabled by hundreds
of thousands of AI agents that are
basically doing the work and that can do
so very quickly and you've sort of
broken the Paradigm that sort of bigger
companies have more employees and
they're larger in size I think
effectively that will become a liability
that it sort of makes you less agile
makes you less able to adapt to changing
circumstances than these AI technologies
will yeah and well and as a oneperson
company I can tell you I'm looking
forward to the day where we can have ai
that helps us do a lot more with much
less so exactly right it's music to my
ears all right Ben if people want to
hear more about invisible Technologies
where can they find more about the
company yes so just go to
invisible. uh is the easiest way to get
us uh online and would love to help um
people transform their operations we
work with all sorts of companies from
the biggest companies in the world um
all the way down to small to medium
Enterprises and so would' love to help
companies on their AI journey and just
streamline and scale their business
operations Ben plumber CEO of invisible
Technologies thanks so much for coming
on thanks for having me all right
everybody thanks so much for watching
we'll have another interview up on the
channel shortly and we'll see you next
time take care