Innovative Applications of Generative AI in the Customer Service Industry

hi everyone thanks a lot I'm Olga I'm
the co-founder of dialogue shift and we
offer AI based chatbots for hotels so
this means Hotel guests can communicate
with the hotel via different messaging
channels and we automate most of the
communication with artificial
intelligence
about six weeks ago we launched a hotel
chat bot based on large language models
and in this talk I will share some
learnings that we had and I will also
share some general thoughts on the
future of customer service customer
experience and customer communication
but I want to start with a movie who of
you has seen this movie
okay I have to the ones who hasn't seen
it you have to you have to watch this
movie and it's a science fiction movie
from 2013 it's called her and the film
Paints the future where every person has
an artificial assistant a personal
artificial assistant in his or her ear
and this assistant can complete almost
any task it can independently write
emails make phone calls contemplate the
meaning of life and even fall in love
spoiler alert sorry
um six about six months ago we uh um a
chatbot that we haven't seen before
amazed the whole world it's a technology
that is different and it is different
than we have expected it
some call it the iPhone moment of
artificial intelligence
generative AI is already transforming
Industries it's changing our searching
Behavior our buying Behavior it's
changing our work life our private life
and it's going to impact our economic
system considering that the cost for
intelligence are going to drop to zero
one day
but I think for customer service
generative AI will do three things it
will
scale or automate customer communication
at scale it will make possible hyper
personalization and it will improve
customer experience overall
tools like childcpt so text and speech
based tools are based on large language
models those are models they are trained
with a huge amount of data they can
analyze text and generate text based on
predictions using statistical methods so
they basically generate or add one word
at a time
so the implications for customer service
are of course first the communication
and I want to share share some examples
from the hotel sector so hotels receive
about 1 000 phone calls or emails every
day of course depending on the size of
the hotel or the complexity of the hotel
product and the services but it's a lot
a lot of communication and most of the
queries are repetitive
um guests ask about when can I check in
can I bring my dog when is breakfast
served where can I park my car I mean no
no problem if you have a lot of people a
lot of Staff who can answer phone calls
and respond to emails but unfortunately
hotels don't have enough staff actually
80 percent of hotels in Europe and the
United States suffer from staff sorted
shortage so what can they do of course
they can automate this repetitive
communication with artificial
intelligence but do you remember that
those dumb chat Bots that couldn't help
you at all yeah
water under the bridge with generative
AI because now we are entering an era of
Next Generation chatbots and AI
assistants the interaction with those
systems are is amazingly fluid amazingly
human-like they can respond to quite
complex queries and to individual needs
of customers they are not limited to
specific topics so you don't have to
pre-train a language model with specific
information anymore but they can
basically
answer anything that is that is on the
internet and they can do something that
is very astonishing it's called
reasoning so they have a broad
understanding of time and size for
example so if I ask the chatbot
I'm arriving at 9 00 pm tonight at the
hotel is the sound is still open
and the chatbot will say yes it is still
open because our sauna closes at 10 30.
so the chatbot understands that 9 pm is
before 10 30. this is something that was
not possible before and the very cool
thing is you can feed the systems now or
the chatbots with unstructured data so
you can give them a website you can give
them spreadsheets PDF documents and
those documents serve as their knowledge
base for the specific information that
they have to to course to communicate
and I think in the future if in the near
future most of customer communication
will be handled by artificial
intelligence and we are not far from
being greeted uh by a hologram at the
reception desk in a hotel and maybe when
we call the hotel we have a nice lady
like Annie who can talk to us but it's
not a person it's artificial
intelligence
um so the question is will we be even
talking to humans in the context of
customer communication in the future I
think we will but before I get back to
that I want to share another very
interesting Insight from the hospitality
sector and it's conversion
about the about 50 percent of website
users abandon an online purchase if they
can't find answers quickly enough
answers that they need in order to make
their buying decision meaning if you
um if you have a quite complex product
and cast your potential customers have
questions before making their buying
decision the quicker you can provide the
information the more money you make
period
uh this is especially interesting in the
hotel industry because about 70 of
queries that are asked in the website
chat on a hotel website are booking
related 70 and here's a huge potential a
huge potential for making money when you
respond quickly
um so what happened when we updated
Hotel chat Bots with large language
models so what we did we started four
weeks ago we started an open Beta with
10 hotels and we chose the hotels we
chose underperformer on our platform so
hotels that had a lower automation rate
slightly higher live chat rate live chat
rate means that a human needs to jump in
into the conversation into and and
chatting live with the customer so we
chose 10 hotels we switched the AI
components so the products stayed the
same we only switch the AI component and
this is what happened
so we increased the automation rate by
10 okay this is something that we
expected
the live chat rate decreased by 20 so
this means 20 percent less people needed
assistance um from a human
and the cool thing is we increase the
conversion rate by 10 in our case
conversion rate means booking a hotel
room in your chat bot so 10 more people
booked a hotel room in the chatbot and
this is actually astonishing because
it's only the AI it's the AI that
understands humans or customers better
and therefore is is this the whole
product is much more powerful
so the challenges that we experienced
especially in customer service it is
crucial that the information
is correct so the information that the
chatbot communicates it's correct and
that it is consistent so the check-in
time is always at 3 pm and it is never
at 2PM for example but at the same time
so we wanted to to the chapter to of
course communicate specific information
but also to have general knowledge so
for example if I ask in Paris by the way
how high is the Eiffel Tower I want the
chaper to give this answer to me why not
it's possible Right but to balance that
specific information of the organization
and general knowledge from a technical
point of view was the biggest challenge
at the beginning
I think we handled it now quite well
with a very elaborated prompt design you
can ask me about that later if you're
technically interested but this was
quite challenging and the other
challenge were the high expectations by
our customers because they use chatgpt
and they expected that their the hotel
chatbot can do anything and we had to
kind of had to educate them and say okay
look it's you're still an application
for hotels it's based on this technology
but it's not LGBT it cannot write your
CV okay and they're like okay
um and then the next a big impact that I
see generative AI will have on customer
Communication customer experience is
hyper personalization
we have been dreaming about hyper
personalization for years now but now
finally we have this technological
technological shift where it is finally
finally possible so beginning with
search for example when I am planning a
trip to Copenhagen I don't have to work
through lists of hotels restaurants and
activities that might might match my
needs or not but I can simply use my own
language and tell the artificial
intelligence what I'm looking for what I
need I will get specific recommendations
I can iterate and I can create a tailor
and a travel experience that is tailored
to my needs and also if you think
further along the customer
customer Journey think about
recommendation systems think about
Dynamic pricing and offers think about
Predictive Analytics and think about
adaptive and intelligent spaces so
imagine AI is connected to the building
how cool is that
so if you think or if we look at
generative AI from a high or abstract
perspective and if we think about what
it can do very well so it can generate
yeah we know that by now it can
summarize very well it can classify or
categorize very well and it can predict
very well but it can is also
comprehensive and by comprehensive I
mean that they will have a language
layer over arching tools applications
workflows so AI will be able to perform
quite complex tasks using tools and
applications by itself like in the movie
and will we talk to people will we talk
to humans and in the context of customer
service in the future or will we only
talk to robots on the eye and I think we
will do both we will also talk to people
because the hotel of the future that I
Envision is an Apple store this means we
have a lot of people a lot of employees
and they have time for me they can focus
on the customer 100 because everything
in the background is perfectly automated
and the more automated a society becomes
the higher the human need for
interpersonal interaction thank you a
lot
[Applause]