Play
Show Navigation
Show Transcript
00:00
Play
Seek 10 seconds backwards
Seek 10 seconds forward
00:00 / 01:38:33
Mute
Disable Captions
Settings
Picture in picture
Fullscreen
Search in video
You can search the video captions for specific words or phrases.
Popout transcript
Hide Transcript
Skip transcript
All right.
Namaste, India.
[CHEERING]
I am Vishwa.
And I'm Aditya.
And we both, together,
welcome you to TDX.
A warm, very welcome to
our incredible Salesforce
developer community.
[CHEERING]
Come on.
You can do
better than this.
[CHEERING]
Yes.
This is the energy
that we are expecting
for a conference that's
come back to India
after six long years.
Awesome.
Are you ready for TDX?
[CHEERING]
We are as excited as
you are being, here
in Namma Bengaluru.
We are bringing you the
excitement of pre-keynote
with a touch of
interviews, fun,
as well as some surprises.
Are you ready for it?
[CHEERING]
And are you ready for
the best keynote ever?
[CHEERING]
OK, Vishwa,
let's start off.
Tell me this.
What has changed in
the past six years?
There are a lot of things
that change, Aditya.
For starters, we
both got married.
I mean, not to each other,
to our better halves.
Thank you for the
clarification.
Oof, OK.
Ha.
So a lot of things have
changed in pandemic.
But on a serious note,
a lot of things like we
went from work from office
to work from home, again,
from work from home
to a hybrid model.
And now we are back,
coming into work
from office.
Yeah, but talking
about the pandemic,
one thing that has
remained constant
is the energy of
the community.
So during the
pandemic, the way
the community
came together
in support of everyone
is unprecedented.
Like, the community has
done vaccine drives.
They have distributed free
food to all the families
that have been impacted
and really showed
the world what this
community is all about.
That's so true, Aditya.
It's amazing to see the
passion and the resilience
of what this community
brings together.
If I remember rightly,
when the pandemic hit us,
everything was closed.
And people were not
getting together.
But our incredible
Salesforce
community leaders--
if you are all the
community leaders,
please stand up,
and get recognized.
Get recognized.
If you are an MVP,
please stand up.
Please get recognized.
Golden Hoodie winners.
Golden Hoodie winners,
please stand up.
And not the least, but all
the community conference
organizers, please
stand up and get
yourself recognized.
So we have seen an
explosion of community
conferences last year,
which is how all of us
came back together after
being apart for so long.
Can you name
a few, Vishwa?
Yeah,
I think Bharat Dreamin,
Architect Summit
Yes, Architect
Summit, are you here?
No?
Mule Dreamin and
Delhi Dreamin.
Guys, come on.
[CHEERING]
Who else, Aditya?
Bangalore Dreamin.
Come on.
We are in Bangalore.
[CHEERING]
Hyderabad Trailblazing.
[CHEERING]
MC Square.
[CHEERING]
We're Dreaming.
See, this is a lot
of conferences,
and we want to thank
all of the community
for making this happen.
Let's give a big round
of applause for everyone.
[APPLAUSE]
All right, so talking
about conferences,
quick show of hands.
How many of you are here
for your first ever TDX?
Wow.
Amazing.
Trust me, you're
going to love this.
All right.
So again, show of hands.
Yesterday was a
holiday, and all of you
are here today.
So be honest with me.
How many of you
told your managers
that you're
working from home?
Show of hands.
[LAUGHS] OK.
All right.
Wave big and smile.
You're on camera.
We will ensure to share
it with your managers.
[LAUGHS]
Yeah, I'm sure there
are some pro developers
in this crowd as well.
And it's Friday.
And it's known
for deployments.
So how many of you have
brought your laptop
to do some deployments?
See?
I said you.
OK, yeah.
You might have
to roll back
if you're deploying such
a high-energy environment,
but nevertheless.
Yeah.
All right.
So thank you for
the response.
And typically,
when you talk
about a conference
like TDX,
it's not always
just a conference.
It's more of
a celebration.
So it's a celebration
of our product
that we all love, yes?
It's a celebration of all
the innovation that we do.
And most importantly,
it's a celebration
of the community.
And one way in which we
celebrate the community
is by sharing
your story out
there so that others
can be inspired.
And I see a
lot of familiar
faces in the crowd.
So, Vishwa, do we want
to go talk to them?
Let's get into the crowd
and pick a few people
to interview them.
All right.
I see people
tensing up already.
all Right, so we have
Sukesh over here.
Sukesh why don't you
stand up and come with us?
[CHEERING]
All right, so the reason
we picked Sukesh--
please face that side.
Smile for the camera.
All right.
By the way, are you
on PTO or working
from home today?
Actually, I'm both.
Oh, you're both.
Oh, you're both?
I'm doing both.
All right.
So the reason
we picked Sukesh
is he was back
here in TDX 2019,
but back then, he was
a student and a Journey
to Salesforce aspirant.
But now he's an integral
part of the Bangalore
community.
And he is encouraging
a lot more students.
So, Sukesh,
how do you feel
being back here in
TDX after so long?
First of all,
thank you everyone,
and I'm so thankful to
Salesforce Epic team
and especially to Vishwa.
So I have enrolled for
the [INAUDIBLE] program.
From then, my
journey started.
And here, continues
a long way.
So I'm working
at a [INAUDIBLE]
as a senior
software engineer.
Apart from that, I'm
a core team member
of Bangalore Trailblazer
Community group.
And I am marketing
and communication head
of Bangalore Dreamin.
So apart from
all these things,
this is the first
TDX I'm attending.
2019, I couldn't
able to attend
because I don't know what
is Salesforce and all.
But me and Bangalore
Trailblazer Community
group and
Bangalore Dreamin
has inspired many of the
Trailblazers across India.
And apart from that, I'm
so happy and glad to be
here.
And I'm so excited to
attend this conference.
OK, thank you.
That was more than
30 seconds, but--
[LAUGHTER]
So what are you looking
forward to in TDX?
Do you have your
eye on any session?
Any key takeaways,
networking opportunities?
First of all, I'm
not sure that I'll
be part of this
keynote, but I'm
so happy and
glad to be this--
here, talking in
front of everyone.
And so, probably,
I want to network
with all the MVPs
and product managers,
attend tech sessions,
and give it back
to the community as well,
so inspire Trailblazers
in the ecosystem.
That is why we
are all here.
All right.
Thanks a lot, Sukesh.
You were definitely
prepared.
Thank you, Sukesh.
Take your seat.
So what will you--
[INTERPOSING VOICES]
--Kavindra here.
[CHEERING]
Oh, I see Kavindra here.
OK.
Everyone excited?
[CHEERING]
That's all?
Is everyone excited?
[CHEERING]
Great.
Yes.
Kavindra, thank you
so much for putting--
helping us, putting this
amazing show for us.
You have been
in the ecosystem
for the last 18 years.
You have seen this
community grow.
You have seen
wonderful Trailblazers.
We just
interviewed Sukesh.
Do you have anyone
in your mind,
one person where we can
go and interview them?
Wow, that's a
tough question.
It's a tough question
because there's
thousands of
you, thousands
of amazing Trailblazers
that I can pick from.
But if I must,
there is one person
that comes to my mind.
This is the
Trailblazer that
has the heart one India
Trailblazer community, who
has jumped in to help the
community and Salesforce
and make an
impact for India.
Also, this is
a person that
has mastered AI and
Agentforce and, in turn,
has helped others
do the same.
This person has gone to
the meetups, Dreamforce,
TDXs and many
other events,
and really helped others
share their knowledge--
their knowledge and
expertise to others.
And this person-- when I
think about initiatives,
things that
come to my mind
are Trailblazer chain.
You [? greet ?] equals
passes to TDX Bengaluru.
And this person-- she
is Shreeya Rashinkar--
[CHEERING]
[INTERPOSING VOICES]
Why don't you walk
with me, yeah?
Aditya, let's get
her onto the stage.
Yeah, so Kavindra
is singing
all praises for you.
So while we walk
on the stage,
why don't you
start telling us
a little bit
about your journey
into Salesforce
and how you've
become an integral
part of the community?
So I think my heart is
really beating very fast.
So I might not be
in all of my mind
to speak what I
have been doing.
But I think
this is really--
I mean, it's very--
I'm so thankful, first
of all, Kavindra,
for all the sweet words.
It's very kind of you
all to be recognizing
a little bit of
efforts that I've
put in for the community.
So basically, one
thing that I've
been doing recently is
the Trailblazer chain.
That is something really
close to my heart.
What happened was when we
saw this TDX India coming
here, I thought,
obviously, everybody
is excited, right?
So I just wanted
to build up
that energy for everyone
to be a part of this.
And when-- I put a
message on LinkedIn
saying that I am
attending TDX.
The question should
be, who's not?
And that is where--
my hands are
literally shaking.
OK.
This generally
doesn't happen.
[LAUGHS]
And I actually got a
response from somebody
saying that I would really
want to be a part of TDX,
but somehow, I'm not
able to make it because
of XYZ constraints.
And that actually hit me.
I just thought we
being Trailblazers,
we being a part of
this community--
and we are doing better
for ourselves, as well
as our families.
Why can't we just take
out some time, be a part--
do something
better, and be
able to help people who
really deserve to be?
And that's where
I just started.
What you've done
is really great,
and you've been doing
lots of awesome work.
Thank you so much.
And since Kavindra has
been singing praises
for you, he has got you
a small, little gift
all the way from the US.
Kavindra, do you want
to present it to her?
Thank you so much.
We have a little
surprise for you.
I will hold it.
What it is?
Yeah, let's look
what the surprise is.
[CHEERING]
That's a Golden
Hoodie, guys.
Let's put it on.
[INAUDIBLE]
Bigger round of
applause, people.
[CHEERING]
All right.
There is one
more small token.
A Golden Hoodie Astro.
[CHEERING]
We are going to
take one selfie.
OK.
It's a selfie time
with the crowd,
so we can face this way.
Yeah?
OK.
Come this way.
All right.
Thank you.
Thank you so much.
Thank you so
much, Kavindra.
Thank you, Shreeya.
Thank you so much.
[INAUDIBLE]
All right.
This cover is for you,
in case you use it.
[LAUGHTER]
All right.
That was exciting.
And look at that.
It's almost time
for the keynote.
And before we
kick it off, it's
actually a
timeless tradition
that we have
where we light
a lamp before
starting anything
significant in our lives.
So in order to
do that, I would
like to call upon
the entire Salesforce
leadership that is here.
So we have Arundhati
Mam, wherever she is, MK,
Kavindra, Sanket,
[? Sanjana, ?]
[? Kristoff, ?] Kiran,
[? Parul, ?] Avantika.
All of you, please
join us on stage.
[TRADITIONAL MUSIC]
[? Harun, ?]
[? Kumar, ?] please.
No, no, no.
Please come with us.
[TRADITIONAL MUSIC]
No, no.
All of us.
This is an auspicious
occasion for us.
[TRADITIONAL MUSIC]
It will burn your hand.
[TRADITIONAL MUSIC]
All right.
So I would
request all of you
to come in front of
the lamp, please.
Mam.
Everybody in
front of the lamp.
All right, so the reason
that we light a lamp
is to signify that any
small spark of fire
can actually illuminate
a world of possibilities
for us.
And sometimes the
people themselves
are the fire that
open up possibilities
for a whole lot of people.
And we have one such
person among us today.
Arundhati Mam,
we want to-- we
have a small
surprise for you.
MK handing it over to you.
Is it on?
Yeah.
All right.
So you know there
is a tradition
in South Indian wedding.
After the couples
are wedded,
they're asked to look
at the Arundhati star
as a guiding light,
and also making sure
that-- how they lead
their transformed
lives together.
But we don't have to look
very far because there's
a star right here with
the same [LAUGHS] name.
[APPLAUSE]
[CHEERING]
Yes.
She's been trailblazing
for over 30 years
and really leading the
digital transformation
of many businesses
in this country.
And you can see
here the government
of India recognized her
exemplary service, awarded
the Padma Shri.
Join me in
congratulating Arundhati
for this awesome award.
[APPLAUSE]
Congrats.
Thank you.
[INTERPOSING VOICES]
[APPLAUSE]
And [INAUDIBLE] for you.
That's the Mysore silk.
That's her favorite sari
and colors, too, so.
[LAUGHS] Congrats.
Congratulations, Mam.
We couldn't be
more proud of you.
You want to say
a few words, or--
Oh, microphone.
Can I-- mic?
Can we get the mic?
First of all, good
morning, all of you.
And thank you
ever so much.
This was actually
something
that they kept
as a surprise.
I was told there is
a little moment when
we won't tell you
what the program is,
which is very unlike them.
We run through every
minute of the program.
So I guessed
something like this.
But-- well, this is well
beyond what I had thought.
Thank you again very much.
I'm really and truly
grateful and honored.
And believe me, it's a
moment of great humility
that I have because to get
something like this, not
only from the government,
from the country,
but also from the
organization that I'm
serving--
it's a huge amount
for anybody.
So I'm truly grateful.
Thank you.
[APPLAUSE]
All right.
So thank you--
[INTERPOSING VOICES]
Thank you so much,
everyone and everyone.
All right.
So I think it's time for
the most awaited session
of the day.
What is it?
The keynote?
Are you all ready
for the keynote?
Yeah!
Keep your phones on.
Silence.
Make sure that you're
looking at the speakers
and the screen.
And get ready
for the keynote.
Let's get the
keynote started.
[CHEERING]
So who are the
Agentblazers?
Well, now, that's a story.
Trailblazers--
they've led the way.
People who saw a
mountain and said,
hey, I'll climb it.
The people who saw a
challenge and said,
hey, let's solve it.
They were pioneers,
innovators, visionaries.
But today, today,
something new
is on the horizon.
It's not just about
blazing a trail anymore.
It's about building
a world where
humans and digital
agents work together,
where empathy
meets efficiency,
where intelligence
is autonomous.
We call them Agentblazers,
a new breed of problem
solvers who cut through
the shiny hype of AI
and answer the real
question of what
AI can actually do.
I mean, what if
workforces had no limits?
What if customer
service could
feel like a conversation
with an old friend?
What if businesses
could make smarter
decisions in real time?
Agentblazers see the
future not as something
that happens to them, but
rather, something they
can create.
Now, look, we've
been dreaming
about agents for years.
We've seen them
in the movies--
robots, avatars,
futuristic assistants.
But this?
This is different.
This is real.
Agents who learn
your business.
Agents who know
your customers.
Agents who can
take action right
when you need it most.
Agentforce is about
empowering people,
freeing humans to focus
on what really matters.
Yeah, our stuff.
And the best part?
The admins,
the developers,
the dreamers in
the Salesforce
ecosystem-- you're the
ones turning this vision
into reality, because
Salesforce didn't just
build this platform
to automate tasks.
No.
It was built
to empower you
and people like you to
become Agentblazers.
So who are the
Agentblazers?
You are the Agentblazers.
And the really cool part?
The future?
It's already
here, right now.
[MUSIC PLAYING]
Please welcome
President and CEO
of Salesforce South Asia,
Arundhati Bhattacharya.
[APPLAUSE]
Good morning, everyone.
Good morning again.
I hope you are
excited to be here
because I am thrilled.
And yes, I did hear
[? y'all-- ?] the last
time somebody
asked you the same.
And I hope that this
day and tomorrow
will come up to your
expectations and way more.
Having said that,
you will all
agree that the pace
of AI innovation today
is like nothing we
have seen ever before.
And that's all
the more reason
why we need to be at
events like TDX Bangalore.
So let's get right down to
this amazing technology.
But before doing that,
in the true Salesforce
tradition, let
me first express
my heartfelt gratitude and
my thanks to all of you
all for turning up here.
To all our customers,
our Trailblazers,
our partners, and of
course, our employees,
you are at the
heart of it all.
And it's because of
you that we are here.
Now, having said,
this is a company
that has always been
guided by its values.
The values, by
the way, are not
something that's just
written on a plaque.
And especially today
in the age of AI,
where along with the
kind of excitement
it has generated,
it has also
generated a lot
of mistrust--
and trust, by the way,
is our greatest value.
It has to remain that
because we deal with data.
And that data, by the way,
is something that's yours.
It's not ours.
It's not anybody else's.
We need you to
believe that we
give equal importance
to it in order
to make our
customers successful
and to innovate
on a daily basis.
This is what
really drives us.
And that's the
kind of promise
we made 25 years ago.
A simple promise--
1% of our time, 1%
of our products,
and 1% of our
profits to go
to the nonprofit sector,
to the community.
And you can see the impact
that it has made globally.
Not only are we a
zero-emission company.
But today, if you look at
the impact only in India,
we have given more
than $5.6 million.
We have contributed
622,000 volunteer hours,
and more than 800
organizations today use
our products.
Not only that.
What I'm really and truly
happy about today is
the fact that we are
declaring a grant
of $300,000 to Mantra for
Change in order to change
the way that we are
actually going to deliver
teachers' training.
And this, by the way, is
going to impact 44,000
schools.
[APPLAUSE]
Therefore, I have
absolutely no problems
in saying that I am
really proud to belong
to an organization
that is one
of the most innovative,
most philanthropic,
and most ethical.
It's what that
has driven us.
We truly and
really have tried
to imbibe the
motto of business
as a platform for
change, and we live it
every single day.
Having said that, what is
Salesforce doing in India?
Do I really believe it's
making an impact in India?
Starting from one
small office, today,
we are a huge
organization,
with more than 13,000
direct employees
and contributing $1
billion in revenues.
Look at the kind of logos
that we are working with,
logos that are
household names, that
are known by all of you
all on a daily basis,
logos like Air India,
Asian Paints, Royal
Enfield, Emami, Godrej.
You name it, and we
are working with them.
Along with that,
we have also
created a huge
Trailblazer community
of as many as 3.9
million people.
And by the way, it's
these Trailblazers
that are at the cutting
edge of innovation
and helping us innovate
on a regular basis.
Innovation, by the way,
is driven by the way
technology changes.
Imagine the coming
of the internet
for the first time.
We could get stores that
were offline online,
and we could make them
available for customers
24 by 7 from the
comfort of their rooms.
Come the cloud
age, and we learned
an entirely new way of
how we could access apps.
And today, we
have Agentforce.
Agentforce will do
for the workforce what
the other kinds
of technologies
had done in order
to totally change
the way we do our work.
Agentforce
Is going to impact
the workforce
like never before.
But tell me, why
is that required?
It's required because
customers today
have learned to
expect much more
than what we could
ever deliver earlier.
They don't want
any hold times.
They want to be
attended to 24 by 7.
They want their
issues resolved
as soon as they call.
They want to
talk to experts.
But over and
above all of this,
they also want
somebody to be
empathetic and
understanding
of the problems that
they are facing.
But what about
the companies?
Are they capable
of taking this on?
Their workforces
are limited.
There are complexities
in many of the questions
that have to be answered.
There is actually
burnout in so many
of their people.
And overall, they are
finding their productivity
is dropping.
And this is actually
out of a survey that's
actually shown that
productivity is indeed
dropping.
Now, with circumstances
like this,
what is it that we need?
What then, if workforces
were not limited?
If our workforces had
no limits, probably,
that would be the
way to get the best
answers for all of this.
And that's exactly where
Agentforce comes in
with its digital
labor platform.
This, by the way, is a
trusted, deeply unified,
easy-to-deploy
platform that's
also open and integrates
with almost everything
that you do.
Agentforce, by the way,
does not only assist.
It can also act.
And therefore,
companies can actually
meet their customers
where they are.
This is really the
requirement of today,
and it's something that
can impact every business,
in every industry.
And therefore, there's
a quantum change
in the way the
workforces can actually
learn to behave.
Imagine if your
employees are
able to solve complex
issues with the assistance
of Agentforce.
Not only that.
Even in their own work.
Say, for instance,
an employee
uses Agentforce to assist
them in actually working
on their own health plan.
Imagine a time like that,
when the employees become
truly empowered
and truly enabled.
This is indeed
something that
can bring about a very
big change in the way
customers experience
the businesses that they
interact with.
But let's not only
talk about theory.
It's good to talk
about actual examples.
Take, for instance,
Hero FinCorp.
You also know that it's
a non-banking financial
corporation that
actually gives loans.
But like all
other businesses,
they are cyclical.
There are times when the
requirement for loans
are really, really
high, and there
are other times when
there is average demand.
Now, how do you
scale your workforce
to take care of
the peak levels
at the same manner
in which you
are taking care of
customers on average days?
Well, it's possible, it's
possible with Agentforce
because with the
help of Agentforce,
they've actually brought
down the number--
the amount of
time they take
to approve the loans from
as much as two-plus days
to just 30 minutes,
a reduction of 99%.
Take another instance.
Look at Air India.
Air India, as you know,
is flying a huge number
of our customers.
In fact, I flew
here in Air India,
and it was a lovely
flight, for that matter.
But the fact of the matter
remains that when there
are issues in flights--
and there could
be on account
of technical
reasons, on account
of climate, which happens
very often these days--
the number of issues
that come up for solving
are huge, whether it be
rerouting of passengers,
whether it be refunds,
whether it be upgrades,
and all of these
are complex matters.
Agentforce is
being used by them,
and they are envisaging
using it more and more
in order to resolve
all of this.
But don't take
my word for it.
We have here somebody from
Air India, Vikram Singh,
and I am very
pleased and honored
to call him up over here.
And you can listen
from the horse's mouth,
so to speak.
Please don't mind me
calling you a horse.
I'm not calling you one.
That's just a
turn of phrase.
But please, come
up over here,
and tell our people here--
tell our customers.
Welcome.
Tell our customers what
impact do you really
see, Vikram, of
Agentforce in the work
that you all are doing?
Oh, well, Mam,
Agentforce is actually
transforming our customer
service completely--
not only the
things that we do.
We are completely
reimagining
all the processes because
it is a complex operation,
ultimately.
The recent example
is the refund
that we tried and
solved with Agentforce.
What used to take weeks
is now taking just days.
And it's mainly because
one of the human
is still involved
in-- because it's
a transaction of money.
But I am very,
very confident
that in about two weeks or
so, once we are completely
confident that
Agentforce is doing truly
an immense job in
calculating and doing
the right thing,
we should be
able to do it in minutes.
That would be
absolutely wonderful.
Imagine getting
your refund back
in a matter of minutes.
And beyond that,
Vikram, tell me,
how do you see
Agentforce supporting you
in your vision
for the future?
Well, Air India
vision is to become
world-class airlines.
And our
technological vision
is also to become the best
technological airlines.
And I think Agentforce
resembles of what we
are trying to do.
I think it will
not only empower
us to reduce
human efforts,
but also rather
increase our presence
with digital agents,
along with human,
to give larger and
better experiences
to the community,
to the consumers.
And I think it will
fast-track our vision.
Thank you very much.
Thank you, Vikram,
for coming and coming
here and talking to
us about your company.
And believe me,
this time, when
I flew Air
India, the people
on the flight, the crew--
they were so good and
so polite and so--
they were so--
what should I say?
Giving so much
of attention
that I was really
and truly moved.
And I hope as
Agentforce takes off
much of their
drudge duty, they
will be able to create
that human connection
far more than they were
able to do earlier.
But with that,
it's time now
for me to call on stage
Muralidhar Krishnaprasad.
He is our President
and CTO for the unified
Agentforce platform.
Huge round of applause
for him, please.
Thank you.
Welcome, Murali.
And thank you.
Thank you for making
it all the way.
Thank you,
Thank you.
All right.
As you saw what
Vikram showed you
and as you saw what
Arundhati talked about,
digital labor is here.
Now, two years
ago, when we all
talked about
digital labor,
the talk was,
you know what?
Just use an LLM.
Just throw all your
documents at it.
It's going to
answer everything.
But the reality is
a little more harder
than that.
Why?
if it's public
data, it's easy.
You can just go
to ChatGPT and ask
any question for the LLM.
That's good enough.
But in an enterprise, you
have a lot of private data
and a lot of private APIs.
So it's not the same.
So the LLM is more like
the CPU in your computer.
What are you going to
do just with the CPU?
You need a hard drive.
You need peripherals.
You need an operating
system to run it.
And that's what we are
building with Agentforce.
It starts with your data
and metadata layer, where
you want to be
able to bring all
the data that you
need in enterprise
and make sense out of
it with the metadata.
And then you get
the triple-A layer--
the apps, analytics,
and all the metadata
around it-- the semantics.
It's sales.
It's marketing.
It's service, it's health.
And all the APIs to
be able to connect
to all your systems--
SAP, Microsoft, whatever
system that you may have.
And then comes
the operating
system-like
layer, which is
going to then use the LLM
to be able to say which
APIs to call, what
data to go retrieve,
and then use the
power of the LLM
to reason and answer.
But all of this, just
like in a computer,
comes in a box, is
wrapped in our platform,
and the platform provides
all that security,
the metadata
and all the CLI
and the developer
tooling experience on it.
But as you all know,
in any enterprise,
you just don't
have one platform.
You have a lot of
different things.
So it's not just about
having a trusted platform.
You also need an
open platform,
which is why with
our platform,
we have opened and made it
extensible at all levels
of the stack.
That starts with
the lowest level,
where we make our platform
available in all geos
possible.
At the data level, we
spearheaded the industry's
first zero-copy model,
where we can share data
at a file level with all
of the other data systems.
And of course,
for our APIs,
with our power of our
platform with MuleSoft,
you can virtually connect
to anything that exists
in your enterprise.
And finally, if you have
your own trained LLM,
you can bring that,
too, very easily
with our platform.
And that's why
we believe we
have the industry's
leading Agentforce
platform.
It starts off at
the MuleSoft layer
for all APIs, our
Salesforce platform that
gives you all
that developer
experience, the
Data Cloud that
binds all the data
in your enterprise
and with all the
applications, [INAUDIBLE]
the ones we ship or the
ones that you custom
build with analytics.
And then Agentforce
pulls all
that together to
give you that agent
experience for every
aspect of your business.
Now, this is all
architecture.
So then your
question would be,
how does it actually work?
The way it works
is very simple.
You have a trigger.
A trigger could
be somebody
coming to your website
and asking a question,
or a trigger could
be an email coming
into your system, or it
could be some data change,
some predictive ML that
predicted your customer is
going to enter.
It doesn't matter
what it is.
The trigger basically
calls into Agentforce.
Then the Agentforce
reasoning engine steps in.
It understands
what the intent is.
And then it has access
to that entire corpus
of data, APIs,
and actions.
And then it figures
out what to do.
And then it uses that to
actually do the action.
And the outcome
could be an answer
to your question.
The outcome
could be calling
an API to create
the refund, as
in the Air India example.
But the core heart of this
is that Atlas reasoning
engine.
And we believe we have
a differentiated product
here.
Why?
Often, you will hear
these terms, RAG,
and everything, thrown in.
People would just
say, you know what?
Just throw some
documents at an LLM.
It's going to answer
all your questions.
What we learned is it's
not that easy because what
you have to do-- just
because you have some HTML
documents or others,
it's not good enough
for you to get the
answers because you
need to prepare them,
make them agent-ready.
That means really
going through
that pre-processing stage,
chunking it, enriching it
with more content so
that you can go retrieve,
because you might have
a really long document.
And if you have
a question,
you're not going
to go pick up
the entire document.
You need to pick
up chunks of it.
And that's on
the right side
with the power of our Data
Cloud unstructured stack
that you can get.
Again, that's not enough
because what you then
need to do is as the
questions are coming in,
you need to go
figure out which
chunks match your answers,
then reiterate on it.
Just like if I talked
to [? Hari ?] here,
I'd be like-- he
asked me a question.
I'd be like, oh, did you--
is that what you meant?
I refine those
questions and try
to get the answers.
Same here is what our
planner and reasoning
engine does.
And that together
constitutes
the powerful
Agentforce platform.
Now, what we have seen
with our customers,
like Air India--
Vikram talked about it.
We've also seen customers
like Vellore see huge
advantage, like, 75% more
accurate results and 16x
faster.
And we've also seen-- we
have now-- by the way,
we sold over 5,000
deals last quarter.
And we have thousands
of customers live.
And what we have
seen all of them
is go through
a progression.
The first phase
is really having
your agent answer simple
questions in your corpus.
And that's things like,
OK, what's your question,
answer, pull up the
answers from the corpus.
We call them
knowledge agents.
And then you evolve.
Now you trust it.
You [? move on ?]
to the next stage,
which is action agents,
which are things
like take actions
on your behalf.
And then if you
trust it even more,
the third phase
is where then you
make it autonomous
so it can actually
work on your behalf.
It can crunch
those documents.
It can answer
emails automatically
for you to do your
SDR, BDRs, and so on.
That is the
journey that we
have seen customers take,
and one such customer
is ourselves.
Now, if you go today
to help.salesforce.com,
that's something
that we actually do.
So you might ask that
question, how do you start
this agentic journey?
Now, on the left
side, you have
traditional companies,
traditional companies
with hard-coded workflows,
with lots of siloed data,
hallucinating AI.
And this right side is
this awesome Agentforce
company.
What does that mean?
You've reimagined every
aspect of your business
to be AI-first, to
be agent-powered.
That means you have a
unified understanding
of who your customer
is and every touchpoint
that they may have.
But then there is
a gap in between.
We call that the
chasm, right?
Jumping the chasm.
How do you jump
that castle?
For decades we've
gone through
various transformations.
I think Arundhati showed--
you started with the
internet, the cloud,
the mobile, the SaaS.
There were so many
different revolutions.
In every one of
them, it's been
you, the Trailblazers,
who really connected
that chasm, who
jumped and made
your company successful
in that transformation.
And in this
transformation,
it's going to
be you again who
are going to make that
leap for your company.
And so we're going
to call you not
Trailblazers anymore,
but, really, Agentblazers.
You're going to
blaze the trail
and really agentify
your company--
not just add AI to
your company agenda,
but really reimagine every
aspect of your business
and make it better.
And to make that
easier, you've
now started a new
community of Agentblazers.
I invite all of
you to join it.
And we also have a
lot of new material
available for you.
And as developers here--
I think a lot of you
brought laptops, too--
your question would be,
OK, can I play with it?
Is it just some slides
that I'm seeing here?
Well, I'm very happy
to announce that we now
have Agentforce Developer
Edition available for all
of you.
And if you note
that small--
I kept telling them
to put it bigger.
It's free.
As long as you keep
using it every 45 days,
it's free for you.
It includes the
entire platform
that we talked about,
all of the Data Cloud,
Agentforce.
And you can play around.
You're going to see
a lot of demos today.
You can create all
of those yourself
and go convince your
boss why you can become
an Agentforce company.
Now, in order
to also do that,
I'm going to take you
through three steps.
First, we're going to
understand, what does it
mean to agentify
your work--
all of your workflows
and everything else?
And second, we're
going to actually see
how you're going to
go build and deploy
these agents,
and finally, how
are you going to
share everything
you've created
with the community.
On the first part--
now, let's dig down.
I kept talking
about agentic AI
is changing software.
What does it mean?
Now, when you build
a traditional app,
what do you do?
We all go figure out, OK,
what relational database
I should use, what
logic I should
do with flows and
other kind of things,
deterministic.
And then you spend
a lot of time
building a high-purpose,
high-cost, probably, UI.
That's our traditional
software development life
cycle.
But now, on
the other side,
on the agentic
software, it's
going to be both
structured and
unstructured data.
It's not just the forms
that somebody fills.
It's going to be all the
PDF documents, the video,
audio, multimodal,
all of that.
And in addition,
you're going
to start having
nondeterminism
in your workflows because
somebody is asking you
a question to you.
It's not like a
fixed workflow
that you can actually
write for it, just like we
have human conversations.
And you're going to
also have the ability
to create multimodal and
almost near zero-cost UI
that's personalized
to every one
of your experience.
And like I said
earlier, we, ourselves,
have tried this now.
If you go to
help.salesforce today,
in fact, the entire page
is just an agent there.
We've actually thrown out
all the website before.
In fact, we think
that's going
to be the future
of website itself.
Why do you need a website?
You just have a
box that can answer
all your questions.
Imagine the change
it's going to cost?
Now, what we saw in just
a matter of a few months,
once we have
implemented it--
we have seen over 32,000
conversations a week.
It's going to
go up even more.
50% reduction
in escalations--
50%.
Now, it's not just
about the cost savings
we're going to get
on the other side.
But imagine the
customer satisfaction.
They're not waiting
to talk to somebody.
Just ask a question.
You have a problem.
You got the answer
right there.
And just like
that progression
I said, we started
with a knowledge agent,
and now you can
also take actions
like filing cases and
other things as well.
And this happened by
indexing over 700,000
documents that we have.
But that's not all.
Just like your
enterprise, Salesforce
is also equally complex
internal implementation.
We had three
different orgs,
a sales org that the
sales team manages,
a service org that the
service team manages,
a marketing org the
marketing team manages.
And what we used
to think-- we had,
like, 250 million
visitors to our website
all the time.
It turns out it
was only 140,000.
And people just
had different IDs,
different email,
and so on.
And so what we
did with the power
of our Agentforce
platform--
we implemented
Data Cloud there,
and now we have a 360
view of everything
that you do on
the website.
We know exactly
who clicked,
who watched video,
and everything else.
And that
information, using
the power of Data Cloud
One in our Agentforce
platform-- we can project
to all these orgs.
No copy-- one single
copy of your data
projected to all the orgs.
And now we have
multiple agents
running-- marketing
agents, sales agents,
service agents, all seeing
the same copy of the data.
Or if you call your
human and you're
talking to them, they
also know exactly what
you've done across
all touchpoints
of your business.
And that is the
power that we
are able to achieve
with Agentforce.
But coming back,
we always see
this as a chat interface.
You're still talking.
You're doing this.
But what if it can
actually extend
beyond the chat window?
What if it can
actually go--
like I talked
about, really
have multi-purpose
UI and so on?
And that's why I'm happy
to announce Agentforce 2dx
today.
[APPLAUSE]
What Agentforce 2dx
will allow you to do?
It replaces
that rigid logic
with dynamic reasoning.
Now you can add agentic
commands in your flow.
That means-- imagine
you have a flow--
you know the
traditional flow, right?
You start here.
You put all the boxes.
One of that box can now
be an agentic experience.
So you're going to
be able to bring
in nondeterministic logic
within deterministic
logic, or vice versa.
The other thing
you can also do
is this thing
called variables.
You can actually set
state in your agent
and really go change
the reasoning logic.
The second thing
is it enables
you to take action.
With MuleSoft
Topic Center,
with one single
click any API--
you can now make it
available in an agent.
That means you can
talk or English
to your IBM mainframe.
And with the
power of Heroku,
you can write in
any language--
Ruby, Python, whatever
you name it-- and make
them available there.
And with the power of
Data Cloud's action,
any change in any data?
You can now
trigger an agent,
take response to it.
And finally,
on the UI side,
we are going
to be allowing
you to actually reuse
all the lwc components
that you created.
Now you can use that
in the agentic output.
You're going to see
Tableau, next, in the demo
soon, where you're
going to actually see
high-five analytics,
actually, part
of your agentic
responses, and many more.
Now, you heard--
many, many times
we've been repeating
Air India-- how it's
an Agentforce company.
But we're going
to actually show
you live how they
implemented Agentforce.
And to do that,
I'm calling
Avantika on the stage.
[CHEERING]
Thank you, MK.
All right.
Good morning, Bengaluru.
So thrilled to be
here all the way
from our HQ in
San Francisco.
And I must say, my journey
here was extremely smooth,
thanks to Air India.
Just as Arundhati
experienced,
I also had a very
smooth journey
and this personalized
white-glove experience
as I was traveling
on my way here.
And as Vikram
mentioned, Air India
is continuing to
elevate that experience
even further.
They're poised to become
the world's best airline,
with technology
at the forefront.
Now, when we think about
agents and Agentforce,
a lot of us are
typically used
to interacting with
agents on a website
in the context
of a chat window.
But as MK just
showed us, we
are starting to bring
in brand new paradigms
of interacting with
agents with background
agents, voice
agents, and servicing
these agents in
other channels,
like mobile and Slack.
So to show you how Air
India could totally
Elevate their experience
with Agentforce,
we're actually going to
show you this completely
live.
And to do that,
big shout out
to our demo drivers,
Nithin and Benny.
So let's get right in.
[APPLAUSE]
All right, so here we
are in Salesforce Service
Cloud.
And we see the
customer profile
for the Khanna family.
Now, if we look
closely, we
see the basic contact
details for our traveler,
Payal Khanna.
But now we go beyond
those basic details
and also include
information
like her loyalty
status, the fact
that she's a
platinum member.
And you'll notice that we
even have a new insight
here, that she's
traveling with
a six-month-old infant
for the very first time.
Now, beyond these
details, we're
even able to
bring in live data
like the flight status
history, the booking
events.
We can even see the reward
transactions and web
engagement data, all
thanks to Data Cloud
pulling in this
real-time information.
Now, this is a lot of
rich context and data.
What if we could
add an agentic layer
on top of this and
actually connect this data
to Agentforce?
Well, to show you
that, let's go straight
into our Agent Builder.
So here we are.
This is Agent Builder.
And we have a various
group of agents here.
And we've clicked into our
customer support agent,
which is meant to
help our customers
with their flight
and booking data.
Now, if we look
into our agents,
it has multiple topics.
Let's double-click into
the airport data topic
right here.
Now, this topic is meant
to assist customers
with any inquiries related
to the airline and airport
operational data.
And you'll notice
that this topic
has several actions.
One of those
actions includes
retrieving information
from Data Cloud.
We have another
action that's actually
going to gather
insights from Tableau
and render those.
On top of these
topics and actions,
something brand new
that we've introduced
includes context
variables.
Now, we talked
about introducing
some level of determinism
into nondeterministic
workflows.
We can do these
with variables.
So over the course
of the conversation,
you can actually save
valuable information
in these variables.
And it's valuable
because you can actually
start creating filters
on top of them.
So these filters include
rule-based logic for you
to determine how
you want to route
to specific topics
based on the value
of that variable.
For example, we may want
to validate a customer's
account ID before we get
into seat assignments
and go through
that whole process.
So now that we've looked
at how we construct
this agent, my
favorite part
is actually simulating
this agent right here
within the builder.
What I can do is I
can ask questions
from the perspective
of a customer.
For example,
Payal can see,
if her flight from
Cochin is canceled,
what her options are.
And the agent gathers
valuable information
like her loyalty
information, some
of the flight
itinerary data,
and offers a very personal
and actionable solution
for her.
At the same time,
I can interact
from the perspective
of an employee.
I can ask about insights
about flight delays,
and this agent is
going to gather data
from multiple sources.
And we're able to leverage
the power of Tableau
to not only provide
these insights,
but also visualize it in
this beautiful rendering
right here in
that chat window.
Now, this is powerful
functionality
is pretty amazing.
But you might be
wondering, how
does this actually work?
We have all this
transparency
built in to how
the agent works
in this middle column
right here, which
is our plan tracer.
You can see exactly as
the customer identity is
passed in, we
start by retrieving
important account details,
like the fact that Payal
is a valued Air
India customer.
Then we're able to fetch
real-time information
from Data Cloud, like
the flight and booking
information.
And finally, we pass
that entire payload over
to Tableau.
And Tableau is
what helps us
generate those insights,
summarize them,
and visualize them
like what you see here.
All right.
It's pretty amazing that
we can conversationally
interact with
these agents.
But what if you want to
take it a step further
and have these agents be a
little bit more proactive?
Why do I always
have to type
or click a button to
invoke these agents?
Why not just have
them be a little bit
more proactive and
react based on an event?
Well, that's where we
have background agents.
Background agents
are able to be
a lot more proactive.
You can invoke these
agents, headlessly,
through an API, a flow,
a scheduled event.
You get to decide
how you want
to invoke these agents.
Now, the best
way to show you
background
agents in action
is with our beloved
Flow Builder.
So let's head
into Flow Builder.
Now, for those of you
who are flow experts,
here is an example of a
flow that we have here.
And what it does is based
on a flight itinerary,
it's able to react
to different changes
in the flight status.
Now, I mentioned experts.
You all probably have
experienced flows before.
It's probably very clean,
elegant and simple,
like what you
see here, right?
Nithin, why don't we zoom
out just a little bit?
Just kidding, I'm sure
those of you who've
used flow know that it
can get quite unwieldy.
I mean, what we see here--
that's a lot of logic,
branches, decision trees,
and conditionals.
I mean, this works
perfectly fine,
but it can get hard
to maintain, scale,
and even debug.
Now, what if
we can instead
leverage agents to
collapse this logic
and make it a lot simpler?
Well, thankfully, now we
can actually embed agents
directly into flows.
That's right.
So right here, we actually
have the exact same flow
that's reacting
to a status
change in the itinerary.
But instead of all
that complicated logic
we saw earlier, we are
collapsing all of that
into a single decision
node by the agent.
All we do is we pass in
the context in the form
of a prompt to the agent.
The agent's already
equipped with a whole host
of actions it can take
based on the situation.
And there we go.
We have this intelligent
reasoning node right here
within the flow, rather
than building all
that hard-coded logic.
Now, I actually want
to show you this
in action in the
context of Air India's
mobile application.
Now, what we've done is
we've set up this flow
to be triggered
as soon as there's
a cancellation in
the flight status.
So let's go ahead
and run that flow.
Now, when Payal's
flight gets canceled,
the agent is
immediately triggered.
And what it does is it
gathers important context
about the flight
cancellation
data, Payal's
flight itinerary,
and alternative
itineraries.
And look, she gets a
personalized notification
informing her about
the cancellation.
She gets more context
about what's happening.
And you'll notice that
the agent knows that she's
a valued member,
so it's immediately
helping her get rebooked
into the earliest flight
possible.
So the agent looks at
the next alternative
itineraries and
renders them
in this very
seamless UI component
that you see here.
So now Payal, at the
click of a button,
can reschedule
her itinerary,
and she's all set.
Now, I'm sure we can
agree this experience is
probably far better
than anything we've all
experienced before
with flight delays
and cancellations.
Now, I don't
know about you,
but when I'm in a
frenzy at an airport
with a flight
cancellation,
I don't even have a
single second to type.
Well, guess what?
Agentforce can also
work with voice,
through Agentforce
Voice, so I can actually
interact with Air India
directly through voice,
and I'm actually
going to show you
what that looks like.
All right.
Thanks, Nithin.
So here I am.
Hopefully, you all can
see what I'm showing you.
I am in the Air India
mobile application
and now I want to interact
with their customer
service team
through voice.
So I'm going to go
ahead and call them up.
Hello, Payal.
I see that you
have recently
rebooked your flight.
Is there anything that I
can help you with today?
Hi, yes.
I just got rebooked
and I can't
seem to find my gate.
Your flight is departing
from gate 36 at 4:00 PM.
Would you like me to send
you a map to the gate?
Yes, please.
I need it pretty fast.
No problem.
I will send you a
link in the chat now.
Wow, look at that.
Isn't that amazing?
Huge round of applause for
this wonderful experience.
[APPLAUSE]
I mean, look
at that, right?
A new way of interacting
with customer service.
And Air India is really
championing and pioneering
this new customer
service interaction
paradigm in the whole
airline industry.
Now, we just saw
the customer service
experience and
what the passenger
sees on their end.
You might be curious, what
about the Air India team?
What about the employees?
What about their
side of the picture?
Well, this is where the
Air Force employee agent
comes into the picture.
Now, this agent has a
slightly different set
of capabilities.
First, within
these actions
we want to connect it to
third-party airline APIs,
and thankfully, we can do
that with the MuleSoft API
catalog.
Now, once we connect
to those APIs,
the next thing
we want to do
is render these
for our employees
in a very visual
manner through Tableau.
So here, we're going to
go ahead and generate
those semantic
layer insights.
And finally, our Air
India teams normally
collaborate in Slack.
This is where all
the action happens.
So we want to continue
surfacing these insights
from the agent in Slack.
So now let's go ahead
and add all those actions
to our agent, and let's
see this live in Slack.
So here we are in Slack.
And as soon as the
flight status changed,
our agent automatically
created a Slack channel
and added all the
right individuals
to this channel, including
the ground teams,
the crew, the customer
service teams.
It summarized
what's happening.
It shows a live
visualization
of the rebookings.
It knows who on
the ground team
is assigned and gives
them a checklist of action
items.
Now, you might
have remembered
Payal had interacted
with our voice agent
because she's
looking for the gate.
Well, our agent
has created a case
to tell the team that
Payal is on her way
to the gate
with her infant.
And now, as you can see,
[? Rohan ?] [? Mehta ?]
from the ground team is
informed that he needs
to be at the gate,
patiently awaiting her
arrival, and the Air India
team is there to support
Payal and her infant as
they onboard their flight
to ensure a very seamless
and pleasurable journey.
And there, my friends,
you have seen,
through Payal's
journey with Air India,
how Air India is
delivering this connected
and elevated customer
experience with the power
of Agentforce and
Agentforce 2dx.
Thank you all so much
for following along.
And with that, I'm going
to pass it back to MK.
Thank you.
Wasn't that amazing.
Now, you saw a lot.
She went through
a lot of stuff.
And every single
thing-- by the way
all these demos are really
running live on an org.
So nothing fake here.
You saw Agentforce API.
You saw how easy
it was to connect
all the existing
APIs into that stack.
You saw Tableau next.
You saw the power of
Data Cloud bringing
all that data there.
And you also saw
how we could bring
that complicated flow.
With structured
deterministic logic,
we were able to now plug
in nondeterministic logic
as well.
That is how you can
agentify every app
and workflow.
And because you have all
the Developer Edition
available to
you, I'm going
to challenge
every one of you
to go try these things.
Everything is available
for you to go try.
But then you're going to
ask the next question--
what is that life cycle
of this agentic app
experience?
What do I do?
Now, it's not like your
traditional software.
We talked about
that before,
because there's a lot of
nondeterminism involved.
You can't even
test it to see
if it's returning the
same text because it's
going to be different.
And so we have devised
a new life cycle
for agentic app dev.
It starts with the
ideating and the plan.
Now, you saw a lot
of those agents
being created.
Of course, now you can
use the Developer Edition
to go try it out
yourself, or you can also
use the production
orgs that you may have.
But the build part is
the interesting one.
Now, it was a
complex agent.
Now, instead of
trying to actually
hard-code and create those
topics and everything
else, we've really
simplified it.
We built an agent
for an agent
so now you can
just use that AI
assistant to actually go
build your whole agent.
So you can literally
just type English
and say, [INAUDIBLE]
create the agent for me.
Now, in addition to
that, you, of course,
have all the configuration
of Agentforce.
You can also use
the APIs and all
the unstructured stack.
And with Heroku
app link, you
can write your actions
in any language
of your choice.
Now, once you have
done the build,
the next step is testing.
How do you test this
nondeterministic thing?
For that, we have
added Testing Center.
With Testing
Center-- again,
you can use AI to
generate all the tests
that you want.
Double check
that as people
are asking questions-- is
it doing the right topic?
Is it doing the
right results?
And then post-testing,
you want to go deploy.
And you're going to see
that in the demo shortly.
We have, now, this
new DevOps Center
where you can use either
visual or CLI to go deploy
all of these agents.
And finally, once
you've deployed,
it's not just
once and done.
You want to continuously
observe the agent.
Maybe somebody is trying
to break your agent.
Somebody is giving
you toxic inputs.
You want to be
alerted on it.
So we have alerting
built right in.
We have full Interaction
Explorer that can actually
tell you, what kind of
topics are people asking,
and which ones are-- is
the agents able to answer
well, and which ones they
are not able to answer?
And it's always on top of
every one of our mind--
cost.
Is it costing
too much or not?
And so that's why we have
digital wallet for you
to be able to
really understand
what the cost that's
going on with Agentforce.
With all of this, you
now have the new 2dx life
cycle for agent building
and testing, deployment,
and monitoring.
And in order to
see that, I'm
going to invite
Christoph to the stage.
Christoph?
[APPLAUSE]
All right.
Thank you so much, MK.
And before we
start, please
welcome our amazing demo
drivers, Aditya and Benny.
[APPLAUSE]
All right.
So I want to go
back to Air India
because they have an
amazing customer loyalty
program, but they
want to build an agent
to make it even better.
And what that agent
will look like,
we don't know yet.
We need to ideate.
Remember?
The first step of the
agent development life
cycle.
That is why we start
in a scratch org,
because it's the perfect
place to experiment.
And to get you
started, we have
some amazing new tools.
First, templates-- they
let you create all sorts
of agents really easily--
for example, an SDR
agent or a service agent.
But then my
absolute favorite--
the brand new AI
Assistants, because here,
you don't have to go
through all the steps
of creating an
agent manually.
You just describe what
you want the agent to do.
So in this case,
we're going
to say, hey, we want
a customer loyalty
agent for Air India.
It should help
customers manage
their points, book
reward travel,
and that kind of thing.
And just like
that, AI Assistants
created the agent for you.
It has a name.
It has a role.
All that looks good.
We can move to
the next phase.
And here, you can see that
AI Assistants is actually
generating topics.
These are the jobs to
be done for the agent.
For example,
manage points,
book reward travel,
like we just asked.
And if we dive into
one of these topics,
you will see that
AI Assistants even
generated actions,
things like,
get points
balance or verify
redemption eligibility.
Now, the first one
is actually a flow,
and the second one
is an Apex class.
And the point
here is that you
don't start from scratch.
We are actually using AI
to scan through your org
metadata and find the
flows and the Apex classes
that you exposed
as actions,
and that can be part of
the agent implementation.
Next, we can review
the topic description
and instructions, and
all that looks good.
So next, we can
choose the channels,
and we want that agent to
work in the Air India app,
in Slack and in email.
Awesome.
So now we can save this.
And the last
thing we will do
is to actually add
unstructured data.
So we have some
PDF files that
describe the rules of
the loyalty program.
You can upload them
here in Data Cloud.
But because we
did it earlier,
we can just select
them here in the list.
And now for
the big reveal,
we can actually
click Create Agent.
And just like
that, we have
built a fully
functioning agent
in just a few minutes
with AI Assistants.
Who wants to try it?
Let's try it?
Yes?
Let's do it.
Let's ask a question.
Yes.
How can I combine
points with my family?
Good question.
So now, the agent
is reasoning
over all the
available actions,
and there is your
detailed response.
Now, how did
the agent do it?
Well, let's take a
look at the planner.
And first you see that
it selected a topic--
Manage Points.
That makes sense.
And then it selected
a bunch of actions.
And after it picked
the right actions,
the agent is
able to find out
what's the best way
to actually present
that answer to the user--
in this case, a
[? rich ?] card.
And the good news
for us developers
is that we don't
even have to code
that card because
the agent is
capable of generating rich
components dynamically,
things like a card or
a carousel or a forum.
And this is going to
work across channels.
WhatsApp, web?
It's all going
to work there.
And best of all,
this feature
is available today.
You can enable it for your
agent with that toggle,
Enable .
Adaptive.
Response Formats.
Awesome.
So now, we created an
agent with Agent Builder.
And we all love
graphical tools.
But I know that many of
you also like the CLI.
Is that right?
All right.
And that is why
we just added
an agent-first module
to the Salesforce CLI.
So Aditya, how do we work
with agents in the CLI?
Absolutely, Christoph.
So CLI fans in the
crowd, give me a cheer.
[CHEERING]
OK, I love it.
So we are going to be
working with the CLI
in our beloved IDE,
which is VS Code.
And what we are
going to do here--
we are in a
Salesforce project,
and we can run
the new agent
commands using
the terminal
right inside VS Code.
So let's start
it off by running
our first command, which
is to create an agent.
So what you see here
is very similar to what
you would have
seen on the UI,
where you can
select from a bunch
of different templates.
But I'm not going
to create the agent.
Instead, I'm going
to retrieve the agent
that Christoph
has just created.
For this, I'm going
to run the command sf
agent retrieve and pass
in the name of the agent
that I want to retrieve.
Now, what this does
is it pulls down
all the metadata
about the agent, which
includes not just
the agent itself,
but the topics, the
instructions, the actions,
and puts it all inside
VS Code for me to either
modify, commit to
version control,
or even deploy to
further environments.
Now that I have my
agent inside VS Code,
I can make
modifications to it.
So what I want to do now
is add a new custom action
to my agent.
Now, in my project,
there's a bunch of code,
and I have an
Apex class that
lets me redeem points
for lounge access.
Now, how can I create
an action created
based on this Apex class?
So for that, no
surprise, it's
going to be
another command.
In this case, it's going
to be generate action.
And here, I am passing
the name of the agent
and also the name
of the Apex class
using which I want
to create the action.
So I give the
action a name.
And what you're
seeing here
is that an action
is created based
on the inputs
and outputs that
are present in the
invocable method
inside my Apex class.
So all of that is built
right into the CLI.
Awesome.
Well, thank you, Aditya.
And now we
created an agent
with low code and
then pro code.
And it's time to move
to the second step
of the agent development
lifecycle, which was?
Remember?
Testing.
Testing.
Thank you.
All right.
And testing
agents is going
to be very different
because agents
are nondeterministic.
So Aditya, how
do we do it?
All right.
So agents can be
tested from the CLI
using what we call an
Agentforce_CLI_Test_Suite.
And within my
project, I've
already created one
such test suite.
And to run this
test, I'm going
to use the command
sf agent test run,
and pass the name
of the test suite.
And while the test runs,
Christoph, why don't you
walk us through the
test suite itself?
Yes, so I want you to take
a look at that test case
here because it looks
very different from
a traditional unit test.
You are not passing
traditional input
variables.
You are actually
passing an utterance.
And that is
natural language.
That's a question
that a user may ask.
And the outcome is also
not a traditional return
value.
It's also
natural language.
It's the description of
what a successful response
should look like.
Now, notice, too,
that you can test this
with a single utterance.
Why?
Because people
are going to ask
the same thing in
many different ways.
So we need to parse
multiple utterances here.
So Aditya, how did
we do with the test?
Let me check.
Looks like we passed.
We passed.
OK, that's
totally awesome.
But what if you are
low-code developer?
You want the same testing
capabilities, right?
And that is exactly why
we created testing center.
So let's take a look.
And this, if you
want, is just
a graphical UI on top of
the same testing engine.
This is why we can see
the tests that we just
ran in the CLI.
But you want to be able
to create your own test.
So let's do that.
Let's create a new test.
We're going to
give it a name.
We're going to
select the agent,
provide a description.
And now, remember, we
need to pass the test
utterances, right?
So what you could do is
upload a CSV file and type
them all in.
But that looks
like a lot of work.
So we can also get AI
to generate these test
utterances.
So let's do that.
And because AI is doing
all the work we can ask
to generate 2,000 of them.
So we're going to have
a great, solid test.
And now we can go ahead
and generate that test
case.
And there we go.
We generated test
with low-code.
So far, we built an agent.
We tested the agent.
It's time to deploy it.
And for that, we're going
to go to DevOps Center
because this is where
admins and developers work
together to manage the
entire release process.
And the first thing that
we are going to do here
is pull the changes
from the scratch org.
And this is really
my favorite part
because here, you
realize that everything
we did so far--
the agent, the
instructions, the topics--
it's all metadata,
which means
that it can be part of
the exact same process.
So now we're going to
commit these changes.
And then we can
promote this
to a staging environment,
because guess what.
We already have a
fully functioning agent
that your business
partners, the people who
ask for that agent, can
start experimenting with.
That means that we
need to start keeping
an eye on that agent.
What is it doing?
Is it doing the
right thing?
Which brings bring
us to the last phase
of the agent development
life cycle, which was?
Observe.
And observing an
agent is going
to be very different
as well because there
are millions of users
who are going to ask
for millions of things.
So how do we
even make sense
of that massive
amount of data?
Well, we have a
new tool for that,
and it's called
Interaction Explorer.
So here, we are looking
at an agent that's
already in production.
And you can see the
number of sessions.
You can see the
number of users.
I can see all my topics.
So let's dive into
one of these topics.
And once again,
we are using
AI to tag and categorize
all these conversations.
So if we dive into one
of these categories,
I can see all
the conversations
in that category.
And if something
doesn't look right,
I can dive even deeper at
the conversation level.
And here, I can see
the number of turns.
I can see the latency.
I can see the
quality score.
And on the
left, I can even
see the
conversation itself.
So there is a ton that I
can see here to understand
what my agent is doing.
But there is one thing
that I cannot see.
There is one thing that
nobody in this room
can see, and that is
private customer data.
And that is a good
thing because developers
like us-- we shouldn't
be able to see that.
So we need to talk
about data governance
for a second.
And the good news is
that you are already
protecting your data
in many different ways
in Salesforce.
You tag PII fields.
Maybe you create
permission sets,
sharing rules.
Maybe you're using Shield.
And the good news is that
all that applies when
you are using agents.
But we also added
a lot of data--
unstructured data,
conversational data.
That is why we just
launched new native data
governance capabilities
in Data Cloud.
As a small example, you
can create a data masking
policy that defines
exactly who can access
your data, when,
where, and how,
which means that if we go
back to that conversation,
that data masking there?
You have full
control over it.
And that is the agent
development life cycle.
That is how you create,
test, deploy and observe
agents with Agentforce.
Back to you, MK.
[APPLAUSE]
Wasn't that amazing?
Great job, Christoph
[INAUDIBLE].
Thank you.
Now, what you saw--
Avantika
demonstrated what it
means to agentify for
your app and workflow.
What Christoph and
Aditya showed you--
how are you going
to use AI itself
to go create all of that?
We saw how to create
the agent itself,
how to use AI to create
your testing, how
to use CLI, and of course,
Agentforce Explorer
to really understand who
was using Agentforce.
Now for the
final, third step,
we're going to be talking
about how you can now
extend this with
an open ecosystem.
Now, if you really
look at Salesforce--
I talked about,
earlier in my talk,
how open and extensible
we are at all
levels of the stack.
At the data level, we
are open and extensible
because we have the
zero-copy model that we
have partnered with a lot
of industry-leading data
vendors.
At the
infrastructure level,
we can now run on
multiple substrates.
We have our first
party substrate, AWS.
We had Alibaba for
our core stack.
And we just announced a
partnership with Google,
where we will be
porting all of this
to run on the Google
substrate as well.
Then comes the LLM.
We already support
OpenAI, Azure OpenAI,
and Anthropic models,
and you can also
bring your own
custom-created LLMs
as well and use
it in actions.
Finally, on the right
side, we have AppExchange.
That's been
our premier way
in which you can share all
the things that you have
created with the community
and make some money as
well.
All of you helped us
create what AppExchange
is today.
We have over 7,000
partners and over 9,000
apps and expert ratings.
And 91% of our
customers use a partner.
And that's why what was
just one million installs
in 2011 is now
13-million-plus installs
in 2024.
Along the way, AppExchange
supported all of it,
from sales to service to
marketing, Tableau, data,
and everything else.
But now, in this
agentic era,
we are also going to be
announcing AgentExchange.
AgentExchange is
the way you're
going to be exchanging
your Agentforce for all
your topics and actions.
So if you publish
all of those things
that you saw-- all
those awesome demo.
You can now
publish all of this
into the AgentExchange.
Now you can easily
use them in your agent
as you're creating it.
You can also
share and sell
to other people
in the community.
Now, in order for
us to actually see
how this works, I would
like to invite Parul
to the stage.
Parul?
[APPLAUSE]
Thank you.
MK.
And isn't it
incredible, everyone?
Today we have seen how Air
India is using Agentforce
to help their customers,
crew, ground staff,
and even their
support teams.
You know what?
They have a vast
network of partnerships,
and it can be very complex
and arduous to manage
it all.
So let's see
how our employee
agent can help Air India
manage their partnerships.
Before I dive deep
into the demo,
I want to give
a huge shout out
to my demo crew,
Aditya and Benny.
[CHEERING]
OK, everyone, let's start.
Here we are in a case
in Service Cloud.
Air India is trying to
onboard a new partner
as their
corporate partner.
We need to
create a contract
and send it to them
for e-signature.
Let's head over
to Agentforce.
We'll ask our agent
to create the contract
and send it over.
I am also wondering what
the agent will come up
with.
It's using its
logical reasoning.
Hmm.
There we go.
Our agent was able
to create the order,
but it was not able
to send the contract.
Wondering why?
This is simply because
right now, our agent
does not have
the capability
of the e-signature
workflow.
So what are our options?
Option one-- spend
endless hours and time
building and debugging
this e-signature workflow
from scratch.
Any takers there?
I don't see many hands up.
This is a wide audience--
wise audience, Bengaluru.
So let's go ahead
with our option
2, which is to find
a trusted solution
from a trusted partner.
And where do we find
trusted agentic solutions?
AgentExchange.
Here is our newly
launched AgentExchange.
[APPLAUSE]
We already have hundreds
of topics, templates
and actions all
ready to go.
You can even install
one directly from here.
But you know what?
It gets even better.
Now you can access
AgentExchange directly
inside the Agent Builder,
right in the flow of work.
Remember?
I wanted to add
a new action.
So let's go there.
And look, at
AgentExchange directly
inside Agent
Builder, everyone.
[APPLAUSE]
So let's go
ahead and search
for an action that can
create the contract
and send it for us.
Now, this is an AI-powered
intent analysis search.
And there you see--
DocuSign comes up as the
first recommendation.
Let's deep dive
into that and see.
I can learn more.
Look at the reviews.
I can also go through
all the topics
and the actions.
This looks perfect
for my use case.
Let's go ahead
and install it.
Now, what this
will do is bring
in the metadata
in your org,
just like any
other package.
We see the topics.
We're going to skip those.
Let's go to the actions.
Let's bring in all
of these actions
and add them to our agent.
Once these
actions are added
we are all ready to go.
So let's go, once
again, to our agent
and ask it to
create the contract
and send it again
for the e-signature.
It's using all the
logical reasoning
and sifting through the
new actions to empower us.
And there we go.
The agent was able to
create the contract
and send it over.
[APPLAUSE]
Let's quickly look
at the plan tracer,
how this got done.
So first, our agent
was automatically
able to query the order
to get the details.
After that, it
was successfully
able to invoke the
Create Contract action.
And then it invoked the
e-signature workflow
all by itself.
That's the power
of AgentExchange.
Now you can extend the
capabilities of your agent
with powerful and trusted
actions from ISVs directly
in the flow of work.
I am sure you
are all itching
to create an agent,
so let me show you
how to package and share
your agent for that.
Let's head over
to VS Code.
Here is the airline
Maharaja agent
that we just created.
You can see all
the metadata,
the agent class itself,
all the metadata
about the Apex classes,
templates, and flows.
So now we can package
it easily using
the Agentforce CLI.
Let's go to command
line and give
our first command.
We will create a
template for our agent.
And in simple
steps, there you go.
The template is created.
Now, let us create the
second generation package.
Again, a very
simple command.
And boom, the
package is created,
and that's all you need
to create your agents
and share them
within your company
or even on AgentExchange.
So everybody, this
is your moment
to create your agents
and share it out and live
and create history.
Whether you are a
Salesforce admin
or a seasoned
developer or you
are starting
your journey new,
you can use the expertise
and your creativity
on AgentExchange today.
So let's build, share
and innovate together.
Back to you, MK.
[APPLAUSE]
That was awesome, Parul.
Thank you.
Again, you saw
a lot that Parul
demonstrated right now.
You saw AgentExchange.
You saw topics.
You saw packaging
and how you can
make all that available.
Now, just to conclude
what we saw--
we saw three things.
We saw how to agentify
every app and experience,
how to use all the new
AI-based tools to go
create those
agents, and how
to extend with the
open ecosystem.
So to wrap it up, I would
love to invite Arundhati
back on the stage.
But before she says--
I thought it would be good
for us to take some selfie
before all of you
start disappearing.
How about all of
you, one, two three,
shout Agentblazer,
and then we'll
take a selfie
together, all right?
1, 2, 3, Agentblazer.
All right.
Back to you, Arundhati.
Wow, wasn't that
absolutely amazing?
Believe me, I'm equally
amazed, if not more so.
What do you all think?
A shout out for
all our presenters?
[CHEERING]
And now let's get hands-on
with the technology.
Your first stop
is Trailhead.
And the first
stage in Trailhead
is the Champion
stage, where
you'll learn all the
fundamentals of AI
and Agentforce and
Cloud, of course,
the Data Cloud piece.
But that's not enough.
You need to then go on
to the really exciting
Innovator level.
That's where you can begin
to use the technology
on all of your use cases.
You can use Agent Builder
and build your own custom
agent.
You can also deploy
Agentforce for all sales
and service cases.
And of course, how can
you not then aspire
for the Legend stage?
Here, of course, you
will need Specialist--
Agentforce Specialist
certification
to unlock this.
By the way, our
aim is to build
one million Agentblazers.
And it all starts with
you and you and you--
in fact, with all of you.
Are we willing to take
on that challenge?
Yes!
Louder one.
Yes.
And we will do it.
So with that, I would love
to now call upon Christoph
because this is a very
special moment when
we declare the winners
of our hackathon.
Christoph,
please, come here.
[CHEERING]
And by the way--
Thank you.
Thank you so
much, Arundhati.
On Tuesday and
Wednesday we
had an incredible
hackathon.
In fact, it was the
largest Agentforce
hackathon ever,
and it was here.
Where else will you do
this, other than in India,
by the way?
The innovation, the
passion was incredible,
but I know that
you all want
to hear the finalists,
so let's take a look.
All right, so
the runner ups--
let's start there.
Automates.
Congratulations,
Team Automates.
We had honorable mentions.
The first one is
for the Best Demo,
and it goes to Warp Force.
Congratulations
Warp Force.
The second one was
for Best Use of Slack,
and it goes to
[? Three ?] agents.
Congratulations,
[? Three ?] [? Agents. ?]
And the last one was for
Best Use of Data Cloud,
and it goes to
Team [? Axel. ?]
Congratulations,
Team [? Axel. ?]
[CHEERING]
But now the big moment--
I know that you all want
to know the winners,
so I'll need a
little bit of help.
A drum roll, please.
[DRUM ROLL]
And the winners are Team
[? Cortez ?] Alchemists.
Congratulations.
[INAUDIBLE]
Please join us on
stage to be recognized.
[DANCE MUSIC]
Congratulations.
Congratulations.
Congratulations.
[INTERPOSING VOICES]
[DANCE MUSIC]
Wow.
OK.
Just one minute before
you all move off
with the prize money.
Tell me.
Tell me a little
about your team.
How did you all
come together?
And what's your
team all about?
Please, any of you.
Great.
Thank you.
I'm really excited here.
So we are from a company
called Bounteous.
Our team name is Bounteous
AI Chemist, which kind
of looks like, alchemist.
We envisioned a
solution where
we wanted to create AI
concierge service that
cuts across all
the industry.
And we have taken a
hospital-- sorry, hotel
situation here,
and that's it.
Well, that's exactly
the second question
I was about to ask you.
What exactly did
you build in order
to get the first prize?
I think you have more
or less described it.
So congratulations.
Congratulations again.
Thank you.
Very well-deserved.
Thank you, Thank you.
Thank you.
Someone can take this.
And with that, we come
to almost the close
of the keynote.
But let me tell
you, Trailhead
is not the only place
where you will learn.
Please understand
that being
part of this community
is a huge learning
experience.
It's something where you
learn from each other,
where the knowledge is
shared, where you do group
learning, and
it's great to be
just part of this
whole huge community
and to exchange
with each other all
of your experiences.
Join us at many of
these community events
that you are
seeing up here.
Join us at some of
our world tours.
And believe me, you will
find your time well worth
what you are spending.
And with that,
we would just
like to remind
you that there
is a whole lot of things
that we are expecting
for today and tomorrow.
I am requesting all of
y'all to go experience
the technical sessions.
There is a huge campground
out there for all of you
to explore.
And don't forget
to join us tomorrow
for True to the Core.
Thank you, all
of you, again,
for being here today and
being part of this really
fantastic experience.
Thank you.
[APPLAUSE]
[MUSIC PLAYING]
Resume AutoScroll
Link copied to clipboard
TDX Bengaluru Main Keynote
Become an Agentblazer with Salesforce and discover how to bring humans and agents together to drive customer success.
Starring
- Christophe CoenraetsSVP, Technical Audience Relations, Salesforce
- Muralidhar KrishnaprasadPresident and CTO, Unified Agentforce Platform, Salesforce
- Arundhati BhattacharyaPresident & CEO, Salesforce - South Asia, Salesforce
- Parul JainVP, Software Engineering, Salesforce
- Avanthika RameshSenior Director, Product Management, Salesforce
Up Next
Questions about Salesforce+? We have answers.
What is Salesforce+?
Salesforce+ is the free streaming service for business. It’s a place to watch live events that help you grow your career and company, on your own time. Salesforce+ is produced by Salesforce, so you’ll find our company’s big events like Dreamforce here — as well as virtual learning made for your role and industry. Everything is free and unlimited.
What can I watch on Salesforce+?
You can watch all kinds of helpful, entertaining, and inspiring videos on Salesforce+. One thing our members love is our events. Whether you catch them live or on demand, Salesforce+ gives you a free front-row seat to our biggest global events, like Dreamforce, TDX, Tableau Conference, and Connections. These entertaining business shows star people in roles and industries just like yours, giving you fresh ideas for how to grow your company and do better work. The best part is, it’s all free.
Do I have to pay for Salesforce+?
Salesforce+ is free. Watch any Salesforce+ events or series at no cost. To enjoy unlimited free access to our entire library of videos, become a member by signing up for Salesforce+ with a Trailblazer account. To join us for virtual events like Dreamforce or Connections, you need to create your own Trailblazer account (it’s free!).
What is a Trailblazer account?
Your Trailblazer account gives you access to a unified profile across the Salesforce ecosystem and allows you to manage your settings and access across multiple Salesforce sites.
How do I sign up to register for live and on-demand events, like Dreamforce?
Sign up or log in to your Trailblazer account and complete a few additional details to watch Salesforce+ events live and on demand. We recommend preregistering for all events on Salesforce+ to get your own personalized journey, reminders, recommendations, and more.
Do I need to be a Salesforce customer to watch Salesforce+?
You don’t need to be a Salesforce customer to watch Salesforce+. Our virtual learning is open to everyone.
