FedEx is an
agentic enterprise
that operates on
a global scale.
But even with
small business
customers signing
up around the
world, they had
a challenge.
Customer signing up
is not the same as
a customer actively
shipping. And let's
face it, FedEx doesn't
make money when
a customer is
signing up. They're
making money when a
customer is shipping.
They had disconnected
shipping data from
marketing data, so it
was really challenging
for them to see the
complete view of their
customers, and that
was slowing down sales.
with intelligent
context and with
agent force, they
were able to get
a complete view
of their customer.
They were able to
take those dormant
customers and
they were able
to turn them into
active shippers.
They did this
with Data360, the
next generation
of data cloud,
unifying your
enterprise data, your
customer data, your
unstructured data through
zero copy. And
what's more is that
now Data360, it stores
your agent-force
conversations, your
customers' preferences,
and that gives
your agents memory.
One more thing that
Data360 does, and this
is one of my personal
favorites, is governance
is not an afterthought
with Data360.
Your compliance teams
are going to love this,
and I'm actually an
ex-IT auditor, so I
love this, is that when
you create compliance
policies and governance
policies, they are
applied to all of your
data automatically.
Now, we know that
agents, they need data.
But more than data,
they need context.
That's what helps
them be intelligent.
But in most enterprises,
that's actually
very, very difficult.
The same data, it sits
in your lake, it sits
over here in a warehouse,
it sits in a business
system, and how
many different formats
are there for date? And
you have to try to
reconcile all of this.
Well, Informatica,
it brings cataloging.
It brings lineage,
data lineage.
And Informatica brings
a golden record.
Now, there's one other
thing that agents
can benefit from.
Agents rely on signals.
They rely on signals
from your systems.
Think of a signal
like, in the
case of FedEx, a
shipment that's
running late or
it's delayed.
Well, MuleSoft,
what it does, it
connects all of your
systems to be able
to surface all of
these operational
signals. And those
signals trigger
your agents to
take secure action.
And Data360, well,
Data360, it wraps
it all around. It
unifies your data,
and it brings
trusted, governed, and
intelligent context
to every single one
of your agents and
your applications.
So let's do a double
-click on Data360
specifically through a
demo. And to do that,
please welcome again,
Kristen and Abby.
Now, I had
shared with you
that FedEx had
a challenge.
They had all these
dormant customers, and
we needed to turn them
into active shippers.
Well, we solved
that problem with
Data360, but
they had a new
problem, but it
was a good problem.
The problem was
they started getting
a lot of questions
from customers
now that they
were shipping.
Questions like this
one. What's the best
way to ship a 70
-pound box of wholesale
electronics from
Austin to London?
You'd think that's
kind of easy. That
should be a simple
question. But the
reality is the answer
is very, very complex
because it depends
on the delivery zone,
it depends on the
shipping services,
import, export, or
any other nuances
that are very specific
to that delivery.
Well, as a FedEx
seller, what I have
access to is
through Slack, I can
access AgentForce.
That's how I can
answer all of these
questions at scale.
that the
customers ask me.
What's the best way
to ship this box?
Now, Agent Force
provides a response.
Excellent. But it's
not very helpful.
Because instead of
giving me the answer,
what Agent Force
is doing, it's
telling me to reference
a 200-page PDF.
All right,
let's humor it.
Let's check out that PDF.
But when I'm scrolling
down, I can see
there's a table.
There's charts. There's
headers. There's
footers. there's an
index, there's a lot.
Now, I can process
this. I can understand
this because that's
a human skill. I
can understand that
this is a chart.
That's a table. I can
see the symbols and
I can see the index
key. That's easy for
me to understand.
But when you take
a document like
this, an unstructured
PDF, and you
process it or
upload it the
traditional way, you
know what happens?
It gets completely
flattened. It turns
into raw text. In this
raw text, it's losing
all of the meaning.
Let's find that same
line. So we can see
England over here.
We see England. We
see UK. We see D.
What does D mean? We
knew through the PDF.
We could see that D
was the zone. We could
see that the check
marks for what were the
services that were
available within that zone.
But now it's
flattened. So it's very
challenging for an agent
to be able to understand
this and make meaning
and context of
it. Well, this is why
we are introducing
intelligent context,
data 360 intelligent
context, because
now when we take an
unstructured PDF like
this and we upload it,
what happens is we get
intelligent chunks.
We can start to see
all of the pieces of
that PDF logically
parsed out into an index.
We scroll down, we
find that same entry.
You know what we
see? We can see that
it is England. We can
see that it's zone
D. We can see the
export and import
laws and all of the
standards that apply.
This becomes a
meaningful slice of
information that an
agent can use to reason.
let's do a quick
test. So we're still
within Data360. We're
going to ask Agent
Force that question
one more time. Now
that we've uploaded
this document using
intelligent context,
look at that.
We get a response.
We get a response
that's accurate and
that's complete. It's
not referencing me
to the PDF anymore.
But it does look like
Agent Force is being
a little too generous
because it's also
exposing some internal
FedEx employee contact
details. The reason
why that's happening is
because that PDF, it's
not an external document,
it's an internal
document. So that's
okay to have internal
contact details for an
internal document, but
that's not something
we accidentally want
exposed to a customer.
Well, like I said
earlier, the inner
auditor in me is geeking
out right now because
with data governance,
it's built in.
We can create a
policy right here. And
instead of having to
do many keystrokes
to be able to
configure that policy,
With AgentForce
embedded within Data360,
all we do is use
natural language to
describe the policy
that we want to create.
In this case, it's
a masking policy.
We want to hide that
information so that
it doesn't get
exposed to a customer.
AgentForce does
the work for us.
We just validate
that it's the
right policy.
We hit activate.
Now that we've got
this policy activated,
let's jump
right back into
production. So as a
FedEx seller, I am going
to ask that question
one more time.
What's the best way
to ship a 70-pound
wholesale electronics box
from Austin to London.
The answer is complete,
it's accurate,
and most of all, it's
appropriate. It's
not exposing any
sensitive information.
And this is how
FedEx has been able
to take their seller
productivity and
reach it to a new
level by taking
raw data, converting
that data into
context and context
into intelligence.