I'm thrilled to be back
here for the second
year in a row since
last year's World Tour.
Now, as the largest
nonprofit dedicated
to serving Americans
over the age of 50,
AARP supports
over 38 million
Americans through
some of life's
most important
and oftentimes
So when a caregiver
calls AARP, it's
usually tied to
a high-stakes
situation related to
health, safety, or
an urgent matter
for a loved one.
So when they call AARP,
it's critical that
they receive clear and
immediate guidance,
because when they don't,
that means they're
left with uncertainty
and time and attention
is pulled away from
those who need it most.
Well, thanks to
Agent 4's voice,
AARP is able to
provide intelligent
support 24-7 to
their members
through channels
that they trust most.
In fact, why don't
we go ahead and
check out a visionary
experience of what
that feels like for
an AARP member. A
moment to
connect to AARP.
Should we give
that another try?
This is the one who
knows it's real.
Let's give them
one more try.
phone. We'll
do the trick.
RP customer
agent. Thanks for
calling AARP and
thanks for being a
member. How can
I help you today?
I recently became a
caregiver for my mother,
and she's actually
moving in next week.
And I need to make my
home safer for her,
but I have a lot of
steps near the entrance
of my home. What do
you recommend I do?
Margaret, that's
wonderful that you're
taking steps to make
your home safer.
I'm going to send
our HomeFit guide to
your email on file. It
has a lot of helpful
information about
improving home safety.
It was good. It was fine.
Agent Force was
able to recommend
the right PDF for
Margaret to read in
order to get her
question answered.
I don't know about
you all, but if my mom
was moving in next
week and I had to prep
my house for her, I
don't know if I have all
that time to read
through a massive PDF.
Why couldn't Agent
Force just give me the
specific answer I
needed from that guide?
Now, if we look at the
home fake guide that
Agent Force recommended,
it's a 36-page
PDF filled with diagrams,
visuals, tables.
And when this gets
sent to the agent
and the underlying
large language model,
meaning it gets
converted to raw text
like what you see
here, which is much
harder for the
large language model
to make sense of.
And this is why
Agent Force was
unable to provide that
specific answer that
Margaret needed.
And this is the problem
that Intelligent
Context has been designed
to solve. So what
we're going to do now
is go ahead and upload
that HomeFit guide to
Intelligent Context.
Now, by a show
of hands, who
here has to deal
with searching
through PDFs on
a daily basis?
I mean, look
around the room.
Here in this city,
in D.C., we deal
with a Potomac of
PDFs flowing through
our systems every
day. And we know
just how critical
every single page
of that PDF is
important for context.
So now that
Intelligent Context
has processed
this guide, you'll
notice that it's
generated chunks.
Chunks are
essentially meaningful
And this is how the
structure and relationships
between all that
data is preserved. So
why don't we go ahead
and ask Agent Force
that same question that
Margaret asked earlier
and see if it comes
back with a response.
There we go. Now Agent
Force was able to
extract the right
answer for Margaret from
the guide. So it recommends
that she installs
a ramp and adds
handrails. This is great.
But there's one thing
that we need to catch
here. If we look
closely at those
generated chunks, it's
actually exposing some
PII. It's exposing
the names and contact
information of
internal employees. And
this is happening because
Intelligent Context
is processing
multiple documents,
including internal
documents. So we need to
make sure that that
data is masked. And this
is where Data360
Governance comes into
the picture. So with
Data360 Governance,
you can actually define
custom policies for
your agent. And Agent
4s makes it even
easier, where you
describe it in natural
language and it
automatically gets added.
And then once we
activate that, you'll see
that it's redacting
that sensitive data.
So, so far, what
did we do? We went
ahead and uploaded
that HomeFit guide.
We made sense of
the data through
intelligent
context, and we
masked sensitive
information.
We need to go
build our agent.
And this is where
the new Agent
Builder comes
into the picture.
So here's our new
Agent Force Builder.
What I'm going to
do here is I can
actually start from
pre-built templates.
So organizations
like AARP can get
started quickly
with topics like
volunteer
management, volunteer
participation,
and go live fast.
Now I want us to
double-click into
this topic right
here around
providing guided
advice to caregivers.
You'll notice that we
actually have explicit
instructions to
ensure the agent never
answers any questions
related to providing
medical or occupational
therapy advice.
And the way we do
this is through
this conditional
expression.
And this is what
ensures a level of
determinism. We're
injecting this deterministic
logic to ensure
the agent never
deviates from non
-negotiable instructions.
And this experience
is powered by Agent
4 script, which is
going to be critical
for industries,
regulated industries,
and organizations
like ARP to ensure
that consistent and
compliant experience.
Now, as a final step,
let's add that context
to our agent. So
we're going to go
ahead and add the home
fit guide that we just
processed through
Intelligent Context.
We're going to
activate our agent.
and why don't
we go ahead and
give that call
one more try
and see what we
come back with.
AIRPE member agent,
thanks for calling
and thanks for being
an AIRPE member.
How can I help you today?
Hi there. I
recently became
a caregiver
for my mother.
She's moving in
next week and I need
to make sure my home
is safer for her,
but I have this
huge staircase for
my home. What do
you recommend I do?
Awesome. Super
clear, super
specific. Thank
you so much.
how was that?
Much better.
And this is a new
experience that agent
force and intelligent
context is really
powering. And now
you'll notice that
Margaret no longer
has to spend that time
searching through
that massive guy. She
can channel that
energy towards really
caring for her mother.
And you'll notice
along the way,
the agent never
compromised safety,
accuracy, or compliance.
With these
grounded agents and
that trusted
system of context,