I'm very excited for our
first guest of Dreamforce.
He's here to give us a
look at the AI innovations
we are going to hear
about this week.
It is my pleasure
to welcome
Salesforce President and
Chief Engineering Officer,
Srinivas Tallapragada
to Dreamforce today.
Srini, thank
you for coming.
Always one of my
favorite interviews.
We have a tradition
of starting
all of our presentations
with a thank you.
But I really think
we need to thank
you and the
engineering teams
because the unbelievable
technology that we're
releasing this
week is really
going to be game changing
for our customers,
Thank you for inviting me.
And I want to also
thank all the T and P
folks of who we call
technology and product,
this is all our
engineering and product
organizations,
who've been working
very long and hard,
as you can see.
And I think you all
will have great fun
So let's talk
about Agentforce.
It is the big announcement
coming out of Dreamforce
What separates Agentforce
from other autonomous
And what makes Salesforce
uniquely positioned
in this space to make
our customers successful?
Let me just, if
you step back,
every agent requires,
first of all, a role,
like what is it's trying
to do, just like a human.
Then, it needs
access to data.
And a lot of
times this data
is trapped in
the companies.
They are not in
one single place.
They are in
different places.
Then, it needs some
tools, just like a human
would do, a lot of tools.
And then, you need to know
what channels it is on.
Where is it
available, either
in WhatsApp, or text,
or web, or something.
All of it is underlined
by trust, Carolyn.
You need to trust
what it's saying.
So if you take
back and then you
see what we have been
building for the last four
years, first of
all, we had to get,
for all the roles,
we had the C360.
So we really understand
the jobs to be done.
Then, we got
the Data Cloud.
And Data Cloud with
our no copy technology
unlocks all the trap data.
And on the tools,
all the investments
you made in Apex,
flow, and also
with our MuleSoft acts,
it is able to access it.
And we have created
a new engine
called Atlas Reasoning
Engine, which
is very advanced
reasoning engine,
And then, of course, we
have the Salesforce Trust
in terms of trust layer,
in terms of preventing
So if you look at what
an agent needs to do
and what we had to
do, it's perfect.
I don't think this unique
combination of the C360,
plus data, plus the
Agentforce layer
underpinned by trust, and
available across the world
with a lot of
certifications
per industry, I
think, is very unique.
Because when you look
at it being built
on the platform,
you mentioned
that extensibility
and the reusability.
So customers can
use the things
that they've built,
the things they love,
those workflows,
the Apex, any
prompts they've
already built. That,
Agentforce is
taking advantage.
Then you've got
data, right?
And it's grounded in our
customers' data, which
is really making a
difference in how
quickly our customers
can build agents.
So it allows them not to
start investing so much.
That's what we think we
can solve it very easily.
So when you're
talking to customers,
What are they
excited about?
What are they
nervous about,
and what advice
would you give them
when they're exploring
building agents?
No, I think I've
been talking
In fact, yesterday
morning, also, or day
So I met a lot of
APAC customers.
And I always ask them,
everybody wants agents.
They understand
generative AI.
They are all very excited.
But I said,
hey, how many of
you have been trying
out, or piloting,
or trying some
experiment with agents,
or co-pilots, with
ChatGPT, with the models?
Almost everybody, all 100
people raised their hands.
When I say, OK, how many
of you have got something
in production, only two
people raised their hand.
So customers are
realizing that this do it
yourself is very
complicated, very costly.
But they know exactly
what they want to achieve.
What they are stuck
is how to get there.
And that's what I
think we will help
What I tell them is,
hey, while you are
at Dreamforce, try it on.
We have our
Agentforce boots here.
If you bring in, you
build your first agent,
you'll see how easy it
is, how it leverages
all the investments
you made,
how we have made
it very easy,
low-code way
to turn it on.
But you got the Salesforce
security and trust
I think one of the most
fascinating things I've
seen is, as you're
in building an agent,
you've got, on one
screen, the data
So that's immediate
guardrails.
You're giving
it instruction
And then, on the
right-hand side,
you're seeing
what the outputs
And so for our
developers and
for our trailblazer
community
to be able to
see that, I think
that's where the
trust comes in.
And then people start
getting more comfortable.
So there is so much
innovation possible
with Data Cloud and
with Agentforce.
When you're
looking at, how
do you decide
with your teams
what you build
for our customers?
So I think,
first of all, we
are a big user of
Agentforce ourselves.
And if you look
at Salesforce,
we get trillions
of database,
trillions of transactions
we are doing every day.
Number one is
trust, correct?
So we use a lot
of AI in what
we call AI ops, a lot
of our language models
in capacity planning
and everything built
by our own research team.
And a lot of them we
even open source, too.
That's to run
the operations.
Then, for our
engineering, our engineers
are also using our
Code Genie models.
And we have got more than
3 million lines of code
which the engineers
are generating
every day through the AI.
In fact, a
version of that,
which, for my internal
devops, which I use,
we also released for
our external customers.
It's called Agentforce
Dev Assistant.
And so if you are
a customer who
is using Apex or
LWC, you can generate
And then, the
third part is, so
that's the evolution
and how fast
Then, how do we invest in?
A lot of times we are
talking to customers.
We know exactly the
jobs to be done.
And then we are trying
to say, how can we
help customers
implement those
And we stay true to the no
code, low code, pro code.
So however a customer
is comfortable,
we've got those
options for them.
And so those customers
that may be a little bit
nervous, we have
out of the box
across the Customer
360 that they
can start to try and
then modify as opposed
And then, once you
get comfortable
with those use
cases, there
That's where
our trailblazers
will be able to
create amazing value.
So when you
think about what
this means for
productivity, real quick,
what are the biggest
things that customers are
telling you where
they're gaining the most
You mentioned our
own engineerings team
and what they're
seeing, but what are
Actually, if you
step back again,
everybody thinks
Agentforce,
the general narrative is
it's only productivity.
Actually, it's more
useful in my mind
to increase your sales
or increase your reach.
Productivity,
everybody understands.
If you look in
a call center,
you're trying to
summarize a case,
it's a very painful thing
somebody has to type in.
Ability to quickly
summarize and help
them do semantic
search [INAUDIBLE],
it's a great
user experience.
The other thing
you will find
is, today, a lot of
times all the leads--
a company, leaves
a lot of leads
on the table
because they don't
have people enough
to chase them down,
This is where our
Einstein SDR really do it
So I think there will be
advantages on both sides.
And we can get
hands on everywhere.
So Srini, President and
Chief Engineering Officer,
thank you so much
for joining us.
