Play
Show Navigation
Show Transcript
00:00
Play
Seek 10 seconds backwards
Seek 10 seconds forward
00:00 / 01:08:24
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
Welcome to Release Readiness Live.
It is important to watch Release
Readiness live
so that you get a snippet
of the coming attractions
of the functionality
that's coming from Salesforce.
We really focus on the features
that are going to be most useful
and impactful with release readiness.
Why have you got the great opportunity
to actually hear
from the mouths of the people
who spent a lot of time
thinking about what exactly
you all,
as Salesforce developers
and admins, are actually wanting to see.
Your feedback is so important.
I think that's
one of the most amazing things
actually about the Salesforce
ecosystem is how involved
trailblazers are with evolution
of our products.
Is RL about to start?
Let's get started.
To.
Welcome to Release Readiness Live.
I'm Carolyn Bathauer, our senior customer
and product evangelist.
And your new host of RL.
To all of our developers out there,
this is your day.
It is developer preview time.
We have an incredible
team of experts here to bring you
the exciting new innovations
coming your way in the Winter
25 release,
and you've probably heard of Agent Force
by now.
And we're excited to walk you
through Agent Force
Reasoning Engine,
also known as the brain of an agent.
You'll learn how topics, instructions,
and actions from the agents
you're building
to make your teams
more productive across the customer.
360 and I do need to remind you
to only make purchasing decisions
on existing features and functionality.
So with that covered,
I want to thank you
for investing your time
with all of us today
and for the feedback that you provide.
It does help our technology teams
build solutions
centered around your experience
and to make this broadcast even better.
And we have a great lineup
of release readiness content for you,
starting with our Developer Preview.
We'll be back here at 11 a.m.
Pacific for Service Cloud.
So you don't need to go very far.
And that's followed by a special bonus
episode featuring Health
Cloud at 130 Pacific.
And day three wraps up tomorrow.
We have CRM, analytics,
and Admin preview.
All of our episodes
will also be available on demand.
So over the next hour
you'll learn how to answer
intelligent queries using Prop Builder.
Plus,
we explore how to manage
performance metrics
to increase your productivity,
and to get the latest on the top
developer features
and strategies in Winter 25.
We've got that coming your way
and along the way
we want to hear from you.
So we want to answer your questions.
You've got the experts here.
So make sure
you're joining the chat on Salesforce.
Plus drop your questions in the Q&A tab.
We've got this team of product experts
standing by to give you those answers.
And then we'll also take
some of your questions live.
And if you still have questions
after the show,
you can always post them
in the trailblazer community.
Hashtag Salesforce rel.
So here we go.
First up
we have Angela
Lee, probably the most popular person
at Salesforce
today, director of Product Management
for Agent Force, Angela,
I can't wait to hear
how Agent Force works
to deliver value
to all of our customers autonomously.
So take it away.
All right. Thank you. Carolyn.
Hi, developers.
My name is Angela
and I'm one of the product managers
on the Agent Force team.
And I'm so excited to tell you
about some of our great updates.
And then really peel back
behind the curtain
and tell you how this
Agent Forest reasoning engine works.
So let's get started.
All right, so
I'm sure you've
heard a lot about Agent Forest
and you're wondering,
well, how does it actually work?
And that's crucial
for you as a developer,
because we want to make sure
that when you're building this,
you know how it's going
to work in your org
and serve your users and customers
very well.
So Agent Force is three building blocks.
The first is topics.
And you can think about topics
as the jobs
that you want the agent to do
or the topics of conversation
that the agent should be be aware of.
And then we have instructions.
And instructions are crucial
because they're actually going
to be the guardrails for your agent.
You want to make sure
that it only handles
the things you want it to do,
and then be able to give it
those, clear instructions
that can escalate to a human.
If you need someone to intervene
and you don't want it to go rogue.
So instructions are really,
really critical for that.
And the last part of, of your agent
is actions.
And actions are really powerful
because they allow that agent
to take action on your behalf.
And this is also where you can go
and customize
some of the things that you've
already built inside
of the Salesforce platform
and repurpose them
inside of your agent,
and you'll see how
all of these three building blocks
come together inside of, an agent
in the demo.
And I know
I have gotten
tons of questions of
what is the Atlas reasoning engine?
And very simply,
it's the brain of Agent Force.
It's the orchestration
engine that communicates with your users
and the elements to generate a plan and
to execute those actions
that you've assigned to your agent.
And it makes sure that you have
the relevant business actions
for any task
that you want your agent to conduct.
All right.
So that's that's,
you know, it's really hard to grasp
how all these things work together.
So let's just go ahead into the demo
and then we can see how the
all of these things come together.
So here I am in Coral Cloud Resort.
So this is a great place for me to start
thinking about some amazing things
that I can do on my next vacation.
And here we already have
one of the agents that we've built,
that connect to this experience
cloud site.
So I'm not going to type everything out.
I have some special keyboard shortcuts.
So just like any, customer
service experience, this is this agent,
is able to handle this query.
I'm going to ask
if I'm Sophia, and I give him my email,
he'll let me know
if there's any interesting
experiences that are happening today.
And what's happening behind the scenes
is that this agent is taking my query
just a natural language,
and processing it
through that reasoning engine
and then providing a response.
And so you can see here
give me a list of activities
that match my interest,
because it also has on my data
inside of the CRM.
So knows the things that I'm interested
in like get yoga or bootcamp
and things like that.
And now it's asking me if I
would like to book
any of these activities,
and I'm going to go ahead
and go into the builder to show you how
we can actually teach this agent
and have this agent
take action for us
by booking an experience for our user.
So here I am in set up,
and I'm going to go ahead
and open this in builder and you can see
I already actually had
two agents in there.
Copilot
the agent
that was formerly known as copilot
and then this coral cloud agent.
So here we are in Agent Builder.
And this is where
all of that customization magic happens.
So you can have
we have an experience management
topic as you saw.
But you can also add any other topic
that is relevant
for your particular use case.
You can have one that answers
generally frequent ask
questions
summarization things like that.
So let's click in here into this topic.
I'm going to expand
a little bit as well.
So you can see.
So as I said before each topic
has a scope descriptions
and then instructions.
So let's go through each of these.
First we have the scope.
So
I'm going to zoom in a little bit.
We have the scope where
we can give the agent a job.
And in this case
the job is to help navigate and book
any experiences
that a customer might ask about.
And then we have instructions.
So, you know,
when we want to make sure
that we're being really prescriptive
for the agent
for how they interact
with these customers.
So I want to make sure that we're
passing in the right records.
If they request to book
a specific session,
that we have to go
and reference a specific,
record, here
assignment it just
locate that session record.
So you can be really prescriptive
in your instructions so that the agent
is acting the exact way
that you would want them to, to be.
All right. So the next part
is actions.
And actions are super powerful
because they actually take a lot
of the things
that you've already built
into the platform in flow's
apex or prompt templates,
and then wrap them and
create them as actions
that your agent can use
to take action on your behalf.
So as you can see here,
we have a few actions
that are out of the box
that we've already built for you that's,
available inside of the platform.
But we also have two
custom actions here.
And then
let's go ahead and take a look
at how we actually built
these custom actions.
And as you can see,
one of them is a flow
and one a popped profitable.
So let's take a look at the flow.
And you might be
wondering,
well, one should be able to reason
through all of these things.
Why would I use flows
which is more deterministic?
Well,
part of the answer is, well,
because it is more deterministic
when you're creating
and booking experiences,
you want to make sure that it's super
prescriptive,
that we're going to grab that right.
Booking ID,
that we're going to create
the right type of record.
You don't want to leave it up to chance,
to an element. Right.
So here we are in flows.
And I'm sure many of you developers
already have tons of flows
inside of your, your org.
So you can repurpose those,
as actions inside of Agent Force.
So here we are in this flow.
We're going to make sure that,
we're going to create that
if someone asked to create,
a book
to ask to book an experience,
we're going to create that booking,
we're going to get those booking records
and then give them,
an output that says,
oh, it's been successfully book
and it has that new booking ID
so just that set flow,
we can wrap it as an action.
And the agent can use this,
to take action if
a user asks it to do so.
Next we have that other second
custom action
which is
built with one of our prompt templates.
And as you can see here,
in prompt templates,
we're using this prompt template.
We've really
simply asked it to provide us
a list of personalized experiences.
And we have an apex class
that we can call
and bring into
the prompt template itself.
And there's no code here.
We're really simply giving it
some instructions,
that will allow it to take action.
So here we're going to ask you
to always use a bulleted list.
You can tell that request
to actually prompt the guests
to reach out and book an experience
if something looks to their interest.
And then giving it, making sure
that we include the name,
time of activity, duration,
things like that,
but don't include the date
because it can always change.
And then the other powerful thing,
within our prop templates
is that you can also choose the model.
So here we have three,
but you can also use GPT for or Azure.
And then we're always continuing
to add more models
that you can use and reference.
All right. So
we added the actions we created them.
And now let's
actually see how it works.
And in our preview.
And then we can also get a
a sneak peek
behind how our reasoning engine works.
So here,
you know, try that request again.
Around.
Hi my name is Sophia.
I actually provided
an email just in case.
Can you let me know
if there's any interesting experiences
that are happening today?
And what's happening is,
we're putting in this request.
It enforces actually
going back
and reasoning through and understanding
what topic should I choose for this?
But in after that,
what action should I take?
So as you can see here, and
make sure I want to make really clear
that this is a reasoning.
It's not a dialog tree.
It's really going through
and semantically understanding
what the request was.
And then first
we selected the right topic,
which was experience management.
Then it picks the right action.
So first we had to identify who
Sophia is.
Does she have a record does she not.
And we're actually able
to grab that correct record.
And then I asked for a schedule.
And there we have
it selected that right action,
which was a custom action
that we created
within that prompt template.
And then it gave her response.
So it answered correctly.
We had that list.
It has a time
and how
long each of these experiences are.
All right. So
actually
oops picked the wrong shortcut key.
I actually want to know more about
that beach bucket challenge.
And you know,
let's say that
I looked at something in the list.
I saw that it was really,
something of interest.
And this is
where has that conversation went back
and forth that you can have
with an agent?
It's not a dialog tree
that I already had.
It's to me
it's just going to respond
in a way that you expect
Ella wants to do do so.
So, again, it
grabbed the record details
for this particular
record about the beach boot camp,
gave me some information about,
you know, the activity level,
where its location is
and things like that.
And then it also asked
if you would like to book it
because we had included that
into the prompt template.
And so actually, yes,
that sounds super fun.
So now I'm going to go ahead
and ask that agent
to book that experience for me.
And what I'm looking for is, again,
hit the right topic.
But now it grabbed that action
that we created inside of flows.
So now we created that new booking idea,
which is right here in that session.
The output is now
this new booking information.
And now that output message
that the booking has been successful.
So now
we can have our,
our customers be able
to not just ask information,
but then actually ask the agent
to do something for them
and be able to fully complete a task.
All right. Back to you, Carolyn.
That was great. And you know what?
You've been so busy.
Amazing technology,
for everyone at home.
If you have questions for Angela,
you can post them in the Q&A,
right on Salesforce Plus.
But let's get to a few.
We do have some coming in.
Angela,
can we have an action
tied to multiple topics?
Yes, absolutely.
And that's actually
one of the most powerful
things about actions
is that you build it,
but you can use it anytime, anywhere.
So true to say oh for you.
Yeah. The other one is everywhere.
That's awesome.
Staying on topics,
how many topics can an agent have?
Yeah. Great question.
So right now,
we are looking at around 20,
but as we continue
to build out features
and things like that,
we're probably
going to be able to add more
and just make sure that we are having
the most important things about topics
is that they are going to
you don't have
want to have overlapping topics.
So it should be really distinct,
like one job to be done per topic.
Once you start mixing them,
then you might have the,
the agent in the planner
might pick the wrong one.
So you want to be really explicit
and make sure they don't overlap.
So granular is really important
in this case.
Okay. Great.
One more question.
Can actions be reused across agents.
Yes.
That's great
because it's just part of that
agent force platform. It's a platform.
Object. Yeah.
So those components,
you build them similar to like you
said, you can bring in flows,
existing prompts. You build apex class.
That's it's pretty awesome.
So thank you for that.
And you'll be back in
just a little bit to answer
some questions across the panel.
Keep the questions coming on chat.
So we've got more
AI innovations to tell you about.
Cyn Shen is here.
Associate product Manager
for Prompt Builder
said,
you're going to show us
a lot of enhancements to Prompt Builder.
That'll make it easier
for developers to build
and deploy all of their prompts,
so can't wait to see it.
Awesome. Carolyn, thank you so much.
Hey everyone.
My name is Shanda and I'm Associate
Product Manager here at Salesforce.
Now you just saw
some of the amazing innovation
with Agent Force
will critical to building eight great
AI agents are actually prompt
and prompt builder.
What this allows you to do
is customize
these agents to adapt to
your business needs in your workflow.
So today I'm
going to talk about some innovations
that really focus on the data, actions
and workflows in a trusted manner.
So that being said, let's
go ahead and get started.
To start off with Einstein search,
Einstein search is our newest innovation
and prompt builder
to really take a natural language query,
search through your structured
and unstructured data,
and then surface those results
within your prompt templates.
This allows you
to use the full power
of retrieval augmented
generation to search through things
such as knowledge
articles,
cases, whatever
external or internal data sources
that you have
that you can also connect
through data cloud.
Now before when it came to template
triggered flows,
automatically inputs were created.
The tide ID in a 1 to 1 relationship
with the prompt template itself.
This meant that you couldn't
use reuse Apex classes or flows in
many different prompt templates
will now with Flow and Apex reuse.
I'm pleased to announce that
as long as your inputs
into the template triggered flow
are a subset of the inputs
into your prompt template,
then you can actually reuse that
in as many different prompt templates
as you'd like,
which really adds
to saving time and effort for our users.
Now, trust is our number one
value here at Salesforce,
and we're adding onto existing
Einstein trust layer
functionality of machine
learning models
that detect and mask PII
to now also include field based masking.
What this allows you to do
is deterministically select
subject fields
and have that data
classification metadata
like compliance categories
or data sensitivity levels
to make sure that none of your data.
And I know that fields
really get through the cracks
and go to the limit.
And this even works
for custom fields and custom
objects as well.
Now, previously when it came to editing
standard templates,
you really couldn't do that.
What would have to happen
is you'd go into prompt template,
open a standard template.
You'd have to save
that as a new template,
and then we connect that downstream
with whatever flows.
AP classes
that were invoking that prompt template
will now with standard override
templates for certain select templates.
What you can actually do
is save that as a new version
in Prompt Builder,
and that gets automatically reference
whatever is invoking it from before,
which saves you a lot of time as well.
Now last but not least
is prompt performance metrics.
What this does is
it allows you to aggregate
your end user feedback
data, store that in data cloud,
and then surface
all of those key metrics
in prompt Builder itself.
This allows you to go on your iterative
prompt engineering journey
to really understand
what are your end users like
and what do they not like,
and then improve upon your prompts
for the best experience possible.
With that being said,
let's go ahead and get into the demo
and see how this all works real time.
So as you can see on my screen
here, I'm going to be playing the role
of a Salesforce developer
at a shoe company. Get fly shoe.
And what I want to do is take an input
query from my customer
and then scan through my shoe custom
object, which is in the back end,
or catalog of a bunch of different shoes
with structured data and unstructured
data such as descriptions.
Scan through all that
information, retrieve
previously ordered products as a flow,
and then package
that all with the prompt
to recommend a specific shoe
for my customer.
So I'm going to walk you
through how this all works in data cloud
and then bringing that
into Prompt Builder as well.
So let's go ahead and walk through
how this works.
So as you can see,
the first thing you do
is bring in your data
through a data stream.
I've already brought that in
with our Salesforce CRM data connector
which is a custom object shoes.
And so I'm going to walk you through
what we do next.
So this automatically creates
a data lake object.
Now the data lake object is key
because this is what you build
your data model object on.
So I'm going to go ahead and click that.
And you can see
it has all the different fields
for my shoes.
Whether it be price material purpose
you have it.
The next thing I do is actually create
custom field mapping.
What this does is
you select the fields that you want
in your data model object.
So my data model object gets created.
And then you can go ahead
and open that data model object
that we want to have.
And it has all of the different fields
that I've selected.
All reference in this data model object.
Now comes the fun part
building our search index.
So I've already gone ahead
and configured this
because it takes about
10 to 15 minutes to run.
But I'm going to walk
you through the end
to end flow
of how you can configure this in Data
Cloud for your use case.
So we go ahead and hit new.
We'd go into Advanced Setup
to show you the full configuration.
And then this opens up
the search index builder.
So what now
I'm going to do
is create a search index on that shoe
data model
object that I just made I'd select shoe
and you can see that over here
I hit next.
And the next thing
is our chunking strategy.
So the descriptions of our shoes
are unstructured your data.
And it's a long text data.
But we want to break that down
into chunks for a search
index to better search on.
So with that being said, I'm
going to manage the fields that I want,
and I want the description
because that is a key unstructured
data field here.
We bring that in
and we choose our chunking strategy.
Well, it's not a call
log, it's not a text box.
It's not really a conversation,
but it is a passage.
That's our passage extraction.
Our shoe object is not an HTML,
so we don't need to strip any HTML.
We'll set that to false.
And then we can actually prepend fields
to junk
to junk
to actually allow you
to really search through that better.
So we go ahead and hit select the field.
We can take the shoe name.
We can add the shoe as well.
And that does it for
our chunking strategy.
The next step is
the vectorization strategy.
This is the process of
turning that textual data
into a numerical representation
for a search index to search on.
So we already have an embedding
model here. Go ahead next.
And the next thing I want to show
you is actually the related fields
for search filtering.
So these are structured data
that you can actually add on to your
to your search fields
to really allow you to enhance
that search process.
So for a shoe, you know,
I could take the color,
I could take the brand,
I could take,
I mean, even the price material,
you have it,
you could build it to
your specific needs
and your customization.
So that being said, we go at it next
and there you have it.
Voila your search index is configured.
You go ahead.
Save gives about ten 15 minutes to run.
And with that being said
your search index is configured.
Now the key is how to showcase this
as a retriever and prompt builder
to pull all that information in.
As you can see on the screen,
I have a prompt to take that input
and then work through the retriever.
So how do I actually find the retriever
that's in Resource picker?
I go into Einstein search.
I pick the data model object
which I have, which is the shoe,
and then I have a default retriever
that pulls up in the shoe
retriever that I customized.
So I go ahead
and take the shoe retriever.
And the next step
is actually configuring
this for Prop Builder.
So you can see the data model object
that acts on an input
query from the user.
And then the chunk
which is the description of the sure
unstructured data
and some other information.
And I only want three results
that are ordered.
So that being said, let's
go ahead and give this a spin
with the test input.
So I want shoes
that are functional, sturdy
and good for work on the factory floor.
So this is an example
use case from one of our customers
that they want to scan through
all the shoes.
Let's go ahead and find our contact
that we want to search on
which is Lauren.
And let's
go ahead and save and preview this.
We're going to give it a second to load.
But what you're going to see
is in the resolution panel
we get a Json view
of the different chunks.
And our structured data.
And we also get similarity scores
based on our search
index to really compare
what is the most similar
to our input query
and what is not as similar.
And then those get ranked
packaged together
with all of the grounding
and enriching data sent over to the LLN.
And then you get a packaged
response with neatly organized
that this is the shoe that you want.
So you can see that here.
Actually,
you can see our unstructured data,
which is the chunk,
the shoe description.
And then you can actually see
the vector search score
and the score,
which is what the LMS
use to rank these options.
So I'm going to show you
how this looks for the end case.
And it recommends the Timberland
six inch boot.
So this is really good
for someone who wants sturdy shoes
as according to our input query.
So that's the power of Einstein
Search and prompt builder.
Now the next thing I want to talk about
is flow reuse.
So as you can see on my screen,
I have a flow to retrieve
the ordered products.
But I'm going to click into this flow
and show you how this can be used
in many different prompt
templates as well.
So as you can see on my screen
here, here we have the flow.
The flow normally had automatic inputs,
but now we've added the ability
to have manual inputs as well.
So this flow takes in a contact
as the input.
We can go ahead and open up the contact.
And what you see here is the available
for input click button.
And that's the availability
outside of the flow.
Now this is really key
because once you select this
any prompt template
that has contact as an input,
it could have as
many inputs as you want.
But if contact is one of those inputs,
then that actually leads to you
being able to reuse
that wherever you want.
So let's
go ahead and take a look at this
and see how this can be used in
a totally different prompt template.
So as you can see in this
prompt template here,
this is a summarize
info prompt template.
But as you can see
in the resource picker
I go ahead and
flow and retrieve ordered
products is right
there and ready to use.
Now this is really cool
because contacts
and accounts are both inputs,
but since contact is
one of those inputs,
you can reuse that flow.
Just have it ready to show.
Another thing I want to call out
is record snapshot.
Now record
snapshot is a single click view
to really get
a Json view
of all the related merge fields
for a specific record,
and then bring that
into your prompt template.
With one click,
you go into the resource picker.
You pick the contact
or whatever record you want,
and then just have record Snapshot
right there with one click.
So I'm going to go ahead
and show you how this works
and show a lot of that data.
And I've also brought
in the account number
for our field based
masking and sensitive data masking,
which I'm going to show you all
when this resolves.
So let's go ahead and take a contact.
Let's go ahead and take Lauren.
And let's go ahead and take the account,
which could be Acme Partners.
Let's go ahead and save preview.
And you're going to see
in the resolution
panel is a resolved version
of that record snapshot,
along with some of our data
and our flows
and I'm going to walk you through
that Json view
and how it's pulled together,
all that information
and important merge fields
that are configured
for the running users layout.
So let's go ahead and give it a second.
As you can see on the screen
there's the Json view
of all that information.
So you have Lauren's business
phone her email
you have it
her owner that she's an owner.
She reports to who yada yada yada.
Okay.
Let's go ahead
and also show you the data masking.
So what I went ahead
and did was included an account number,
which I configured in Object Manager
to actually be config essential.
So this means it's any object field
that's always
confidential will always be masked.
So you can view our data
masking details here.
And so it caught the machine
learning model
email address and phone number
because it knew that
those are pattern based.
But the account number was field based.
So that got masked over here.
Moving on.
We're going to
talk about a few new features.
But I one thing I want to call out here
is actually the data sensitive level.
So this is an Einstein trust layer.
And you can pull the field based masking
and set confidential to always
be masked.
So moving on.
One thing I want to talk about
the standard override templates.
Normally you can see these
standard templates on the screen
and they're over there.
But with override templates
you see that in the category over here.
Let's go ahead and click on a test
standard templates override.
So instead of saving
this as a new template
you can actually just save it
as a new version.
And as I mentioned before,
this allows you to reference it
without any of the additional work.
So the last
but not least, as you can see on
the screen,
is prompt performance metrics.
You can see the generation count,
the thumbs up feedback,
the thumbs down feedback
all within the list view
for prompt builder
to really allow you to iteratively
improve on your prompts
and improve them
from that information
from your end user.
Stored in data, Cloud and Surface
and Prompt Builder.
Now you just saw
some of the amazing innovation
with Prompt Builder,
and we're so excited
to see how you use them
in your business workflows
and to really add value
to your use cases.
Thank you so much. And back to you.
Okay, that was great.
I think the technology is amazing,
but you're also speaking
to the sneakerhead in me
with your example.
So I got really excited.
And also we've got
some reactions online.
Avantika says,
wow, it's great to see it all live.
Search is so powerful.
So that's awesome.
So we do have questions right now
for Sade,
but if you have more questions,
please post them in the Q&A
tab on Salesforce
Plus because we will get to a panel
as well.
So Sid, how does Prop Builder connect
with Agent Force?
Yeah, it's a question
we've been getting a lot.
And as Angelo was talking about,
you have topics, instructions
and actions with an agent force.
I think what's really key here
is that your actions
are actually just a wrapper
on your existing Salesforce processes,
like Angela had mentioned.
And so with prompt templates,
what you can do is really add
to the customization of this.
So you can go ahead
and customize these actions.
You can change the model,
bring in extra data sources.
You can change the language
and the wording that you want it
to respond with,
and really just add to that key
customization to really fit it to
your business needs with the action.
That's awesome. Right?
And you know, we
we will get to a bit
more data
cloud to later in this episode,
but what is Data Cloud useful for
and how are you enhancing the roadmap
with Data Cloud for Prompt Builder?
Yeah.
So I think data
cloud is really important
because the big thing I would say
is that it really enriches your prompts.
So if you think about it, data
cloud is really critical
to a few different things.
So you can do with data.
Cloud
is you bring in a bunch
of different information
so you can have,
you know,
in the future
we're going to have on a roadmap
data model objects, data graphs
within data Cloud that you can bring in.
But right now
a lot of the features
that I just showed you Einstein
trust layer, field base masking,
data base masking,
all of that is through data cloud
Einstein search your similarity search.
That's data cloud performance metrics.
That's also data cloud.
So you just have
a lot of different ways
to really enrich your prompt
and make it more,
you know, grounded on your business
needs and your business data. Yeah.
And I think too, it's
you want to be prepared.
I just just your example of the way
the customer was searching,
we're not used to that as consumers,
because most of the time what we do
is we search for one
very specific thing.
I have to go
find the name of being able to say,
this is my use case
for a particular thing.
You recommend to me
what it is I should purchase.
That's pretty powerful,
and you need a lot of
different data sources
in order to make that a reality.
So because data stored everywhere
so you guys can thank you
and you will be back in just a bit,
and we will keep rolling.
Next up
we have souped up magma
product Management director,
and she is here to show you
the latest
in lightning web components
and mobile enhancements,
including the number one
developer request.
And I'm excited to see that.
So let's see what you've got for
the audience. You've got over to you.
Thank you.
I'm sort of the, director of product
experience services.
And it's a pleasure to talk to you
about DC
and mobile enhancements
that available now in the winter 25.
So I'll start with the number of
on the developer request.
And what's this is LW c local dev.
So this is
we've heard your feedback and
we have delivered a new and improved
local dev experience
that is in the beta version now,
and this is a replacement for,
previous, iteration.
And it comes with, unified
development experience
and previews of how your component looks
and behaves in production.
So it local previews are great
for iterative development.
It makes it a lot faster.
So the magic is
you don't have to deploy your changes
to org in order
to preview all your changes there.
You don't have to,
reload your browser
or clear your cache or killed app.
It just works.
So let's see it in action.
I have a demo for you here.
So first thing you want to do is.
Head over to the setup of your org
and search for a local dev preference.
And make sure
to turn on that switch there.
I've already done that.
And you can see
there is this, yellow statusbar
at the top showing
that your org is ready
for a local dev experience.
Now let me head to, vs code
and make sure that you have,
already deployed your, initial,
application, source code to your org.
And then, another thing
you want to make sure is, that, sorry,
you have installed, a lightning
dev plug in.
So you're going to use this,
sorry about that.
You're going to use this command
from CLI as if plugins install
and, install that plug in.
And then now you're ready to connect
your VS code to your local dev org.
So I'm going to, do two
versions of that.
One is for desktop app.
So it will show my lightning
app in the browser.
And another one is for, mobile app.
And it will show, the preview of my,
mobile app in the, simulator iPhone.
Simulator.
So let me run this, preview command it.
And it will
bring up my app in the browser.
So I'm using this recipes app
that is available
in now, public repository.
And it has a wealth of sample data
for all kinds of new features.
So I'm gonna go to a styling tab here
within this app.
And you see are two components.
And I will make
changes to this to component and
do instant preview.
One is
I will change this orange
background of this text
into different color.
And then second is I will change these,
icon and style and hoax component.
Right now it's a gear icon.
Okay.
Let me head back to this code and run
the preview command for,
simulator,
for iPhone 15, simulator.
So it's loading my mobile app
and I'm using the same app.
The recipes.
I'm gonna go into
styling tab and you see the same,
two components here.
And,
I will
give you the same change
in both platforms.
So let me head back to my, code
source code in VS code
and the first changes
in CSS file of styling sheet.
And this
I'm going to change the background
into green
and hit save.
And then in HTML file
for style invokes component
I'm going to change the icon here
and hit save.
So let's see.
Well so in the mobile app
you can see that, those changes already
took effect.
The background color here is changed.
And then the,
icon in styling
hooks and change into heart from a gear.
Let's see it in our, browser.
So scene changes took effect here.
So here we go.
So this is how you can use,
a local
dev experience to,
for instance, previews.
So just to recap quickly, first
you want to, enable your local dev or,
preference in setup and then,
you want to make sure
that,
the lightning dev plugin is installed
and then,
you want to connect your VS code
to your local dev or,
and then you're good to go
and you're good to go.
So now I'm going to share,
some new releases and new features
now for mobile developers who built
these for offline use case.
So for example,
if you use field service mobile app,
then you could try out these features.
So for mobile development,
we focused on
the more streamlined experience
with a number of guardrails
that steer developers
towards successfully building
that work offline.
So here are some highlights.
The first one is for apex.
So apex has some special considerations
when to work in offline.
So when we detect that to usage
we inform the developers.
But these considerations
and point them to helpful documentation
to understand more.
So the second cars are
guardrail is for GraphQL.
So there are some hard limits on
the GraphQL element is used offline.
So for example,
some features, such as attributes,
fiscal aggregations
or aggregations and
physical data filters and
mutations are not quite supported
for sign use.
So that's something
that you need to be aware of.
And we give warnings
to with that defect.
So all of these ESLint
rules are available in, Salesforce
Mobile extensions for VS code
and can be applied to any
c offline project.
So here is a more, mobile enhancement.
Now GraphQL pagination
is supported for sign use.
It was already supported for web.
So what it will enable
developers to do is you can return
all the records
that you desire for,
going beyond the 2000 record limit.
And you can, prime
these, for offline usage. Right.
And then you can show
your record list in smaller increments,
to users
for better
performance and better user experience.
So one thing, cool thing to note, about
this is
if the user creates new records
while the device is offline,
then this new records are reflected
in your,
paginated records
list in the correct order.
So that's pretty nifty.
So and last
but not least is, now we're giving you,
capability to have built, lightning,
web components that can be customized in
field service, mobile app builder.
So you might have seen
the similar feature
in, Lightning App builder
not available for field service.
Mobile builder.
So as a developer, to leverage
this feature,
you need to set, the properties
you want to expose in
the Json, XML file.
And then your admin can customize and,
set them in, the mobile app builder.
So here is an example of
setting the properties for
inner component.
So you can set the number of teeth,
the name, the,
size of the diner and,
dietary preference.
That's,
how easy it is with no code changes
and how it looks like in,
the metadata file is,
you here, we targeting,
lightning surface.
So, Acme Corp..
And then, added to five properties
that you saw in, Mobile Builder.
So as you can see,
we have a number of great enhancements
for C and, mobile development.
And we cannot
wait for you to try them out
and share your feedback with us.
Thank you. And back to you, Carolyn.
Thank you. So moved.
So the chat's going crazy.
So we've got some stuff for you.
Make sure you keep
posting your questions, in the chat.
But let's get started.
Sue said she's very much
looking forward to local dev,
so that's great.
And Jason Ross,
I hope I get this comment right.
Well,
I don't know what part
that was for,
but I hope I,
I hope I represented that well.
There's actually one question
that has come up
multiple times from the audience.
It looks like both Kelly
and Caleb are asking,
Will the local dev experience
work with Code Builder,
and if so, how do they set that up?
I have to
get back to you on that
in the trailblazer community.
Okay, we will do that
because that seems to be
a popular question. So we'll mark that.
But let's talk a little bit
about, caching.
Do you need to turn off caching
to take advantage of the local dev?
So not anymore.
In the previous iteration you had to.
But you know,
we made a lot of changes
and kind of rebuilt,
you know, this whole experience.
So now you don't have to make
any changes to your, caching. Great.
Use local dev. Great.
And then tell the audience
a little bit about the plan
for GA release of the new local dev
because they're excited.
Yeah.
We actually, as we speak,
we are working on, making it
generally available.
We target in spring 25, so.
Great. Yeah, great.
Thank you. Thank you very much.
I know you'll be back in
just a bit to do general Q&A,
so keep those questions coming
because the innovations keep coming.
So up next we have Daniel.
Sorry, Daniel.
Larry senior developer
advocate who is going to dazzle us
with the latest data cloud updates.
Daniel, over to you.
Thank you. Carolyn.
Hello Data Cloud
developers, I'm
so excited to come to you today
to talk about all of the innovations
that are coming out in the Winter
25 release for Data Cloud.
The very first innovation
that I want to cover is Data Cloud one.
A lot of questions used to come in
with people
that wanted to use their data,
cloud data,
and another org
that was in a multi org type situation
where they had multiple Salesforce orgs,
and previously you only had the ability
to use data cloud in your home org.
Well, now
with Data Cloud one,
you can share your data cloud data
that's stored in your Salesforce home.
Org to multiple companion works.
Also,
we get a lot of questions
about encryption.
When is encryption coming to Data Cloud.
And now in the Winter
25 release,
I'm excited to announce that we have
platform encryption available
for Data Cloud.
So now your data
will be even more secure
than it ever was.
And this will encrypt,
data in Data Cloud at rest.
And then, sandboxes.
Sandboxes have been a huge craze
ever since they released
in beta over the summer.
And we have the ability
now for you to merge your changes back
from sandbox
to other sandboxes,
as well as your production environments.
So let's go ahead and talk about
Data Cloud one
a little bit more in depth.
So with Data Cloud one
a lot of you all are multi org type
customers using, multiple
data, multiple Salesforce orgs.
And you want the ability
to see your data cloud,
data and multiple companion orgs.
And so that data will never ever
be physically moved from your home.
Org.
It will always just be virtually shared,
to one or more companion orgs.
I'm really, really excited for you
all to start using data, cloud
and all of your Salesforce orgs
with this new enhancement.
And speaking of using all of your data
in all your Salesforce orgs,
what are some of the
things that you can do
now that your data is accessible
in your companion orgs?
Well,
you can use all of those cool things
that you're using before
in your homeworks.
Like segmentations calculated insights.
You can start building reports
on your data cloud data.
So really, really excited about this
new data cloud
one enhancement
that came out in the Winter 25 release.
And I can't wait to see
all of the things that you build
with the Data Cloud one.
All right.
Let's take a quick look at a quick demo.
Look, how Data Cloud one works.
So here I am in my coral
cloud resorts org and Coral
Cloud Resources, one
fun, fun hotel chain located in Hawaii.
And, Angela might have
been there already before.
And, I'm
going to just go over here
to set up two Data Cloud.
Shut up and click in here.
And you can see now in set up,
I have this new thing
under Salesforce Integrations
that says Data Cloud One.
I'm just going to go ahead and click
Data Cloud One.
And you can see that
Coral Cloud has already connected
two of their other hotel
chains, to their data cloud home org.
So now they're sharing their data
cloud data with their companion orgs.
All right.
Let's look at this
Pixel Harbor companion org.
So right over here
now I'm in a separate Salesforce org.
So Pixel Harbor is a different
Salesforce org.
So again
Coral Cloud Resorts is in
this multi walk type of situation.
And I now have this app here
that says Data Cloud one.
And I can see all sorts of things.
I would previously only
be able to see it myself.
Force data cloud home org
which are data models.
I can see Data Explorer.
So I can explore some data,
calculate insights and more.
So really really awesome
to be able to share that data.
Cloud data. Now to companion orgs.
All right
let's look back
at our other feature enhancement.
So platform encryption for data cloud.
So you are now have the ability
to create your own customer manage
keys with platform encryption
for data cloud.
These customer managed keys
will encrypt your data in Data Cloud.
You also have the ability
to rotate these keys
as often as you would like.
And when you rotate
the keys that you've created,
that is going to leave
all of your previously encrypted data
encrypted under the previous key.
And then each data that you create
as of the
as of the time or the time
endpoint that you created,
your new customer manage key
will then be encrypted
with your new customer managed key.
Now, one of the things
I do want to call out
is that you do need Shield
in order to use this feature.
That is a prerequisite
for being able to use platform
enrichment for data for data cloud.
But this will allow some of you
that had a little bit more,
stricter compliance regulations
that you had to comply to,
to be able to finally
use Data Cloud with,
the ability
to encrypt your data and data cloud.
All right.
Now let's take a look at the demo.
So here I am and set up and under
platform encryption
I now have the ability to see managed
data cloud keys.
So it's already turned on in my org.
So that's why you already see this
toggled as on.
And now after it's turned on in my org,
what I have
the ability to do is
see this new tab here.
Right
next to Salesforce
platform called Data Cloud.
So I'm going to just click
right here on the Data Cloud tab.
And you can see here
I have my first key
that got created
as a result of enabling
platform encryption for data Cloud.
Now I can always generate
a new route key.
And that new route key
will deactivate this current key
that you're seeing right now
on my screen.
And it will create again a new key.
And all the data
that's in my data cloud
org will start being encrypted
with the new key.
So excited.
I know that a lot of people
were wanting this feature
for a long time,
so I'm excited for you
all to be excited as well.
All right.
And lastly, sandbox merge back.
So with sandbox merge back
you now have the ability
to create outbound chain
sets and add
both data
kits data model
objects, data lake objects.
Really anything that you're developing
and designing for data
Cloud can now be added to an outbound
changeset and can be packaged up.
Then you can retrieve that metadata.
You can retrieve that metadata
using the Salesforce CLI and VS code.
And you can create a package
that you can
then promote to other, Salesforce orgs
that have,
other Salesforce production orgs.
You can, promote it to other
Salesforce sandbox orgs.
You can also
even, put this up
onto a source code control
so you can start doing some versioning
and really creating
that dev ops pipeline
and having something to be able
to back up to or recover to.
So really, really excited.
With sandbox merge back releasing.
Let's take a look at the demo.
So here I am in VS code
and I have actually already
retrieved my package.
So if I just scroll up here,
you can see right here
is where I retrieved my package
I already created in my
Salesforce sandbox.
And that gave me the metadata.
So here's some of the metadata
that was retrieved.
And all of that is also right here
under my package name as well.
So you can see I have some data objects,
some data streams,
and some,
some, some data model
objects all retrieved. And my package.
So excited for developers
to be able to start
backing up your code
using source code control,
promoting it
through all the different environments
that you're using
and creating your real DevOps
pipeline here.
All right,
so as you guys are trying
some of these things out,
remember to message me,
take some pictures and screenshots.
I love to see all the things
that you're doing.
And back to you Carolyn.
Thank you.
That's awesome.
I we get a lot of questions
around data cloud.
And I think when you start
to see the amount of innovation
and making it easier,
and the developers
are just getting more
and more productive with it.
So thank you for showing everyone
we do have some questions,
but please
keep posting them in the chat.
So first I'm going to go to you.
I think you covered this,
but are there any prerequisites
to enable platform encryption
for Data Cloud?
I think there was one.
Yes, yes it's shield.
Shield is is what you need in order to.
Great. Yes. Great.
And then adding data kits to a package
that is something that you can
do, correct. Yes, yes.
You have the ability to add data kits.
And this might be a little bit
confusing to people
because traditionally
what you had to do
was you had to previously use,
data kits.
That was the only thing
you had
to move your,
your org data
between other data cloud orgs.
But now you have this new ability
with creating packages
and outbound chain sets,
and you can add those as well.
So you don't have to completely
blow away your data kits,
but you can package those up. Awesome.
So consistently
making our developers more productive.
Thank you, thank you, thank you.
Before we jump into the panel of Q&A,
we just want to make sure
that we are
answering
some of your toughest questions,
but also the request.
So we know that everybody wants to get
hands on.
So please scan this
QR code
and get the new sample app
for Coral Cloud.
I have mine,
I've had it for a couple of weeks,
so if I've beaten you. I'm sorry.
But this is a great way
for you to really get hands on and test
some of the things that you've seen.
Start building some of those agents.
It's a really, really neat way to test.
And then make sure
you check out trailhead
for more information
and more guides on how to set these up.
And, check out this QR code.
Tons of resources around
being release ready.
So everything that you've seen,
we want to make sure
that we are packaging
for you
to take back to your teams and use.
So make sure you
check out those resources.
But let's open it up to more questions.
You're all back on the hotseat.
I've got a few.
I am going to start at this
end of the table. Angela.
One of the questions is from Alex.
Are around the models.
Are they containerized
by production instance?
Like, is the info processed?
It's not used for training
and exposed to other clients.
Can you talk a little bit
about those guardrails?
Yeah.
So everything it's
built on top of our trust layer.
So anything and part of
it is what I said covered as well
with the data masking
is that it's going to,
be encrypted
before it goes out into the elements.
So even as we add different models,
we always have that trust layer,
guardrail
before it goes out to any of the,
the public models. Yeah.
And specifically the zero retention rate
is making sure
that we're not storing
any of that information.
So it
the answer is no,
it does not use your your company data.
But yeah that's great.
And so let's stay with that
theme of trust
because there is a question,
from Naveen,
how does trust come into picture
with the default models being used?
I think this is kind of similar.
But how are we actually doing that
with the trust layer
and go
on a little bit more detail into that.
Yeah, I can
so I think the big thing,
as we mentioned earlier,
the first thing is zero data,
zero retention,
with our vendors
such as OpenAI or anthropic.
So we have those contracts,
and then within the trust layer as well,
you can kind of customize that based on,
you know, the field based masking.
So certain objects
like if you have maybe
revenue as a field
and you really don't
want the model to see that,
then it actually gets slotted and masked
and then sent that to the lab.
And when it gets sent back,
then we replace it
with your Salesforce data
to really ensure that.
And then we also have machine
learning models to detect that.
So I think when you combine
the power of field based
masking, database masking, it
adds on to our existing
trust layer layers and layers. Layers.
You just get better and better.
Yeah. Yeah.
Well yeah.
If I could add like
in let's say
let's take an address for example.
Right.
If it in our Salesforce instance
it might look like 1234 School Street.
But when it goes out
it will look like x y
blah blah blah blah blah.
Like it's not going to be anything
that's really.
Yeah, like a nice label at all.
So you can feel rest assured that like,
there's not going to be anything
that someone could reverse engineer
and think about what it was.
Yeah, sure. Great.
Thank you for that.
Okay, I'm going to pop down to,
one of the questions from Caleb.
He's wondering it
clearly, local
dev saves you from having to deploy
to deploy to sandboxes or anything else.
It helps with in terms of productivity.
Well, you know,
instant previews are great.
Now for iterative development.
So you're making a lot of changes
and you don't have to waste
time, you know,
you know, deploying etc..
Mainly it's enough for that
to now purpose. Right.
So and
it's a great that you can do all of that
in your local environment.
Yeah.
And,
that is the unified,
developer experience.
So you can see, see,
the same changes take effect
now in the mobile app
and also in, lightning app
in the browser.
Okay.
So unified
development experience is a great great.
So Caleb,
if you have any other suggestions
on what you're seeing that might help
you be more productive,
make sure to post that
in the Trailblazer community
and we will. We'll take a look at that.
So let's see here I love my handy pad.
Danielle, do you need shield in each
org that has encrypted
fields and data cloud
or only the primary org?
Or is shield
added to data Cloud as well?
And that question comes from backup.
That's a great question.
So you will need
short shield enabled in each data cloud
org that you want to use
platform encryption.
And Grant, thank you for that.
And then let's stay with you
because Jason wants to know,
does data cloud encryption
sync with Salesforce encryption.
So you don't need to set up
twice for each field.
So that's a great question.
So Data Cloud does not have field
level encryption available
under platform encryption.
So everything will be encrypted.
It won't be on a field level basis.
I have talked to the product
managers about this
and that's something
they're looking into.
But it's not something
that's going to release, right away.
Sounds like an idea exchange idea.
And let's get that loaded up for sure.
Okay.
Coming back around the table,
Angela, Angel wants to know,
should the flow be auto
launched or record
triggered for Agent Builder?
It doesn't have to be.
But most of the flows
are because you're asking it
to do a specific action and and be.
Yeah.
Triggered in some way.
So usually they are great. Okay.
I'm actually going
to bring in everybody
to this next question
because I've heard it a lot.
I heard it a lot at Dreamforce.
We've been hearing it often,
and the topic is around,
is it necessary to have data
cloud to power agent force.
So you just want to get
everybody's opinion on that,
because I know
there are varying opinions,
and I think we should have a discussion
about it.
So Danielle,
let's start with you
from the data cloud perspective.
And then we can
we can move ranking number one. Yeah.
So as you know,
data cloud
powers I so my opinion would be yes.
You need data cloud
because AI is nothing without data.
So my opinion is yes.
Okay, I like that. I like the opinion.
Let's go down to tech.
Yeah.
So I think that
prompts are really unique
because your prompt
and like
Danielle said honestly
your prompts are only as good
as the data that you give it.
I think what data cloud.
It's not necessary,
but what it does is enrich
the entire experience.
So if you think about
all the different sources
you have in data cloud,
whether that be on the roadmap
coming soon, data
model objects,
data graphs
that you can just go
into the resource picker,
embed that into your prompt
and have that ready to go.
I mean, this really cool functionality
of unstructured data search
that's only doable through data cloud,
because you have to search up
that search index.
What you can also do is you can like,
fine tune your models
and even bring your own
large language model.
And that's only available
through data cloud.
So it really is is that next step
for you to ensure success
and really that your prompt is relevant
to your customer data
in your work for sure.
And I know we're talking
to developers today,
but I think to you,
when we're talking across
all of the customer 360,
there are out of box agents
that are tied to the CRM data.
And so the conversation becomes,
if you know
you're thinking about data cloud,
but you're not quite there
using those out of box agents
to try that agent force
technology could be a step.
And then when you start hitting limits
because your data
might be in other places,
that's a good indicator
that it just makes that stronger.
So any any advice for yeah.
I mean, simple answer is yes.
And for all the reasons that,
Sid and Daniel already mentioned.
But I want to just like,
think about one thing. Right.
Agent force, right
now, as it is today is just
the we're just beginning.
Yeah.
It's gonna become
we're we're we're shipping monthly.
We're developing things
all the time, like on a consistent basis
and more and more of our features,
including the one I own,
which is AI based recommendations
that's built on data cloud.
And if you don't have data Cog,
you won't get that functionality.
And actually, that's
what's going to make it
even more powerful, is that you're going
to see,
become better
at giving you recommendations.
So that's just one example of a feature
that would not
you won't even be able to use it
without data cloud.
So that's one thing I think the
the second piece is,
if you think about as a developer,
you especially
if you're at a large enterprise company,
you have probably tons
of different data sources
in different places.
And how do you bring that in? Is Apex.
But with Data Cloud, it
makes it so much easier
to make sure that your data is clean,
that it's not duplicative.
And, that's the way
we're going to be able to ground,
all of those agent capabilities
in a way that it's
going to give the best experience
and they're
going to get
the most accurate responses.
Great, great.
I we keep talking,
but we are going to data cloud to
I keep turning my chair to you.
I think that's a really,
really good point.
You know, you're thinking about it.
You mentioned some roadmap
things, roadmap for data cloud
and how that kind of enhances
some of these things.
Can you speak to some things
you're excited about?
Yeah.
So data Cloud already has Model Builder
or Einstein Studio
that some of you may know about,
which is,
an AI platform that's built
for predictive and generative analytics,
with being able
to use your data, cloud
data and large language models,
as well as in predictive analytics.
And that is, like, like Angela
said, that is dependent
upon having data cloud.
And there's been all types of new things
that have come out
for that, like being able to fine
tune your large language models.
There's, playground
that's released
where you can start, you know,
running your data
through the large language models
and just experimenting
with the type responses
that you're getting.
And they're also adding
and more models
to be able to work on
the predictive side as well.
So you won't get any of that either.
You don't have data cloud, as well.
So okay. Great.
Well, Andrew says hashtag Daniel knows
Data Cloud is giving you those shout
out hi, Andrew.
Okay.
Tell me a little bit
about what you're excited about.
When we're talking about
like, roadmap as well,
I know, you know,
you've been really busy.
But what should developers
be excited about?
Because you are
we have a lot of, newer,
and mobile enhancements coming up.
So for one thing, we're working,
as I mentioned
before, on the GA version for local dev.
Right.
And, a lot of improvements
like supporting more of TypeScript,
programing, and many more.
Yeah, it's it's amazing
how many times
I'm on my laptop or my desktop,
and then I'm also on my phone
at the same time.
So you're having an experience on
both is super important.
So, we've got a couple more questions.
Paul Fox is asking Angela,
this is actually a really good example.
He was trying the demo in trailhead.
He asked it about rock climbing
and it said it couldn't help with that.
He had to specifically ask
about rock climbing experience
to get it to help.
What are some of the best practices
for instructions or topic descriptions
to help combat that and avoid that?
Yeah. Great question, Paul.
And I get that a lot.
I mean,
we actually make this joke
pretty often for those of us
who are parents on the team.
It's like
when you think about instructions,
you have to think about like,
how would you tell?
Like, I have a five year old,
how would you tell a five year
old to answer this question?
You had to be so specific, right? Yeah.
And so maybe in the instruction,
and I had this before
even testing for myself, like,
how I write the name,
for example, like, is it,
is it a name field
inside the custom object
or the object that you're referring to?
Or are you specifying
that it has to match like, email
or things like that?
So being really expensive,
like always use
the name field
or always use
the email field when confirming the,
the request
right in the before you give a response.
And you can always ask
you can always put that
in the instruction.
Like always ask for
an email confirmation
before moving on to provide a response,
things like that,
so that it can be really
prescriptive essentially.
And, and we found that
with instructions,
it's kind of hard to say this,
but like saying always versus don't do
this is better
because when you say don't,
it has to kind of infer
what you mean as the opposite thing.
Right? Okay.
So kind of like going back to the kid
example, it's like, don't sit there.
It's like, okay, but then where do I do
like what do I actually say,
always sit
in this chair versus,
you know, don't sit anywhere else.
So having to think about it
in a different way,
we're so familiar with, decision trees
when it comes to bots. Right.
It was
it was very specific
in doing those things.
And so
trying to figure out how to open it up
a bit,
but be specific in a way, that is, it
almost seems like it's
more of a positive thing,
like do these things.
Yes, exactly. Okay.
So I think with the specific question
around,
oh, we had to say rock climbing,
specific gradients. Yeah.
If you want it to open up you can.
Right.
You can say like, oh,
you know,
give me like looks searched through, or
refer to all of the models
or data objects in
this particular object.
Like you can
always do that
if you want it to be more open.
Great. Yeah. Great.
So we're all learning together.
Test and test and test.
Okay.
Said one more question.
I'm hoping I'm
asking this the right way,
so if we need to reframe it,
let me know.
But can you map a Salesforce
object to a DL object
back to a DM object in
search it
because you can't search
native Salesforce objects.
Okay. Yeah.
So if I'm understanding this right,
you can through
search through native search,
Salesforce objects.
So you go into the resource picker,
you pick that object,
you choose the merge fields
you want gets resolved at runtime.
And over.
The reason
I went through all that
process in my demo
was because shoes had unstructured data
that you couldn't just search.
So the description field was a large
chunk of text
that we wanted
to create the search index for.
And then the retriever is actually
what shows up in Prompt Builder,
because that retriever is what searches
through all the different shoe
descriptions and says,
hey, I like, you know, flashy shoes
and then it understands what does flat.
Yeah, exactly.
Understands what does flashy mean
and which description actually fit it,
even if that word is
an explicitly saying.
So like flashy doesn't mean flashing
like a flashlight straight.
It means there are different
kind of decisions.
That's exactly
what Einstein Search is all about.
Natural language
search and understanding.
Sort of the semantics of it,
like what does it really mean?
And then searching through your,
you know,
it could be knowledge
articles, it could be cases
with just a bunch of data, a text data,
and then pulling that
into prompt builder.
So that's the charm.
So before the answer
would have been no, Carolyn,
we don't have shoes with flashlights.
And now it could be yes Carolyn
we have glitter shoes. Slate.
So you got exactly exactly.
That's what that whole process was.
Yeah.
That's awesome.
Okay, we have two questions.
Danielle, we have one question.
Does, I don't know,
I just maybe
you know what this is referring to,
but does it only
in in change
that will also support DevOps?
Not sure what part that that I think.
I mean, know what they're asking.
So DevOps center is one of those things
that is on the roadmap.
So for DevOps center, right now
you can use it to promote your changes
with Salesforce platform.
And we are working on being able
to integrate data cloud into that.
In terms of the change set piece,
I'm not quite sure
what part of the question
they were referring to there, but,
you you can retrieve a change set,
using the Salesforce CLI
and you can add
all the components for Data Cloud.
They're great. Great.
And I think we had
Kelly's question before,
but she moved,
well, the local dev experience
work with Code Builder.
And how would they set that up of that
to check on that.
And you get that's the one.
Yeah.
As a community,
I just wanted to remind you. Yeah.
They're validated.
Yeah, absolutely.
Well,
I really, really appreciate
all of you taking the time
to show the new enhancements.
And even though we are wrapping up
this particular show,
you can still ask your questions
in the trailblazer community.
Hashtag Salesforce URL.
We'll make sure you get an expert answer
and thank you for tuning in.
Your feedback does matter
and it helps us
make our products better.
So tell us
what you thought of this episode.
Scan the QR code
link on the screen
to take a short survey,
and as a thank you,
you'll be entered
into our sweepstakes
to win one of five Salesforce
certification vouchers.
And there's more release
readiness content to come,
so stick around.
Service cloud is next at 11 a.m.
Pacific,
and then that's followed
by a bonus episode of Health
Cloud at 130.
Come back tomorrow for CRM
analytics, and our admin
preview teams will be here.
All of our episodes
will also be available on demand
and coming up later
this month is Discover Agent Force
what I was meant to be?
You don't want to miss that,
so scan
the QR code on your screen
to register now.
This is a live
launch event
that kicks off on Salesforce
Plus on October 29th at nine
a Am Pacific.
We're going to hear from companies
using autonomous agents
to take action
and drive results 24 over seven.
And then you can get
ready to build your own.
And you really don't want to miss it.
So I'm Carolyn Bathauer
that is it for now
and we'll catch you next time
in the cloud.
Resume AutoScroll
Link copied to clipboard
Developer Preview Winter '25
Learn about the latest developer features including how to capture more events with new Data Cloud integration upgrades. See how you can improve developer productivity and code quality with new Lightning Web Components and TypeScript. Plus discover our top innovations in Prompt Builder, Agentforce, and more.
Starring
- Angela LeSenior Director of Product Management, Agentforce, Salesforce
- Siddharth ShendeAssociate Product Manager, Salesforce
- Danielle LarreguiSenior Developer Advocate, Salesforce
- Suvda MyagmarDirector, Product Management, Salesforce
Developer Preview Winter '25
Learn about the latest developer features including how to capture more events with new Data Cloud integration upgrades. See how you can improve developer productivity and code quality with new Lightning Web Components and TypeScript. Plus discover our top innovations in Prompt Builder, Agentforce, and more.
Health Cloud Winter '25
Discover how you can improve productivity and effectively manage patients with the latest Einstein for Health Cloud innovations. Learn how to enhance your CRM Analytics dashboards with interactive, engaging, and actionable content by incorporating Widget Flow Interactions. Find out how to build dashboards and calculate insights with Health Cloud’s Connected Ecosystem.
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.
