Hello, everyone, and
welcome to Dreamforce.
And hello to everyone
on Salesforce Plus.
Today we're going to
go beyond the hype
and see agents in action.
We're going to look
at some forward
looking technology so I
need to ask that you make
your purchasing
decisions off of products
and services that are
readily available today.
But first, I want to thank
all of you in the room.
Thank you for
attending Dreamforce.
Thank you for
being our customer.
I truly believe that
it is your feedback
and your input that
shapes our products
and makes the company
that we have today.
So please keep that
feedback coming,
whether it's
after this session
or throughout Dreamforce.
So for those of you
that don't know me,
my name is
Krithika and I'm
a product marketer focused
on our sales cloud AI
I'm joined by two
brilliant product experts
here who are at the
forefront of our cutting
edge Agentforce
technology,
Sean Keen, who's
building Agentforce SDR,
and Chelsey Smith,
who's building
So this morning and
throughout Dreamforce,
you've heard us talk a
lot about Agentforce.
Today we're going to
talk about Agentforce
For those of you that
aren't familiar with Sales
Cloud, let me
introduce you
to what we call the
Sales Cloud fan.
At the core or the
foundation of this fan
is a lot of our core
CRM capabilities
And every blade
of this fan
represents
capabilities that we've
built into the CRM to
give you a complete growth
platform from
pipeline to paycheck.
So these are
capabilities that
help you prospect, that
help you collaborate
with your team, that help
you analyze your sales
data all the way to
billing your customer.
And AI is woven into
every blade of this fan
to make your teams
more productive.
But we're at an
inflection point
right now with
AI where we're
moving from a
productivity revolution
to a workforce revolution.
If we look at the
last 10 years,
when we started this
journey at Salesforce,
all the AI that
we've built
is about assisting
individual sellers.
It all started with
predictive intelligence
where Einstein helped you
predict deal outcomes,
helped you
understand where
your forecast might land.
All these predictions
helped our sellers
We're now entering
the world--
or sorry, before
that, actually, I
should go back and talk
about generative AI
because we took it
one step further
after productive
intelligence
And generative
AI didn't just
recommend the next
step for a seller,
but instead it helped
Sellers take actions.
And our generative
AI capabilities
help sellers automate
account research,
generate emails on demand,
and summarize records.
So this was AI that
helped you sell faster.
Now the exciting stuff is
all about autonomous AI.
That's the new era
that we're approaching.
And it's no longer
about assisting
an individual
seller but instead
about augmenting
an entire workforce
or an entire organization.
Autonomous AI agents, they
scale your sales teams
because they do
tasks for you,
like booking meetings
for your reps
or conducting role
plays with your reps
The future will be
humans with agents
working together to
unlock new levels of sales
So I get to introduce
to you today
two sales-specific
agents that we
are launching in October.
Agentforce SDR is going
to maximize pipeline
by nurturing inbound leads
24/7 and Agentforce Sales
Coach is going to allow
you to give every rep
a dedicated coach to
make them your best rep.
So before we dive
into these agents,
let's talk about the five
attributes that make up
Every agent has a
goal or an outcome
that they're
trying to achieve.
In our case, for
our sales agents,
they're trying to
nurture pipeline
And the way your agent
does this is through data.
Similar to how humans,
we use the knowledge
in our brain to achieve
our roles or our actions,
an agent uses
the data in CRM
And as they're making
these decisions,
These are the actions
or the capabilities
that an agent does,
whether that's
booking meetings,
conducting role plays,
providing feedback, and
they take these actions
within the guardrails
that you set for them.
These are the
rules in which
how you tell this
agent to operate
based on your
business practices.
And agents interact
with your employees
and your customers
wherever they are,
whether that's in the CRM,
email, SMS or elsewhere.
And every agent
interaction
goes through the
Einstein Trust
Layer for ultimate
trust and security.
So keep these five
principles in mind
as Sean walks you
through Agentforce SDR.
And I want to note
that one challenge
that sales organizations
can have is
after a successful
marketing campaign
or after successful
event, like Dreamforce
here today, that it can
be difficult to keep up
with the volume of leads.
And that's where
Agentforce can help.
Agentforce SDR partnered
with the human cellars.
The idea is
that Agentforce
can automate many of those
top of funnel activities,
help to qualify the
leads, progress the deal,
and then pass
off to the humans
This is where we get
to see the ROI of AI
and I'd like to jump
into a demo of this.
And I will note, I
want to do something
a little bit
different here,
because a lot of
times when we demo,
we like to show
the end product,
what the end user
uses, but today I'm
going to start out
with the setup.
And you'll see, as
product managers,
we normally don't
demonstrate setup
because that's not
as fun, but you're
going to learn a lot
about how Agentforce works
from how it's configured.
So you see here, I'm
setting up Agentforce,
I've got two options, my
SDR and my Sales Coach.
In this case, I'm going
to jump into my SDR.
And four steps
for activation.
The first one,
there's a variety
of different
features that need
to be on in order for
Agentforce SDR to work.
Now, in the past we
might have told you,
like, go to all these
different places
to turn those options on.
We're going to
make it simple.
We're just going to say
enable all to save you all
Now, next, I'm
going to configure
I start out by
creating a user record
that the agent is going
to be associated to.
So let me go
ahead and do that.
And the reason
that I do that
is because I get all
the benefits of there
being a user record tied
to this agent in terms
of analytics
and monitoring
the use of the agent,
as well as being able
to audit what's happening.
So you will also notice
that I have the ability
to assign a manager
to an agent.
Now the manager becomes
responsible for a couple
Manager is responsible for
training the agent, which
I'll show in
a few moments,
as well as monitoring
the usage and the output
Now, before I move
to my next step,
I'm going to have
a little fun here.
Let's select an avatar and
make it look like zebra
and that will be my agent.
Now, here's where I'm
defining some of the rules
and the guardrails
around the agent.
How it's going to operate.
My language, this agent's
can operate in English.
It's going to use
more of a casual tone.
And then I set my rules
of engagement to say,
this is when the
agent reaches out.
This is how many
times they'll
And this, again,
goes to the idea
Krithika was talking
about, guardrails.
This helps define
what the agent will
have a conversation
about versus,
if it's not on this list,
that they'll punt over
And then finally,
lead assignment
where I'm defining the
parameters around what
leads will be covered
by this agent.
So I'm going to go ahead
and use these defaults.
And then finally,
I've got to give
the agent some knowledge.
And Krithika talked about
the five different columns
and one of them
was knowledge.
And you had made
reference to the knowledge
from the CRM, which makes
these agents powerful,
but when you hire
a new seller,
you're not saying,
hey, learn everything
you need to know
inside of the CRM.
You're bringing in
external sources, as well.
You're bringing in
assets, materials, slide
decks, et cetera,
so you can
do the exact same thing
with Agentforce SDR.
Where in this case,
just as an example,
I'm going to say,
give the agent
the pricing PDF as well
as some case studies.
And this could be
as much or as little
as you want to,
but remember,
the more knowledge you
arm the agent with,
the better it's going
to be at handling
Moving on, it's
ready to roll.
So I'm going to
switch gears here
and I want to jump
into the runtime.
What actually happens
when it's engaging
And on this
screen you'll see,
left hand side
of the screen
you're kind of seeing
the workflow in terms
of where the agent is at.
That's kind of
our wayfinding.
And then the right
side of the screen
you're seeing what
the prospect will see.
And we start
out, the agent
knows how to reach
out to this prospect.
Hey, you downloaded
this e-book.
Check out this demo
video and would you
be interested
in setting up
a meeting with Heather,
who's our human sales rep?
And you'll notice
I've got the signature
at the bottom, "Sent
by Heather Cartino's AI
That is super important
because our goal here
isn't for these agents
to personalize humans.
The goal is for
these agents
to augment the work of the
humans, work with them,
and in order to
maintain the trust
of the prospects,
we want to be clear.
Hey, you're working
with an agent here.
What ends up happening
is the agent knows,
hey, it's time to
follow up with Simon.
It's been, within my
rules, three days.
Would you be interested in
connecting with Heather?
How much does your
solution cost?
Now, we set guardrails
for this agent to say,
hey, hey, don't give
out a pricing guide.
When pricing comes up,
talk about packaging.
So the agent comes back
and says, hey, here
are the different
editions of the product.
Do you want to talk
to Heather in order
to have a deeper
dive conversation?
You might see a theme
here in terms of what job
the agent is trying to do.
Next, we get an objection.
Hey, do you have
a free trial?
Well, the answer
is no, but there's
a way to handle this
objection, which
is to say, hey, we
don't have a trial,
but there are other
ways to go hands on.
You know who you can tell
you about that is Heather.
Would you be interested
in setting up
And finally, Simon
bites and says,
yes, I would like to set
up a meeting with Heather.
And the agent can
provide some of Heather's
And when Simon
selects a date
and a time, the agent
books the meeting,
sends a meeting invite to
Simon and to Heather, gets
it on their calendars,
and, doesn't end here,
behind the scenes, Heather
gets a notification that
says, hey, a new meeting
has been scheduled.
And the agent does
a clean handoff
So what the agent is
doing is not saying,
hey, look at this
email thread.
Figure out what
this is all about.
The agent is giving
a quick summary
of what was talked
about, a proposed agenda,
as well as highlights in
terms of any objections
or challenges
that may come up
in the conversation
so Heather can briefly
read that and jump
in that conversation
prepared to move
the deal forward.
Now, we talked a
few moments ago
about the
different channels
that you're going to
see agents work in.
I just want to show
that it works inside
of SMS, as well, and
it works the same way.
I've got my initial
outreach in SMS.
I've got a question
that comes up.
The agent can
answer the question
and nudge towards setting
up a meeting with Heather.
And then maybe
the prospect
comes back and says, hey,
don't message me anymore,
which isn't the
outcome we want,
but it happens
from time to time.
And I point this
one out because I
want to note, remember,
this isn't a bot.
This is an
autonomous agent.
We've all seen bots where
it's like, text "stop"
to stop receiving
messages.
The agent understands
natural language,
understands that
this person doesn't
want to receive
any more messages,
It says, great, if you
ever want to engage again,
hit us up on
this SMS channel.
And then finally, I could
switch gears and talk
And again, this is
illustrating the idea
that agents can work
across different channels.
And one advantage
an agent has is
it doesn't have to work
one lead at a time,
an agent can
multitask, and an agent
can work across multiple
channels simultaneously.
You will also notice
that this is in Spanish.
And I want to note
that we didn't say,
OK, we've trained
this agent in English,
now let's train it to
have Spanish language
You train your agent
one time and the agent
is multilingual, so
it can be deployed
across different languages
without you needing
to say, OK, let me get my
French language materials
So again, it has
some flexibility
that human
sellers might not
have that can, again,
augment the efforts
Now, again, to
recap, what we
saw here was the setup
of Agentforce, defining
the parameters of how
the agent will work,
and we saw
multiple channels.
We saw how the agent
worked inside of email,
inside of SMS, and
inside of Whatsapp.
Now let's talk
about agents
actually interacting
with your sales team.
And for that, I'd like to
pass it over to Chelsey.
I am Chelsey Smith, she
and hers, a senior product
manager on Sales Cloud,
and excited to be talking
with you today about
the world's first Sales
Now at Salesforce, we
believe every seller
With Agentforce
sales coach,
they're going to augment
the coaching workload
of sales managers,
supporting every rep
on every active deal,
providing personalized
guidance to build that
seller confidence.
And Coach does this again
and again, consistently,
in a non-judgmental
way, coaching at scale.
Now let's go ahead
and take a look
Now, Sean showed
us how you
can get started
with your agents
So our admin has
rolled out Sales
I'm a seller working
on a deal for Pacifica
and right here,
prominently I
can see my Coach on
my opportunity record.
Now this Coach
is always on,
always available, and
perhaps most importantly,
always up to
speed on what's
going on with my deals
thanks to the CRM data.
Coach sees that this deal
is in the discovery stage
and gives me
an opportunity
to practice my pitch
and get feedback.
Coach sets me
up for success
giving me some
best practices when
it comes to doing a pitch
at this stage in the deal
I get my notes together
and get started.
Now this is super simple.
I'm right here using my
microphone, my camera,
no third-party
tools required,
and I'm going to
record my pitch.
Now, I know
during discovery
that it's key to
articulate the customer's
problems and how
my product is going
to solve them
and how I'm going
to do that better
than my competitor.
Now, at the end
of my pitch,
I'm able to rewatch
what I've captured,
do another take, or
go ahead and provide
it to Coach for feedback.
Coach is taking the
pitch I just completed,
the richness of my
opportunity data in CRM,
a prompt that we have
provided out of the box,
and with the
help of LLM is
generating this
feedback just for me.
Coach provides me
a deal, summary,
feedback on my pitch,
areas that need attention,
Now, Needs Attention
is interesting.
Coach is actually looking
at my activity timeline
on my opportunity
and letting
me know, hey,
you're behind
Let's get back
up to speed.
And Next Steps
are intended
Knowing where I'm at
in this deal cycle
and what the
next stage is.
Coach is looking
closely at what
I've identified
as the next step
on my opportunity
record and evaluates
Is it appropriate at
this stage in the deal?
Well, I've got some really
great next steps here
Now fast forward and I'm
in the negotiation stage
and have a call coming
up with my customer.
Coach has been there with
me every step of the way
on this deal and knows
that the best way for me
to prepare for
negotiation is to practice
So I'm going to do a
role play with my Coach.
Now, again,
Coach sets me up
for success with some
tips and best practices
on how to do an
effective negotiation.
Now, in this
interaction, Coach
is playing the
part of my buyer.
They're going to
ask me questions
and make objections just
like my buyer might.
They're going to ask me,
can I offer a discount?
Can I improve any deals
of the term-- excuse me,
can I improve any terms
on this deal, right?
And all of this
is happening
I am speaking
into my machine
and my Coach is
talking back to me.
That was a
great role play.
Now, again, Coach is
going to give me feedback
on how that session went.
So it's taking the role
play I just conducted,
the latest CRM
data about my deal,
prompt shipped
out of the box,
and, again, with a
Large Language Model,
generating this feedback.
Now you can see I have a
couple of new sections.
Overall Impression of how
the conversation went,
specifically What
Went Well, some Growth
Opportunities, and the
all important Next Steps.
Well, my Coach is
a tough negotiator,
but I'm feeling confident
going into my customer
As you can see, Sales
Coach has joined me
and is actively listening
to my conversation
Over on the right, Coach
is on the sidelines giving
me real-time
insights in how
I can respond to things
that are coming up
How could I handle that
competitor mention or tips
on the negotiation that we
just practiced in our role
Well, I'm feeling
really great
coming out of that
call with my customer.
I feel confident
I'm going to be
Now, you've just seen
how Coach has been there
with me every
step of the way,
providing me
hyper contextual,
personalized
feedback to help
me build momentum and
move this deal forward.
So you've just seen two
amazing Sales Agents
Our agents are
effective because they
are supported by the
world's number one sales
And with our
Atlas Engine that
handles the complex
reasoning of these agents,
we know with confidence
our agent behavior is
We empower customers
with no code or low code
experiences to
configure their agents
and achieve deep
customization.
And with monitoring
and auditing
of agent activities,
customers
have a crystal clear
picture of the work
that the agents are
doing on their behalf.
And all of this
is integrated
with existing
systems of action
that your sales teams
live and breathe in.
Well, on behalf
of Krithika, Sean,
and myself, I
want to thank you
for joining us
today and seeing
And quickly-- quickly
to close us out,
right across the hall
we are giving out
Zig is our mascot
on Sales Cloud.
So if you scan
this QR code,
there are two
ways to win a Zig.
One is to become
a Sales Blazer,
another is to leave
us a review on G2,
and then across the
hall in the Sales Lodge
So I'll let everyone
do that QR code.
And then the last plug
that's on the next slide,
but I know everyone's
still doing the QR code,
please attend the Sales
Cloud keynote tomorrow
at 1:00 PM to hear all of
our latest innovations.
And then tomorrow
at 3:15 PM,
there's a great
session about hearing
our Salesforce plans on
how we are going to use
It's our lead of IT, our
lead of sales strategy,
and they're going to
be talking about how
we are going to use
Agentforce internally