The advertising industry is entering a new "agentic era" driven by advancements in AI, necessitating the adoption of open, interoperable standards like the Ad Context Protocol (ADCP) to ensure transparency, efficiency, and innovation.
Mind Map
Click to expand
Click to explore the full interactive mind map • Zoom, pan, and navigate
Hello and thank you for joining us for
the launch of a new industry protocol,
ADC CP. Over the next hour, you'll hear
from the pioneers building the future
advertising as Aentic AI transforms the
ecosystem. We will dive into the
technical architecture of ADCP, see live
demos of the protocol in action, and
discuss the transformative impact of
this technology. Let's get started.
Please join me in welcoming Amanda
Dvito, chief marketing officer of Butler
Till. Hello everyone and welcome. We are
standing at one of those rare inflection
points, the kind you only recognize in
hindsight. Except this time we know it's
happening right now. Advertising is
entering a new phase. Not next year, not
5 years from now, today. For decades,
our industry has evolved through big
defining shifts from print to digital,
from desktop to mobile, from manual
buying to programmatic. But what's
unfolding in front of us now is bigger
than any of those transitions. Because
we are stepping into something
fundamentally new, the agentic era of
advertising. I'm Amanda Dvito, chief
marketing officer at Butler Till, where
I lead marketing and growth. We're an
independent, woman-owned and led,
employeeowned benefit corporation media
and communications agency. And we're
here because our clients count on us to
see what's next before it's obvious and
to help them turn change into advantage.
For us, this isn't about chasing the
next shiny object in adtech. It's about
making sure the buy side, advertisers,
planners, and strategists have a real
seat at the table as this transformation
unfolds. We're here because this
evolution toward agentic advertising has
the power to create more meaningful
work, more transparency, and more value
than ever before. Not just for brands,
but for the people behind them. That's
why we're involved. Because our clients
deserve a world where innovation
actually works for them, not around
them. We find ourselves at the perfect
storm of timing, the intersection of
capabilities and momentum. On one side,
there's the explosion of AI capability.
Foundation models like Claude, ChatgBT,
and others have reached a level of
reasoning and planning that makes true
agentic behavior possible. Agents that
can think, plan, and act autonomously.
And with Anthropic's launch of the model
context protocol late last year, the
technical rails are finally here. The
infrastructure that allows agents to
communicate directly with systems,
platforms, and each other. That's the
capability part. On the other side, the
momentum undeniable. Advertising itself
is evolving in real time. What started
as agents who co-pilot with us on
creative and personalization has now
reached the next frontier, media buying
itself. And here's what's so exciting.
This isn't just about automation for
automation's sake. It's about unlocking
focus. Agentic systems can handle the
optimization and orchestration that
consume so much of our time today,
freeing humans to spend more time on
what really moves the business forward.
Insight, creativity, and strategy.
That's not the next shiny thing. That's
a shift towards smarter work and better
outcomes. From Google exploring MCP for
its ads API to Amazon launching Aentic
creative tools, major platforms are
already moving fast. This isn't
theoretical. It is happening right now.
And that's what makes this moment both
so incredibly pivotal and exciting.
We're not just observers. We're
architects of what comes next. Let's be
honest, the opportunity is massive. I
know you're all feeling it. But so is
the responsibility to get it right.
Today the advertising ecosystem is
deeply fragmented. A plethora of tools,
platforms, and systems that barely
communicate with each other. They create
complexity, inefficiency, and
duplication. Every AI integration is
bespoke and customuilt. We're building
in silos when what we need is
interoperability. The window is open
right now for the industry to come
together and define shared standards
before proprietary walls go up again.
This is that rare moment when we can
choose collaboration over competition.
When we could decide to build a system
that works for all, all of us, not just
a select few. And that's what today
represents. You are here at the founding
moment of something new. a coalition of
industry leaders, innovators, and
builders all coming together to launch a
shared framework for agentic
advertising. It's being built by the
industry for the industry, open,
interoperable, built to evolve. Over the
next hour, you'll see what that looks
like. live demos, real use cases, and
the beginnings of a model that can
transform how we work across every part
of the ecosystem from yield and
measurement to planning and buying. This
is not about one company solution or one
side of the marketplace. It's about
collective progress because the future
we want will only exist if we're going
to build it together. The question isn't
whether aic advertising is coming. It's
already here. The real question, the
real question, the one that will define
this next decade is whether it unfolds
through fragmented proprietary systems
or through open interoperable standards
that create a transparent, trustworthy,
thriving ecosystem. This moment, this
event is about choosing the latter. So
good for you for being here today. So
consider today your front row seat, not
to watch the future unfold, but to help
build it. Because the agentic era isn't
something happening to us, it's
something happening through us. And it
starts now.
Welcome to the start of the agentic era.
Now, please welcome Anne Cochland, COO
and co-founder of Scope 3 to tell us
more about the agentic era of
advertising, followed by Bashko Milikch,
chief product officer at Optal for a
deep dive into the protocol itself. Wow,
I'm excited and we're just getting
started. Hi everyone, I'm Anne Cochland
and I'm the COO and co-founder of Scope
3. For us to really understand the power
of aic advertising, it's helpful for us
to start with what marketers really
want. So, let me paint that picture for
you first and then we can walk through
why the three existing paradigms of
digital advertising are not helping a
marketer achieve those needs. So, here's
a list of what marketers want.
Fundamentally, the first one on the top
of the list here is scale. They want to
be able to buy media everywhere that
makes sense for their campaign. But I
would add that they want to do that with
no marginal cost. So, no additional
money or time or focus just because
they're adding another channel or a
publisher. They want creative and media
to be really really connected. Not
bidding on ad space and then separately
choosing creative but generating the
exact creative for each spot, each
moment, each audience. And so that kind
of probably what ties in there is the
idea of being native. No lowest common
denominator. Everything feels like it
makes sense to be together. They want
outcomes, not just impressions. So real
business results, not proxy metrics. And
continued rhetoric about what that means
within the industry. And then these kind
of all tie together a little bit. uh but
you know thinking about the quality of
execution how creative and media being
together native works making sure that
you're driving to the right outcomes and
scale also I think it's really important
to make sure that there is no difference
between a10,000 buy or $100,000 buy or a
million dollar buy so everything the
quality of execution the ability to
execute should be the same and then of
course they want sustainability
transparency and working media that
actually works for them so that is the
vision that is what digital advertising
should deliver so why hasn't it so let's
start with the insertion order model. So
with the direct IO as a brand, you
definitely get the full power of the
underlying media. You can do really
truly custom campaigns with rich bespoke
executions, custom formats, native
integrations, content partnerships.
There's a real richness and diversity in
what publishers have to offer. This
obviously leads to better outcomes. Um
there's a real direct connection here.
So you you're getting the sustainability
benefits, the transparency, and the
working media, but it really doesn't
scale. Every single deal requires human
negotiation. There's a sales team with
finite resources. It's powerful, but
fundamentally it's oh that's pink. Let's
go for red. It is very limited by human
bandwidth. Uh the publisher can only
sell as much as their sales team can
handle. The second paradigm trying to
fix the big red circle on the previous
diagram. For a publisher to be able to
offer an ads API, they do need to have
the technical infrastructure to expose
their inventory targeting capabilities
and be able to give campaign management
capabilities to the marketer. And we've
seen that, you know, that can really
demonstraably drive scale in the case of
Meta and Google. But the problem is the
reason why I have to unfortunately mark
this red still is that APIs are custom.
Buyers can't integrate with many APIs.
So they choose and prioritize Google and
Meta because those are where they see
the biggest reach and the maximum
performance drivers. and then buyers
just stop. So the engineering return on
investment isn't there. It's not enough
to be able to integrate with everyone
else and that is a big problem for the
publisher. The third paradigm is
programmatic and I finally been able to
color this circle in green here because
we do have automation at scale. The
technical ecosystem that we've been able
to build through standardization through
different protocols means that we can
connect in theory any buyer to any
publisher. But fundamentally what we
should be doing as an ecosystem is
trying to translate the brief from the
marketer here and make it work so that
the marketer can buy inventory that
really fits their campaign brief in a
way that the publisher gets paid fairly.
And there are lots of things that break
down in terms of these circles here.
First of all, this brief is rich. It's
detailed. It's strategic. It needs to be
compressed and simplified to fit through
these plates. And then on the publisher
side, all of that first party data
richness, all the unique capabilities
are not accessible up through the
programmatic stack in the way that they
are through direct IO buying. So while
the DSP here that I've colored in uh
pink is the focus point for decisioning,
it's making decisions without access to
the best information. Buyers have to
spend significant money on third party
data because they can't access that
first party richness that actually
exists on the publisher side. So working
media is around 41%. I'm going to mark
this one red here. And I'm going to mark
all of these an orangey color. You can't
do all of the fancy stuff. You can't
access unique inventory, the custom
capabilities that make publishers
valuable. The promise of programmatic
largely has gone unfilled. And my
concern is that if we just sprinkled AI
on top of each one of these existing
boxes without actually rethinking the
fundamental architecture, we just
perpetuate all of the same problems at
an even higher cost. So these are the
three existing paradigms. Fundamentally
what we're left with is that brands want
outcomes. They want advertising that
actually works across all the existing
capabilities and platforms. And we talk
a lot about the future of advertising,
influencers, AI, chat, all of this stuff
that didn't exist 10 years ago that are
not going to work within any of these
three paradigms because of the
challenges that I've outlined.
So, what if a publisher could build an
agent? Instead of auctioning every ad
and hoping someone bids, instead of
having to create a sales team that can't
work at scale, you can actually take
orders all the time and make the best
decision about which ad to show to each
person at the right moment, the full
vision of what we thought digital
advertising was going to be. And for a
buyer, this means that you get the full
scale of programmatic, but you don't
lose any of the richness of the
opportunity and of the connection that
you have with the buyer. So this is why
agentic advertising is powerful. One,
it's about helping publishers monetize
better. Leveling the playing field
against Meta and Google once a buyer
agent is set up. Adding another
publisher is just another conversation
between agents. No new engineering
integration. No additional sales
negotiations. It helps advertisers get
better performance and outcomes than
they can from wall gardens. It makes it
possible for anyone, whether you're a
massive holding company or a small
business, to get access to the full
breadth and power of digital
advertising. And the way that we've made
this possible is by building a protocol,
a way to connect buyers and sellers.
There are lots of people working on this
and I'm now going to turn this over to
one of the architects of this new
protocol, the ad context protocol to
show you and talk to you about how it works.
works.
>> Hi everyone, my name is Bosko Milik and
I'm a longtime adtech nerd and the chief
product officer and co-founder of
Optible. Optimal is an identity
management and data collaboration
platform and we're committed to the
development of open aentic advertising
standards and together with a growing
community. We've been collaborating on
the development of the ADCP protocols
which we're really excited to share with
you today. So I'd like to invite you to
join us and contribute to ADCP's design
and development. So what's ATCP? ADCP is
short for AD context protocol. It's an
open extensible set of protocols
designed to support intelligence
software agents in planning,
transacting, and optimizing advertising
campaigns across the digital media
ecosystem. It defines a standardized
interface for both agent to platform and
agent-to-agent communication. Now, why
do we need a protocol for agenic
advertising anyway? Well, the rise of
autonomous agents in advertising is no
longer theoretical. Advertising agents
are software that communicate with
people, large language models, data, and
ad platforms to get work done. And all
corners of the ecosystem are working
towards and adopting agents, making
better use of time, resources, and
ultimately driving better outcomes. Now,
agents not only know things, they can
now do things. Marketers are piloting
brand agents today and platforms are
integrating agent logic for a variety of
use cases in digital advertising
spanning from creative design to signal
building activation to workflow
automation. But we've got a problem. We
lack the common language to connect
these proprietary and siloed aentic
advertising systems. And this results in
a fragmented and inefficient ecosystem.
And so ADCP is that missing layer and
common language. It provides a means by
which buyers and sellers of advertising
can connect their agentic systems uh
which enables scaled execution of
relevant ad campaigns across an
inherently fragmented landscape. Now
ADCP is designed to allow advertisers or
agencies to connect to media companies
or publishers directly so that they can
run and optimize ad campaigns toward a
common goal. So who's AdCP for? Well,
for starters, advertisers and marketers.
AdCP creates more working media to drive
outcomes and intelligent software agents
can now plan and buy media on your
behalf, but they need a shared
infrastructure to operate across
channels. And so agents that are built
on ATCP can deploy, optimize, and adjust
campaigns continuously and operate
outside of proprietary silos. They can
tap into publisher signals to improve
performance. So if you're an advertiser,
you should evaluate Agentic buying use
cases and ask your media platforms and
partners if they plan to support ADCP.
Now for publishers and media owners,
ADCP creates more opportunities to
monetize your audiences. It defines a
standard way of representing inventory
pricing and packaging. So, if you're a
publisher, you should ensure your
inventory and data are represented
according to NCP standards and prepare
your monetization strategy for an
agent-driven marketplace. Consider
partnering with platforms that are
offering sales and signals agents and
integrations. Now, for agencies, ADCP
enables you to bring your proprietary
planning, data insights, and
partnerships to bear through customuilt
buyer agents. It establishes a standard
that ensures campaigns can be executed
by intelligent systems across publisher
platforms. Ultimately, it accelerates
campaign planning, activation, and
enables continuous optimization. So, if
you're an agency, you should consider
piloting some agent-driven use cases
with ADCP uh with select client accounts
to prove out the efficiency gains. And
finally for adtech platforms, ATCP
enables you to expose your capabilities
to externally agentic platforms through
a common standard. It keeps ade
platforms central by standardizing how
agents access inventory decisioning and
measurement. So if you're an ad
platform, you should consider exposing
more of your capabilities to external
agentic platforms through the common
standard and evaluate your platform's
compatibility with ATCP. Now let's
briefly look at how ADCP works today. So
ADCP today is made up of two
foundational protocols. The media buy
protocol is one and it's ADCP's core
advertising automation interface
providing standardized tasks for
managing the complete advertising life
cycle. So it specifies a set of tasks
enabling discovery and planning,
campaign execution, creative management
and optimization workflows. So a media
agent operated by a buyer can use the
media buy protocol to connect to
multiple sales agents enabling
end-to-end campaign execution. Now the
protocol is asynchronous by design. So
it's designed to enable a publisher to
require human approval at any stage of
the workflow. So for example, creatives
may require publisher approval before
being activated in their ad server.
The second protocol in ADCP is the
signals activation protocol. It enables
agents to discover, activate, and manage
data signals through natural language.
It's designed so that both buyers and
sellers can interact directly with their
signals agents in order to build and
target signals and inventory that's
relevant to the advertisers's brief. So
the media buy and signals activation
protocols are designed to work together
to enable intelligent signal and product
discovery and to enable direct campaign
execution between buyers and sellers.
The goal is to enable a transparent and
efficient marketplace where both buyers
and sellers work together to deliver to
the marketer's brief and desired
outcomes. Now let's look at some of the
design goals of ADCP. A key ADC CP
design goal is to make the protocol
compatible with both MCP and the
emerging A2A open protocols. Now these
protocols are asynchronous and taskbased
by design and this is because some
operations may take more time like those
where a human intervenes. So for example
when a seller requires creative approval
before campaign activation. Now the idea
is to have agents automate repetitive
and errorprone human tasks and add
intelligence to ad product and signal
selection. Basically we want to give ad
ops superpowers. And speaking of uh
human intervention, it's important that
publishers be able to require manual
approval before any operation. So for
example, creative auditing and pricing
approval are good examples. The current
media buy protocol codifies a generic ad
product and campaign structure that can
be made to work with sellside ad
decisioning systems like publisher ad
servers like GAM for example. It can
also plug into specialized ad servers
like Triton and others. Now, finally,
these protocols, they're designed for
both agenttoplatform and agent-to-agent
communication. And that implies the
ability to share prompts and context
between buyer and seller agents. The
mental model I like is that of agents
talking to each other in natural
language, collaborating as colleagues as
opposed to just making API calls or
clicking buttons in a UI. Now, here's a
picture of what multi-publisher campaign
execution looks like to a buyer media
agent from the standpoint of the AdC
protocol. So, Acme Corp is a buyer and
they have a buyer agent that chats with
the media company's sales agent to plan
and activate ad products that are
powered by relevant signals from both
Acme Corp and its media partner. Now, in
this example, campaign activation and
reporting is enabled using the
publishers ad server. Reporting data is
made available to Acme Corp through the
media buy protocol. Now in parallel,
Acme Corp's media agent also chats with
another media corporation sales agent to
plan and activate ad products backed by
different signals and inventory. So a
similar reporting feedback loop exists.
Now there might be differences in
format, signals, and inventory mix
naturally across publishers and the
buyer can refine her campaign plan and
execution along the way. Each party is
given the opportunity to create and
propose ad products and signals that
strive to optimally deliver the
marketer's intent. So where are we today
and where do we go from here? Well,
several founding group companies are
starting to test and roll out
activations using the ADCP media buy
protocol today. Many others have started
to look at the protocols and build buyer
sales and signals agents to
interconnect. The community is growing
quickly. Beyond growing adoption, we
want help from those interested in
refining, improving, and expanding the
current ATCP protocol specs and open
source projects. An initial technical
steering committee has been formed and
will help establish guidelines and
governance. And we have some dedicated
discussion channels for the various
protocols that make up ADCP and are
public and you can access them uh in the
Agentic Ads Slack community. Now,
finally, several community members are
also interested in contributing to the
open- source sales agent in GitHub. And
there's a lot of work to do on this
front to get to a productionready open
source sales agent. Uh we have a really
good start and we would like to invite
interested and capable companies and
individuals to help out with this
effort. Well, thank you for joining us
today and stick around to see a working
demo of ADCP in action.
>> So much work has already been done in
bringing ADCP to life. Now let's see the
protocol in action with leaders from
scope 3 and swivel.
>> So what we're going to see today is the
interplay between a variety of systems
as we go through an agentic media buy.
Um so you'll see us start um with an
agent powered by scope 3. Um the agent
will then communicate with the seller
agent on the swivel side. Uh and then
you will see finally as the agenda
transaction comes to a conclusion you
will see the objects made um in the
downstream ad serving platform. All this
orchestration is done by agents uh it's
a real demonstration um and it also
includes a real publisher uh our
publishing partner LG ads. Um you'll see
us demonstrate a human in the loop to
approve the media buy. Um and lastly
you'll see creatives traffic uh for the
end of the execution. Um, so we're
excited to show you what this what this
looks like.
>> Awesome. So I'll be acting as the
principal buyer. Um, and just to start
off with, um, I've created a brand agent
already in our system. And so I've named
it Ematini. Um, it's a non-alcoholic
brand. Um, and I could be doing this
through our, you know, scope 3 buying
agent portal. But as we're here today to
talk about ADC CP protocol, I wanted to
do it through a, you know, generic
interface, which is why I'm using cloud
desktop today. Um, but we are still
using the underlying scope 3 API. And so
you'll see that I was able to set up
these really cool standards and stories
which help add more targeting and uh,
information to my brand agent. um like
you know who I'm targeting in terms of
no one underage, people in their mid20s,
late 40s. Um and so that is going to
really help us when we go and create the
media buy later on. And then the other
thing that I've set up for us in advance
of us creating the media buy is
obviously our campaign on the scope 3
side. So I've done that with a budget of
$20,000 and we want this to raise
awareness for my brand. So before we you
know dig into actually creating the
media buy Joe do you want to share
what's going on in the swivel side right now?
now?
>> Sure absolutely thank you. So now we're
over here in the swivel seller agent. So
right now we haven't received any
proposals. So we we have blank
proposals. Uh but inevitably this
proposal will be submitted from the
buyer agent scope 3. We'll receive it in
this little holding pin. Um, we're going
to return ad products. The way the
workflow works, we receive the media
brief. The media brief includes details
about the objective of the campaign, uh,
who we want to target, outcomes we're
trying to achieve, and then we suggest
ad products. These are ad products
powered by LG. See things like LG
channels, LG content network, etc. Um so
um after we receive the proposal and we
approve it uh we'll then we'll then
navigate to LG's downstream ad serving
platform Springerve. Uh today this at
this moment this area is empty but we
will see as the objects are created from
the proposed media buy into the ad
serving platform. So,
next up we
>> Let's see whether or not my campaign
likes any of your products.
>> Let's do that.
>> Okay. So, I'm going to discover the
applicable products for my campaign. Um,
and this is going to actually talk to
the sales agents I have registered
within scope 3. So, in this case, it's
going to reach out to Swivel and say
what products you have available. And
what should happen is the products you
just saw configured within Swivel should
appear here. um if they're applicable to
my campaign.
Okay, awesome. It did find the two
guaranteed video products. Um and this
is so important because it means that
when we're creating these campaigns,
we're creating um our media buys. We're
really able to actually target inventory
that's out there that we want to run on.
Um yeah, Joe, anything else you want to
say about these products? I I think ad
product discovery is a a net new
innovation for media buying in general.
I think out of all the inventory in the
world, if if you think about kind of how
programmatic is transacted today, you
have a bid stream with a neverending
array of um you know bids to choose
from. Um in an agentic buy, there's much
more kind of curation. the the
probability of being successful in an
agentic media buy is exponentially
higher than if you created a
programmatic campaign and you just spent
some budget on some random requests. Um
I think that's part of the beauty of the
agents is that they're working to
fulfill the outcome of the advertiser.
Um and they're doing that in an
orchestration by both combining
intelligence on the buy side and the
sell side.
>> Yeah, it's awesome. Okay, so now we've
create found out what products are
available for us. Let's actually go
create our media buy. So, I um didn't
want to type this, so I've got it copy
and pasteed. Um but we're going to
create a new buy on my um awareness
campaign. So, I remember I've got my
campaign up here. Um we're specifically
going to target the LG content network.
Um so, I wanted video content. This is,
you know, a guaranteed product format of
video. So, this is perfect for me. So,
I'm going to create a new buy. gonna do
a 10k budget, $20 CPM, and I want even
pacing. Um, and then the other thing
that I'm going to do at the same time is
I'm gonna create a tactic. Um, and a
tactic on our side is really allowing us
to create more specific targeting from
our campaign. So, we've got our
campaign. That's like generally what we
want to do. And then our tactic is going
to allow us to have specific targeting
like us, premium CTV, bring in all that
greatness in terms of our stories and
standards. So, we've got our tactic.
We've created our media buy. And this is
just in draft, so I could look at it
before I actually execute it against
Swivel. So, now I'm going to go and
execute uh this media buy against uh
Swivel to get approval. And so, awesome.
It's been submitted. Joe, why don't we
see what's going on in Swivel. All
right. So, we are back in
the seller agent UI. Uh don't see any
proposals. Let's refresh the screen. Ha,
we have a proposal. So, you can see we
have a buyer reference here. We have a
promoted offering. If I navigate here to
uh this proposal, you can see the brief
that was submitted by the buyer agent.
Um, and then importantly, you can see
the opportunity to approve the proposal
um in the bottom right. So, let's go
ahead and approve this media buy. Awesome.
>> So, approval sets into motion a series
of agents on the swivel side. Um, most
of those agents um are trapping agents
specifically for the purpose of creating
the downstream objects that will power
the media by. So the expectation here is
that inside the LG ad server spring
serve where we saw no objects created
earlier um we should begin to populate
um the the hierarchy that would execute
the media buy. Uh so depending on what
ad server that is you know that includes
things like campaigns line items um and
inevitably creative. Uh so I'm now going
to navigate back to the spring serve
instance. Uh this is LG Spring serve
instance. Right now it's empty. Let's
give it a refresh. And
And >> awesome.
>> awesome.
>> Voila. We uh we have objects. Um here is
the campaign. Um underneath the
campaign, it appears as though we have
line items. Um and on top of that, the
line items include the targeting. Um and
importantly, it includes the association
to the correct supply or the correct
inventory um to kind of close the loop
on the trafficking. Um so now that this
element is complete and the objects are
made in the downstream ad platform, you
know, theoretically we're we're in an
asynchronous process. The media buy has
been created yet we do not have
creative. Um frequently you see media
buys are executed and creatives are sent
after the media buy is complete. Um so
now we're going to fast forward towards
our start date um when we receive
creative from the agent buyer.
>> Cool. So let's go back into scope 3. Um,
and first thing I'm going to do is I'm
just going to get it. Um, so you can see
here before is that it was pending. Joe
went and approved it. And so now in our
system, we should have gotten that back.
And we can see that it was approved and
activated. So this is that async
interaction that's happening. Okay. So
let's take its advice. Let's add some
creatives to our media buys. So the
first thing that I'm going to do is I'm
going to create a creative. So in this
case, it's just going to be a very
simple creative. It's an MP4 with just a
media URL. But of course, you can do so
much more with creatives. Um Joe, you
want to give us some examples?
>> Sure. I mean, I think one of the
advantages of an agent agent transaction
is that you're not limited to what is or
is not supported. Um in in a bidstream,
you can execute on things like home
screens or sponsorships or pause ads. Uh
because everything's trafficked into the
downstream ad server. Um this the seller
agent knows what to do with the the
creative files. that knows what to tell
the buyer agent in order to uh correctly
execute and uh many many different
things that have never been available
commercially to uh a buyer without using
kind of a direct sales technique will
will now be enabled.
>> Yeah. And one thing to also keep in mind
is, you know, I wrote in this prompt to
create the creative, but in future this
might not be a human. This might be a,
you know, agent that's actually going to
go and create the creative because the
next thing that we do is we need to
assign it to our media buy. So, we've
assigned it and then we've gone and
updated the media buy that we created
before um and we showed you in swivel.
Um, and at this point, right, it still
then has to get um actually approved um
through the creative process. Um, but as
you can see, we've created the creative.
So, now let's uh jump over to Swivel
with Joe um to see what happened.
>> We're back to Swivel. Look what we have
in our creatives tab. We have some
immatini video creative. >> Um
>> Um
a human in the loop will approve this
video creative. Um I think human in the
loop is kind of a critical thing. I
think I think everybody that's watching
probably wants to know what the humans
continue to do. I think you can see
that, you know, prompting is still a
skill. Approval is still a skill. Um,
but more importantly, it's it's
important that um, especially for things
like premium video or things like a home
screen execution that that someone looks
at and approves a creative. Um, I'm now
navigating back to the LG Spring Server
instance. Um, and what I'm looking for now,
now,
um, is a creative in the creatives tab.
And there it is. The swivel agentic
crowd matini video creative is present.
Um, and thus completes a successful
agentic media buy. Um,
>> amazing. So, we did that from end to end
with our agents and let's go make some
money with this now. So, just to recap,
my agent working behalf of our buyer,
our brand was able to communicate with
the Swivel sales agent. He was able to
figure out what products inventory was
out there. He was able to go and create
that media buy and ultimately create and
add creatives to it. And the the seller
agent, you know, today is is
representative of one seller, but you
can imagine a future where an individual
seller could have hundreds, thousands,
millions of ad products. Um, as well as,
you know, a full array of different
sellers to choose from. Um, and this the
Sweller agent will will select ad
products based off the the media brief
such that even in an infinite world of
ad products, we're selecting ones that
have the highest probability of
producing an outcome. Um, so today, uh,
small scale, but you can see how in the
future there's a a scalable architecture here.
here.
>> Yeah, this is so cool. So cool to
actually do the first, um, Aentic media
buy with you. Um, so really awesome.
>> All right, Emma, really appreciate this
time. Look forward to many more of these
in the future. >> Fantastic.
>> Fantastic.
>> Thank you, Emma and Joseph. That was
incredible. And finally, please welcome
Kyle Dozeman, founding member of ADC CP
and CRO of PubMatic to round out the
launch with Next Steps before we're
joined by a special guest.
>> What you've seen today isn't
theoretical. It's not a vision for 2027
or 2030. This is happening now. Those of
us who have been in the industry long
enough have lived through the major
platform shifts from print to digital,
desktop to mobile, manual buying to
programmatic. But what makes this moment
different is the speed and the stakes.
If we look to history for context, the
world has seen about 20 technological
waves in the last 10,000 years. So a
wave is a set of technologies coming
together with profound societal
implications. Those that enable seismic
advances in what human beings can do. So
whether it was the combustion engine or
agriculture or the internet, these waves
transform everything. But the speed of
these waves is also accelerating. The
first seven waves occurred over 10,000
years. The next seven only took about
200 years and the most recent seven less
than a hundred years. The speed at which
AI will reshape our entire industry will
surprise all of us. Agentic advertising
isn't coming to disrupt us. It's coming
to empower us, but only if we build it
right and only if we build it together.
So what we're announcing today, ADCP and
this independent industry group is our
collective answer to one simple
question. Can we build the next phase of
the open internet better from the start?
Or will we make the same mistakes that
we made in the past? I believe that we
can get it right. And here's why I'm
confident. Because you are here. Over a
thousand of you said yes to this moment.
You recognize that being in the room
where standards are being set is more
valuable than watching from the
sidelines as proprietary walls go up
around you. So let's be direct about
what's at stake. So without a protocol
like ADCP, we're headed towards a world
where every platform builds its own
agentic solution. Every integration
requires custom development, the walled
gardens get higher, toll booths continue
to proliferate, and costs rise, and we
once again lose transparency and
effective working media dollars to drive
growth. or and this is the path that
we're choosing today. We create an open
standard where innovation accelerates
because everyone can plug in.
Competition thrives and the barriers
come down. Advertisers and publishers
can connect directly with true trust and
transparency and the entire ecosystem
grows stronger together. This is the
founding moment. This is where we decide
which future we're building. So, here's
what happens next. And there's a place
for everyone in this movement,
regardless of where you sit in the
ecosystem. First, join the community.
So, if you're a practitioner, a
strategist, someone who just wants to
understand how Agentic Advertising will
change your day-to-day work, join our
community. We have Agentic Advertising
meetups already scheduled. These are
where we'll roll up our sleeves and talk
real use cases, real challenges, and
opportunities. Join the LinkedIn
community. That's where we'll have
ongoing discussions and updates. This is
where you can stay ahead of the curve.
The link is live. Make sure you sign up
before you leave today. Second, start
building. So if you're a developer, an
engineer, a technical team, ready to
implement, it's time to build.
Publishers, the open source code is
available now. Access it directly or
contact us for implementation support.
Let's get you up and running as quickly
as possible. And for any team building a
buying agent or a platform, join the
ADCP Slack channel. Documentation,
testing tools, and a community of
builders are waiting for you. This is
how we prove the protocol works by
putting it into production together. And
third, let's shape the future through
governance. So if you want to influence
the direction of this protocol, the
standards that we're setting, the
priorities that we are pursuing, get
involved in governance. We're forming a
nonprofit entity to govern ADCP
industrywide. This ensures no single
company controls the protocol and that
it belongs to all of us. If you want a
seat at that table, just reach out. This
consortium is only as strong as the
diverse voices that help shape it. And
so let me leave you with this. Today we
have a major opportunity in front of us.
The foundational models are ready. The
infrastructure is built. And the market
momentum is undeniable. And most
importantly, we are ready. This room,
this community, this coalition. So what
happens next isn't up to any single
company. It's up to all of us. To the
publishers. You have a unique
opportunity with inventory and audience
insights that brands and their agents
need. Open your doors through AdCP. To
advertisers and agencies, push your
technology partners to adopt open
standards and get building and testing
your brand agents to be out in front. To
the builders and developers, ship code,
test protocols, break it, improve it,
make it stronger. And to everyone in
this room, tell your networks. This is
only effective if the industry moves
together. Share what you learned today.
Invite your colleagues to the next
meetup and help make this movement
contagious. The window for defining
standards is narrow, but it is open
right now. Let's not look back in 5
years and wished we'd acted. Let's look
back and say we were there. We built
this and we got it right. So, thank you
for being here, for building this future
with us. And now, let's get to work.
Before we close, all the links that you
need will be in a follow-up email that
hits your inboxes within the next hour.
the community sign up, the Slack
channel, the documentation, the
governance information, it's all there.
So, don't wait. Dig in and get started.
So, thank you again. This is just the beginning.
beginning.
>> Thanks, Kyle, for underscoring the next
steps for how to get involved. Before we
go, please welcome a hot take from the
industry's one and only Terry Kawaja.
>> AI and advertising disruption of
tectonic proportions. Cuz this is the
big one, folks. AI's effects will be
broad and deep. It will alter business
models and change the ecosystem at its
core. And this coming from someone not
afraid to call balls and strikes on
trends, unlike other shiny new object
trends that undisiplined observers
willingly went along with despite zero
evidence that they were nothing but a
steaming pile of horseshit. So AI has
been on Lumis agenda for over a decade,
calling it one of the top trends in
advertising in 2016. And look, we've
advised on all major developments in
adtech, take data, mobile app, and CTV
as examples, but now doing a lot of work
in AI. And while we've been able to
garner above market outcomes, high to
singledigit multiples of net revenue for
those sectors, AI is bigger, greater
valuations, and greater still. There's
only one problem with this green box.
It's not big enough. We're going to be
doing AI AI deals for quite some time.
Um, we like to think of Maslo's
hierarchy as a construct to take a look
at AI's application to advertising. And
as you go up the triangle, just like in
Maslo's original, uh, it gets, uh, the
disruption and the impact are greater.
So, at the base level, you've got
workflow, audience measurement, and
media buying, the meat and potatoes of
adtech. Next up is content, ad creative,
and communication. uh further up still
the big daddy internet navigation and
finally who knows powerful technology we
don't know what it can do. So to take
them one at a time at this core layer
essentially you know this right ad tech
is a largely manual largely sequential
largely linear sort of process from
planning targeting buying optimization
reporting uh and measurement and what AI
brings to the table is the ability to
bring all those together and do them in
real time and if you were noticing with
a lot less heads how many more heads I
don't know we'll see uh next up is
content, ad creative, and communication.
Here, I think of it like an iceberg.
What you see above the surface are the
no-brainer reduced costs of asset
production in the name of efficiency.
And that's all well and good. Everyone
should do that. But I think what moves
the needle is what is below the water
surface around data to determine which
ads uh creatives will work for driving
effectiveness. Think of the top as
tactical, the bottom strategic.
And we've had this 25-y year construct
of navigation where we try and find
people and potential customers and drive
them to our website to convert them.
Hell no. Go to where the people are.
Take your message to them. The benefits
are you go to them. It's a one-on-one
conversation. It's a shorter path to
conversion. All right. Uh internet
navigation. This is the big one. This is
search blue links going to an answers
economy. And what's driving this is the
more capable LLMs, the unprecedented
consumer adoption, emergence of new AI
interfaces uh from chat to assistance to
uh uh coming soon uh browsers and now a
paniply a spectrum of monetization uh
options ranging from crude data
licensing all the way through to
contextual ads and more that we don't
even know about. And a lot of companies
doing this, large companies, small
companies. We welcome Microsoft who just
announced they're getting in this game
as well. You can think of these as tech
traction services and models. uh and
based on the amount of dollars, the
massive amount of dollars invested uh in
LLMs to date and the necessary required
return, there's zero chance they're
going to derive all of that return from
a subscription business considering less
than 10% take up today, which is all
roads lead to advertising. So whether
it's new entrance getting into media or
LLMs, advertising will be it. Uh and we
don't know by the top, that is the
ultimate unknown unknown. Um, AI, if you
think about it, is like search intent
data on steroids, right? You type in
Cleveland divorce lawyer, we know three
things about you. You live in Cleveland,
you're getting a divorce, and you need a
lawyer, right? Think about that in an AI
context. Uh, the user starts with on
average a 25word prompt and has multiple
interactions. The reality is it's a
higher intent signal. Um, Zuckerberg
even thinks this is going to change the
ratio, the perspective of uh advertising
spent to global GDP. We'll see. So, one
thing is clear. Doing nothing is not a
viable strategy and but yet there are
significant consequences from corporate
development to the ecosystem revolution.
Let's try and get through these.
Discerning true ad tech is challenging,
right? We characterize these companies
as you're either AI first, a fast
adapter, or you're an AI poser. I know
it's a harsh term, but let's be real.
Uh, and based on the feedback I got from
my LinkedIn post, a lot of people are
focused on this. We all know that
outcomes are better than proxies and
we're living in the outcomes world.
We've been talking about this phenomena
for over a decade of how uh scale and
automation can bring uh operating
leverage and look we think of it as a
continued gestation as we said at the
ANA in 2018 that started with gaming and
is moving to all uh marketers. It's you
could think of gaming as canary in the
coal mine here. Historically performance
defaulted to deterministic measurement
and we know all the problems of last
click the perverse incentives the
shenanigans the lack of scale of
firstparty data airgo the probabilistic
world being better although complex slow
and expensive. So you bring AI into the
equation and all of a sudden you can
turn that probabilistic into predictive
measurement buying and ultimately
outcomes like the big guys do. We
believe that supply side decisioning is
the next evolution for optimization.
what started as waterfall uh and then
header bidding and then we had SPO and
curation and we believe supply side
decisioning is the next phase that
leverages AI to not take take a a view
of the entire spectrum of inventory not
just the signal coming from a QPS uh
throttled uh demand side view and we
believe this will help the open internet
compete against the walled gardens or as
in the words of my friend Snoop Dogg
outcomes get high with your own supply.
>> There you have it. Uh by the way, uh uh
uh yes, what do I mean by compete with
the wall gardens? Well, you know, we
know traffic is moving in the wrong
direction towards wall gardens. And how
do we preserve that in the open web and
the uh hedge gardens? Well, let's take a
look at the attributes of the wall
gardens. Here are the five. And we
believe that with uh uh third-party
inventory and supply side decisioning
and maybe a little bit of M&A in there,
we can replicate most of those
attributes and replicate those results.
Don't believe me? Apploven's already
done it. They have executed both organic
and inorganic strategies that have
allowed them to replicate all five of
these attributes. How do we know this in
such detail? We advise them. How did it
turn out by the way? Uh unbelievable.
Since launching Axon 2.0 companies on a
76% compounded growth rate on revenues
combine that with uh their 80% EBIT
margins, it's a rule of 150 plus company
with a market capitalization of $235
billion which is greater than every
other ade competitor that's public and
every single advertising holding company
period. full stop with enough leftover
to buy a few, you know, fun companies
like Pinterest and Sirius and Snap and
the New York Times and depending on the
day, Warner Brothers. We view this as
the best M&A case study ever. The other
thing that AI does is it reframes what
performance means. Connecting brand to
performance. Probabilistic AI connects
these two worlds so that you end up with
something combination of the two. We'll
call it brand performance that measures
impact on outcomes and assesses
predictive KPIs. The implication being
it allows you to buy on guarantees and
it shifts ad spend uh from a
discretionary expense to a cost of goods
sold. And the other impact is that it
effectively shortens the path to
purchase. Think about it. And when you
can show a tight connection between
spend X and revenue Y with LTV to CACM
math, you change the nature of
advertising from a discretionary
expansion up the income statement to a
cost of goods sold. And then there's no
off button on campaigns. This is why
Adam Fero, founder of Apploving, is
>> Again, canary in the coal mine. He saw
it first. AI is a gamecher for
advertising. I don't know about you, but
I'm sick and tired of this incremental
approach we have. Our product is 10%
better. No, I want X. I want our product
is 10x better. And when you think about
it, right, every order of magnitude
change in advertising has come from
native formats on native infrastructure.
Think Google Search, Facebook, Tik Tok,
AppLov, and coming soon, Open AI. That's
right. They've already announced they're
building out infrastructure, which begs
the question, will AI catalyze a long
overdue great reckoning in adtech? We
would submit that AI will separate the
wheat from the chaff and drive much
needed rationalization and
consolidation. The Lumaccape, after all,
is a fragmented ecosystem of largely
undifferiated companies built in the
Zerp era, which ek out their existence
on revshares and kickbacks. Harsh, but
true. The result is an inefficient
supply chain with economic opacity,
perverse incentives, shenanigans, and
outright fraud. What's needed are fewer
players doing more volume at lower take
rates with better quality. In other
words, like every other industry. This
one just refuses to grow up.
>> Some advocate for a revolution that
blows up the existing programmatic
ecosystem and starts over.
Yes, comrade O. Kelly driving charge
with ubiquitous post on LinkedIn about
how AI is going to reinvent supply
chain. Okay, enough of that. Point is,
he's not wrong. Uh we are due for a
great reckoning and AI may be the
catalyst that pushes us over the edge.
By the way, by the way, given the
ubiquitous ofveness of his posts, one
wonders what kind of automation
technology he's utilizing. ah explains
everything. And if you think about it,
we are entering what we call the fifth
phase of adtech, right? We had the yes
digital ad spend has grown from 13
billion to half a trillion over the
course of these 25 years. And Luma was
launched in 2010 to take advantage of
this very opportunity that we saw
coming. But it hasn't been a smooth
ride. Right? The first 10 years we had
the ad network era and what I call the
early programmatic era. We started with
contextual, we moved to audience. We
started with bootstrap companies, we
moved to venture financed. Then we had
the next 10 years, we hit strong
headwinds from wall gardens and a
variety of issues in the ecosystem. We
actually reached peak lumiscape uh uh
fragmentation and then we entered the
scale and PE era while you know CTV and
mobile app and commerce media continued
to grow but we had the cookie circle
jerk and we had all kinds of shenanigans
going on in the marketplaces. Boom and
bust. The key point being innovation
waned and here we are. Thank God finally
we are in phase five we're calling AI
reinvention to see a rapid innovation
cycle lot of disruption outcomes will
accelerate that and of course Google's
on their back feet because of antitrust.
We believe this will bring a muchneeded
return to strategic M&A imperatives and
finally sector consolidation or as my
partner Connor said to me the other day mortimer
mortimer
calls me a mortimer. So uh you get the
idea and hopefully you're on board for
this as well. But the AIdriven future
holds much promise of an exciting new
reality with emphasis on the word new.
Think about it. $500 billion market cap
of a private company makes them the most
valuable. They've got a new consumer
interface, new product, new pipes, new
value proposition, new business model,
and a new ecosystem. At the end of the
day though, we think this is a win for
consumers. We think this is a no-brainer
win for advertisers. agencies could go
either way depending on whether you move
fast or not, organically or
inorganically. Same thing with the ad
tech ecosystem. The winners will be the
ones that move fast. When it comes to
publishers, I think the large ones will
do fine uh and the small ones and the
news ones, I'm worried about them. Uh we
have to find a solution uh for them. So,
what are my key takeaways? Changes on
hyperdrive rethink everything in this
environment. Legacy assumptions, they're
out the window. Trust but verify has a
whole new meaning. Separate the wheat
from the chaff. Finally, thank God with
a better business model for the
surviving intermediaries. And
ultimately, we believe inorganic growth
key to winning. Now, I know what you're thinking.
thinking.
>> I'm shocked. Shocked to see Luma
pitching M&A is a strategic solution.
>> You're winning, sir.
>> Oh, thank you very much.
>> Thank you all and good luck on this very
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
