0:02 Hello and thank you for joining us for
0:04 the launch of a new industry protocol,
0:06 ADC CP. Over the next hour, you'll hear
0:08 from the pioneers building the future
0:11 advertising as Aentic AI transforms the
0:14 ecosystem. We will dive into the
0:17 technical architecture of ADCP, see live
0:19 demos of the protocol in action, and
0:21 discuss the transformative impact of
0:24 this technology. Let's get started.
0:25 Please join me in welcoming Amanda
0:28 Dvito, chief marketing officer of Butler
0:31 Till. Hello everyone and welcome. We are
0:34 standing at one of those rare inflection
0:37 points, the kind you only recognize in
0:40 hindsight. Except this time we know it's
0:43 happening right now. Advertising is
0:46 entering a new phase. Not next year, not
0:49 5 years from now, today. For decades,
0:51 our industry has evolved through big
0:54 defining shifts from print to digital,
0:56 from desktop to mobile, from manual
0:58 buying to programmatic. But what's
1:01 unfolding in front of us now is bigger
1:04 than any of those transitions. Because
1:06 we are stepping into something
1:09 fundamentally new, the agentic era of
1:11 advertising. I'm Amanda Dvito, chief
1:14 marketing officer at Butler Till, where
1:16 I lead marketing and growth. We're an
1:18 independent, woman-owned and led,
1:21 employeeowned benefit corporation media
1:23 and communications agency. And we're
1:26 here because our clients count on us to
1:28 see what's next before it's obvious and
1:31 to help them turn change into advantage.
1:33 For us, this isn't about chasing the
1:36 next shiny object in adtech. It's about
1:38 making sure the buy side, advertisers,
1:41 planners, and strategists have a real
1:43 seat at the table as this transformation
1:46 unfolds. We're here because this
1:48 evolution toward agentic advertising has
1:50 the power to create more meaningful
1:53 work, more transparency, and more value
1:56 than ever before. Not just for brands,
1:58 but for the people behind them. That's
2:00 why we're involved. Because our clients
2:03 deserve a world where innovation
2:05 actually works for them, not around
2:08 them. We find ourselves at the perfect
2:10 storm of timing, the intersection of
2:14 capabilities and momentum. On one side,
2:16 there's the explosion of AI capability.
2:19 Foundation models like Claude, ChatgBT,
2:21 and others have reached a level of
2:23 reasoning and planning that makes true
2:25 agentic behavior possible. Agents that
2:28 can think, plan, and act autonomously.
2:31 And with Anthropic's launch of the model
2:33 context protocol late last year, the
2:36 technical rails are finally here. The
2:38 infrastructure that allows agents to
2:40 communicate directly with systems,
2:42 platforms, and each other. That's the
2:45 capability part. On the other side, the
2:49 momentum undeniable. Advertising itself
2:52 is evolving in real time. What started
2:55 as agents who co-pilot with us on
2:57 creative and personalization has now
3:00 reached the next frontier, media buying
3:03 itself. And here's what's so exciting.
3:06 This isn't just about automation for
3:08 automation's sake. It's about unlocking
3:11 focus. Agentic systems can handle the
3:14 optimization and orchestration that
3:16 consume so much of our time today,
3:19 freeing humans to spend more time on
3:22 what really moves the business forward.
3:25 Insight, creativity, and strategy.
3:27 That's not the next shiny thing. That's
3:30 a shift towards smarter work and better
3:33 outcomes. From Google exploring MCP for
3:36 its ads API to Amazon launching Aentic
3:38 creative tools, major platforms are
3:40 already moving fast. This isn't
3:44 theoretical. It is happening right now.
3:47 And that's what makes this moment both
3:49 so incredibly pivotal and exciting.
3:51 We're not just observers. We're
3:54 architects of what comes next. Let's be
3:56 honest, the opportunity is massive. I
3:59 know you're all feeling it. But so is
4:02 the responsibility to get it right.
4:04 Today the advertising ecosystem is
4:07 deeply fragmented. A plethora of tools,
4:09 platforms, and systems that barely
4:11 communicate with each other. They create
4:13 complexity, inefficiency, and
4:16 duplication. Every AI integration is
4:18 bespoke and customuilt. We're building
4:20 in silos when what we need is
4:23 interoperability. The window is open
4:25 right now for the industry to come
4:27 together and define shared standards
4:30 before proprietary walls go up again.
4:32 This is that rare moment when we can
4:35 choose collaboration over competition.
4:37 When we could decide to build a system
4:40 that works for all, all of us, not just
4:42 a select few. And that's what today
4:45 represents. You are here at the founding
4:48 moment of something new. a coalition of
4:51 industry leaders, innovators, and
4:54 builders all coming together to launch a
4:56 shared framework for agentic
4:58 advertising. It's being built by the
5:01 industry for the industry, open,
5:04 interoperable, built to evolve. Over the
5:07 next hour, you'll see what that looks
5:10 like. live demos, real use cases, and
5:11 the beginnings of a model that can
5:14 transform how we work across every part
5:16 of the ecosystem from yield and
5:18 measurement to planning and buying. This
5:22 is not about one company solution or one
5:24 side of the marketplace. It's about
5:28 collective progress because the future
5:30 we want will only exist if we're going
5:33 to build it together. The question isn't
5:35 whether aic advertising is coming. It's
5:37 already here. The real question, the
5:40 real question, the one that will define
5:43 this next decade is whether it unfolds
5:45 through fragmented proprietary systems
5:48 or through open interoperable standards
5:51 that create a transparent, trustworthy,
5:54 thriving ecosystem. This moment, this
5:57 event is about choosing the latter. So
5:59 good for you for being here today. So
6:02 consider today your front row seat, not
6:04 to watch the future unfold, but to help
6:07 build it. Because the agentic era isn't
6:09 something happening to us, it's
6:12 something happening through us. And it
6:15 starts now.
6:18 Welcome to the start of the agentic era.
6:21 Now, please welcome Anne Cochland, COO
6:23 and co-founder of Scope 3 to tell us
6:25 more about the agentic era of
6:28 advertising, followed by Bashko Milikch,
6:30 chief product officer at Optal for a
6:33 deep dive into the protocol itself. Wow,
6:35 I'm excited and we're just getting
6:37 started. Hi everyone, I'm Anne Cochland
6:39 and I'm the COO and co-founder of Scope
6:41 3. For us to really understand the power
6:44 of aic advertising, it's helpful for us
6:46 to start with what marketers really
6:48 want. So, let me paint that picture for
6:49 you first and then we can walk through
6:51 why the three existing paradigms of
6:54 digital advertising are not helping a
6:56 marketer achieve those needs. So, here's
6:58 a list of what marketers want.
6:59 Fundamentally, the first one on the top
7:01 of the list here is scale. They want to
7:02 be able to buy media everywhere that
7:04 makes sense for their campaign. But I
7:06 would add that they want to do that with
7:08 no marginal cost. So, no additional
7:10 money or time or focus just because
7:11 they're adding another channel or a
7:13 publisher. They want creative and media
7:15 to be really really connected. Not
7:17 bidding on ad space and then separately
7:18 choosing creative but generating the
7:21 exact creative for each spot, each
7:22 moment, each audience. And so that kind
7:24 of probably what ties in there is the
7:27 idea of being native. No lowest common
7:30 denominator. Everything feels like it
7:32 makes sense to be together. They want
7:34 outcomes, not just impressions. So real
7:36 business results, not proxy metrics. And
7:37 continued rhetoric about what that means
7:39 within the industry. And then these kind
7:41 of all tie together a little bit. uh but
7:43 you know thinking about the quality of
7:45 execution how creative and media being
7:46 together native works making sure that
7:48 you're driving to the right outcomes and
7:50 scale also I think it's really important
7:52 to make sure that there is no difference
7:55 between a10,000 buy or $100,000 buy or a
7:57 million dollar buy so everything the
7:59 quality of execution the ability to
8:00 execute should be the same and then of
8:02 course they want sustainability
8:04 transparency and working media that
8:05 actually works for them so that is the
8:07 vision that is what digital advertising
8:10 should deliver so why hasn't it so let's
8:11 start with the insertion order model. So
8:13 with the direct IO as a brand, you
8:15 definitely get the full power of the
8:18 underlying media. You can do really
8:20 truly custom campaigns with rich bespoke
8:23 executions, custom formats, native
8:24 integrations, content partnerships.
8:26 There's a real richness and diversity in
8:28 what publishers have to offer. This
8:30 obviously leads to better outcomes. Um
8:32 there's a real direct connection here.
8:34 So you you're getting the sustainability
8:35 benefits, the transparency, and the
8:37 working media, but it really doesn't
8:39 scale. Every single deal requires human
8:41 negotiation. There's a sales team with
8:43 finite resources. It's powerful, but
8:46 fundamentally it's oh that's pink. Let's
8:50 go for red. It is very limited by human
8:51 bandwidth. Uh the publisher can only
8:53 sell as much as their sales team can
8:56 handle. The second paradigm trying to
8:58 fix the big red circle on the previous
9:00 diagram. For a publisher to be able to
9:02 offer an ads API, they do need to have
9:04 the technical infrastructure to expose
9:06 their inventory targeting capabilities
9:08 and be able to give campaign management
9:10 capabilities to the marketer. And we've
9:12 seen that, you know, that can really
9:13 demonstraably drive scale in the case of
9:15 Meta and Google. But the problem is the
9:17 reason why I have to unfortunately mark
9:19 this red still is that APIs are custom.
9:21 Buyers can't integrate with many APIs.
9:23 So they choose and prioritize Google and
9:25 Meta because those are where they see
9:26 the biggest reach and the maximum
9:28 performance drivers. and then buyers
9:30 just stop. So the engineering return on
9:32 investment isn't there. It's not enough
9:33 to be able to integrate with everyone
9:35 else and that is a big problem for the
9:36 publisher. The third paradigm is
9:38 programmatic and I finally been able to
9:40 color this circle in green here because
9:42 we do have automation at scale. The
9:44 technical ecosystem that we've been able
9:45 to build through standardization through
9:48 different protocols means that we can
9:50 connect in theory any buyer to any
9:52 publisher. But fundamentally what we
9:54 should be doing as an ecosystem is
9:55 trying to translate the brief from the
9:57 marketer here and make it work so that
9:59 the marketer can buy inventory that
10:01 really fits their campaign brief in a
10:03 way that the publisher gets paid fairly.
10:05 And there are lots of things that break
10:07 down in terms of these circles here.
10:09 First of all, this brief is rich. It's
10:11 detailed. It's strategic. It needs to be
10:13 compressed and simplified to fit through
10:15 these plates. And then on the publisher
10:17 side, all of that first party data
10:19 richness, all the unique capabilities
10:21 are not accessible up through the
10:22 programmatic stack in the way that they
10:24 are through direct IO buying. So while
10:27 the DSP here that I've colored in uh
10:29 pink is the focus point for decisioning,
10:31 it's making decisions without access to
10:32 the best information. Buyers have to
10:34 spend significant money on third party
10:36 data because they can't access that
10:38 first party richness that actually
10:40 exists on the publisher side. So working
10:42 media is around 41%. I'm going to mark
10:44 this one red here. And I'm going to mark
10:48 all of these an orangey color. You can't
10:49 do all of the fancy stuff. You can't
10:51 access unique inventory, the custom
10:52 capabilities that make publishers
10:54 valuable. The promise of programmatic
10:57 largely has gone unfilled. And my
10:59 concern is that if we just sprinkled AI
11:01 on top of each one of these existing
11:04 boxes without actually rethinking the
11:05 fundamental architecture, we just
11:07 perpetuate all of the same problems at
11:10 an even higher cost. So these are the
11:12 three existing paradigms. Fundamentally
11:14 what we're left with is that brands want
11:16 outcomes. They want advertising that
11:18 actually works across all the existing
11:19 capabilities and platforms. And we talk
11:21 a lot about the future of advertising,
11:24 influencers, AI, chat, all of this stuff
11:26 that didn't exist 10 years ago that are
11:28 not going to work within any of these
11:30 three paradigms because of the
11:32 challenges that I've outlined.
11:33 So, what if a publisher could build an
11:35 agent? Instead of auctioning every ad
11:37 and hoping someone bids, instead of
11:39 having to create a sales team that can't
11:41 work at scale, you can actually take
11:42 orders all the time and make the best
11:44 decision about which ad to show to each
11:46 person at the right moment, the full
11:47 vision of what we thought digital
11:49 advertising was going to be. And for a
11:52 buyer, this means that you get the full
11:53 scale of programmatic, but you don't
11:55 lose any of the richness of the
11:57 opportunity and of the connection that
11:59 you have with the buyer. So this is why
12:02 agentic advertising is powerful. One,
12:04 it's about helping publishers monetize
12:05 better. Leveling the playing field
12:07 against Meta and Google once a buyer
12:08 agent is set up. Adding another
12:10 publisher is just another conversation
12:12 between agents. No new engineering
12:14 integration. No additional sales
12:16 negotiations. It helps advertisers get
12:18 better performance and outcomes than
12:19 they can from wall gardens. It makes it
12:20 possible for anyone, whether you're a
12:22 massive holding company or a small
12:23 business, to get access to the full
12:25 breadth and power of digital
12:26 advertising. And the way that we've made
12:28 this possible is by building a protocol,
12:30 a way to connect buyers and sellers.
12:31 There are lots of people working on this
12:33 and I'm now going to turn this over to
12:34 one of the architects of this new
12:37 protocol, the ad context protocol to
12:38 show you and talk to you about how it works.
12:40 works.
12:42 >> Hi everyone, my name is Bosko Milik and
12:44 I'm a longtime adtech nerd and the chief
12:46 product officer and co-founder of
12:48 Optible. Optimal is an identity
12:49 management and data collaboration
12:51 platform and we're committed to the
12:53 development of open aentic advertising
12:54 standards and together with a growing
12:56 community. We've been collaborating on
12:58 the development of the ADCP protocols
13:00 which we're really excited to share with
13:02 you today. So I'd like to invite you to
13:04 join us and contribute to ADCP's design
13:07 and development. So what's ATCP? ADCP is
13:10 short for AD context protocol. It's an
13:12 open extensible set of protocols
13:14 designed to support intelligence
13:15 software agents in planning,
13:18 transacting, and optimizing advertising
13:21 campaigns across the digital media
13:23 ecosystem. It defines a standardized
13:26 interface for both agent to platform and
13:28 agent-to-agent communication. Now, why
13:30 do we need a protocol for agenic
13:32 advertising anyway? Well, the rise of
13:34 autonomous agents in advertising is no
13:36 longer theoretical. Advertising agents
13:38 are software that communicate with
13:41 people, large language models, data, and
13:44 ad platforms to get work done. And all
13:45 corners of the ecosystem are working
13:48 towards and adopting agents, making
13:50 better use of time, resources, and
13:52 ultimately driving better outcomes. Now,
13:55 agents not only know things, they can
13:57 now do things. Marketers are piloting
14:00 brand agents today and platforms are
14:02 integrating agent logic for a variety of
14:04 use cases in digital advertising
14:06 spanning from creative design to signal
14:08 building activation to workflow
14:11 automation. But we've got a problem. We
14:12 lack the common language to connect
14:14 these proprietary and siloed aentic
14:17 advertising systems. And this results in
14:20 a fragmented and inefficient ecosystem.
14:22 And so ADCP is that missing layer and
14:25 common language. It provides a means by
14:27 which buyers and sellers of advertising
14:30 can connect their agentic systems uh
14:32 which enables scaled execution of
14:34 relevant ad campaigns across an
14:37 inherently fragmented landscape. Now
14:39 ADCP is designed to allow advertisers or
14:41 agencies to connect to media companies
14:44 or publishers directly so that they can
14:47 run and optimize ad campaigns toward a
14:50 common goal. So who's AdCP for? Well,
14:53 for starters, advertisers and marketers.
14:56 AdCP creates more working media to drive
14:58 outcomes and intelligent software agents
15:00 can now plan and buy media on your
15:02 behalf, but they need a shared
15:04 infrastructure to operate across
15:06 channels. And so agents that are built
15:09 on ATCP can deploy, optimize, and adjust
15:12 campaigns continuously and operate
15:14 outside of proprietary silos. They can
15:16 tap into publisher signals to improve
15:19 performance. So if you're an advertiser,
15:21 you should evaluate Agentic buying use
15:24 cases and ask your media platforms and
15:26 partners if they plan to support ADCP.
15:28 Now for publishers and media owners,
15:30 ADCP creates more opportunities to
15:33 monetize your audiences. It defines a
15:36 standard way of representing inventory
15:38 pricing and packaging. So, if you're a
15:40 publisher, you should ensure your
15:41 inventory and data are represented
15:44 according to NCP standards and prepare
15:46 your monetization strategy for an
15:48 agent-driven marketplace. Consider
15:50 partnering with platforms that are
15:52 offering sales and signals agents and
15:55 integrations. Now, for agencies, ADCP
15:57 enables you to bring your proprietary
15:58 planning, data insights, and
16:01 partnerships to bear through customuilt
16:03 buyer agents. It establishes a standard
16:06 that ensures campaigns can be executed
16:09 by intelligent systems across publisher
16:11 platforms. Ultimately, it accelerates
16:14 campaign planning, activation, and
16:16 enables continuous optimization. So, if
16:18 you're an agency, you should consider
16:20 piloting some agent-driven use cases
16:23 with ADCP uh with select client accounts
16:25 to prove out the efficiency gains. And
16:28 finally for adtech platforms, ATCP
16:30 enables you to expose your capabilities
16:32 to externally agentic platforms through
16:35 a common standard. It keeps ade
16:37 platforms central by standardizing how
16:40 agents access inventory decisioning and
16:42 measurement. So if you're an ad
16:44 platform, you should consider exposing
16:46 more of your capabilities to external
16:48 agentic platforms through the common
16:51 standard and evaluate your platform's
16:54 compatibility with ATCP. Now let's
16:57 briefly look at how ADCP works today. So
16:59 ADCP today is made up of two
17:01 foundational protocols. The media buy
17:04 protocol is one and it's ADCP's core
17:06 advertising automation interface
17:08 providing standardized tasks for
17:10 managing the complete advertising life
17:13 cycle. So it specifies a set of tasks
17:15 enabling discovery and planning,
17:18 campaign execution, creative management
17:21 and optimization workflows. So a media
17:24 agent operated by a buyer can use the
17:26 media buy protocol to connect to
17:28 multiple sales agents enabling
17:31 end-to-end campaign execution. Now the
17:33 protocol is asynchronous by design. So
17:35 it's designed to enable a publisher to
17:38 require human approval at any stage of
17:40 the workflow. So for example, creatives
17:42 may require publisher approval before
17:45 being activated in their ad server.
17:47 The second protocol in ADCP is the
17:50 signals activation protocol. It enables
17:53 agents to discover, activate, and manage
17:56 data signals through natural language.
17:58 It's designed so that both buyers and
18:00 sellers can interact directly with their
18:02 signals agents in order to build and
18:05 target signals and inventory that's
18:08 relevant to the advertisers's brief. So
18:10 the media buy and signals activation
18:13 protocols are designed to work together
18:15 to enable intelligent signal and product
18:17 discovery and to enable direct campaign
18:20 execution between buyers and sellers.
18:23 The goal is to enable a transparent and
18:25 efficient marketplace where both buyers
18:28 and sellers work together to deliver to
18:30 the marketer's brief and desired
18:32 outcomes. Now let's look at some of the
18:35 design goals of ADCP. A key ADC CP
18:37 design goal is to make the protocol
18:40 compatible with both MCP and the
18:43 emerging A2A open protocols. Now these
18:45 protocols are asynchronous and taskbased
18:47 by design and this is because some
18:49 operations may take more time like those
18:51 where a human intervenes. So for example
18:53 when a seller requires creative approval
18:56 before campaign activation. Now the idea
18:59 is to have agents automate repetitive
19:01 and errorprone human tasks and add
19:03 intelligence to ad product and signal
19:06 selection. Basically we want to give ad
19:08 ops superpowers. And speaking of uh
19:10 human intervention, it's important that
19:12 publishers be able to require manual
19:15 approval before any operation. So for
19:17 example, creative auditing and pricing
19:19 approval are good examples. The current
19:22 media buy protocol codifies a generic ad
19:24 product and campaign structure that can
19:26 be made to work with sellside ad
19:28 decisioning systems like publisher ad
19:30 servers like GAM for example. It can
19:33 also plug into specialized ad servers
19:36 like Triton and others. Now, finally,
19:38 these protocols, they're designed for
19:41 both agenttoplatform and agent-to-agent
19:43 communication. And that implies the
19:46 ability to share prompts and context
19:48 between buyer and seller agents. The
19:50 mental model I like is that of agents
19:52 talking to each other in natural
19:54 language, collaborating as colleagues as
19:56 opposed to just making API calls or
19:58 clicking buttons in a UI. Now, here's a
20:00 picture of what multi-publisher campaign
20:03 execution looks like to a buyer media
20:05 agent from the standpoint of the AdC
20:08 protocol. So, Acme Corp is a buyer and
20:10 they have a buyer agent that chats with
20:12 the media company's sales agent to plan
20:14 and activate ad products that are
20:16 powered by relevant signals from both
20:19 Acme Corp and its media partner. Now, in
20:21 this example, campaign activation and
20:23 reporting is enabled using the
20:25 publishers ad server. Reporting data is
20:27 made available to Acme Corp through the
20:30 media buy protocol. Now in parallel,
20:33 Acme Corp's media agent also chats with
20:36 another media corporation sales agent to
20:38 plan and activate ad products backed by
20:40 different signals and inventory. So a
20:43 similar reporting feedback loop exists.
20:45 Now there might be differences in
20:47 format, signals, and inventory mix
20:49 naturally across publishers and the
20:51 buyer can refine her campaign plan and
20:54 execution along the way. Each party is
20:55 given the opportunity to create and
20:58 propose ad products and signals that
21:00 strive to optimally deliver the
21:02 marketer's intent. So where are we today
21:04 and where do we go from here? Well,
21:06 several founding group companies are
21:07 starting to test and roll out
21:10 activations using the ADCP media buy
21:12 protocol today. Many others have started
21:15 to look at the protocols and build buyer
21:16 sales and signals agents to
21:19 interconnect. The community is growing
21:21 quickly. Beyond growing adoption, we
21:23 want help from those interested in
21:25 refining, improving, and expanding the
21:27 current ATCP protocol specs and open
21:30 source projects. An initial technical
21:31 steering committee has been formed and
21:33 will help establish guidelines and
21:36 governance. And we have some dedicated
21:37 discussion channels for the various
21:40 protocols that make up ADCP and are
21:43 public and you can access them uh in the
21:46 Agentic Ads Slack community. Now,
21:47 finally, several community members are
21:49 also interested in contributing to the
21:52 open- source sales agent in GitHub. And
21:53 there's a lot of work to do on this
21:55 front to get to a productionready open
21:57 source sales agent. Uh we have a really
21:59 good start and we would like to invite
22:01 interested and capable companies and
22:03 individuals to help out with this
22:04 effort. Well, thank you for joining us
22:07 today and stick around to see a working
22:09 demo of ADCP in action.
22:11 >> So much work has already been done in
22:14 bringing ADCP to life. Now let's see the
22:15 protocol in action with leaders from
22:17 scope 3 and swivel.
22:20 >> So what we're going to see today is the
22:22 interplay between a variety of systems
22:25 as we go through an agentic media buy.
22:28 Um so you'll see us start um with an
22:31 agent powered by scope 3. Um the agent
22:33 will then communicate with the seller
22:36 agent on the swivel side. Uh and then
22:38 you will see finally as the agenda
22:40 transaction comes to a conclusion you
22:42 will see the objects made um in the
22:45 downstream ad serving platform. All this
22:48 orchestration is done by agents uh it's
22:51 a real demonstration um and it also
22:53 includes a real publisher uh our
22:56 publishing partner LG ads. Um you'll see
22:58 us demonstrate a human in the loop to
23:00 approve the media buy. Um and lastly
23:03 you'll see creatives traffic uh for the
23:06 end of the execution. Um, so we're
23:08 excited to show you what this what this
23:09 looks like.
23:11 >> Awesome. So I'll be acting as the
23:14 principal buyer. Um, and just to start
23:17 off with, um, I've created a brand agent
23:19 already in our system. And so I've named
23:23 it Ematini. Um, it's a non-alcoholic
23:26 brand. Um, and I could be doing this
23:28 through our, you know, scope 3 buying
23:29 agent portal. But as we're here today to
23:32 talk about ADC CP protocol, I wanted to
23:34 do it through a, you know, generic
23:36 interface, which is why I'm using cloud
23:38 desktop today. Um, but we are still
23:41 using the underlying scope 3 API. And so
23:42 you'll see that I was able to set up
23:45 these really cool standards and stories
23:48 which help add more targeting and uh,
23:51 information to my brand agent. um like
23:54 you know who I'm targeting in terms of
23:56 no one underage, people in their mid20s,
23:59 late 40s. Um and so that is going to
24:01 really help us when we go and create the
24:04 media buy later on. And then the other
24:06 thing that I've set up for us in advance
24:07 of us creating the media buy is
24:09 obviously our campaign on the scope 3
24:11 side. So I've done that with a budget of
24:13 $20,000 and we want this to raise
24:18 awareness for my brand. So before we you
24:20 know dig into actually creating the
24:22 media buy Joe do you want to share
24:23 what's going on in the swivel side right now?
24:25 now?
24:30 >> Sure absolutely thank you. So now we're
24:33 over here in the swivel seller agent. So
24:35 right now we haven't received any
24:36 proposals. So we we have blank
24:39 proposals. Uh but inevitably this
24:41 proposal will be submitted from the
24:43 buyer agent scope 3. We'll receive it in
24:46 this little holding pin. Um, we're going
24:48 to return ad products. The way the
24:50 workflow works, we receive the media
24:54 brief. The media brief includes details
24:56 about the objective of the campaign, uh,
24:58 who we want to target, outcomes we're
25:00 trying to achieve, and then we suggest
25:01 ad products. These are ad products
25:04 powered by LG. See things like LG
25:08 channels, LG content network, etc. Um so
25:11 um after we receive the proposal and we
25:13 approve it uh we'll then we'll then
25:16 navigate to LG's downstream ad serving
25:19 platform Springerve. Uh today this at
25:21 this moment this area is empty but we
25:24 will see as the objects are created from
25:26 the proposed media buy into the ad
25:29 serving platform. So,
25:32 next up we
25:34 >> Let's see whether or not my campaign
25:36 likes any of your products.
25:37 >> Let's do that.
25:40 >> Okay. So, I'm going to discover the
25:43 applicable products for my campaign. Um,
25:45 and this is going to actually talk to
25:46 the sales agents I have registered
25:48 within scope 3. So, in this case, it's
25:50 going to reach out to Swivel and say
25:52 what products you have available. And
25:54 what should happen is the products you
25:56 just saw configured within Swivel should
25:59 appear here. um if they're applicable to
26:01 my campaign.
26:04 Okay, awesome. It did find the two
26:07 guaranteed video products. Um and this
26:08 is so important because it means that
26:10 when we're creating these campaigns,
26:12 we're creating um our media buys. We're
26:14 really able to actually target inventory
26:17 that's out there that we want to run on.
26:20 Um yeah, Joe, anything else you want to
26:22 say about these products? I I think ad
26:24 product discovery is a a net new
26:27 innovation for media buying in general.
26:30 I think out of all the inventory in the
26:31 world, if if you think about kind of how
26:34 programmatic is transacted today, you
26:38 have a bid stream with a neverending
26:42 array of um you know bids to choose
26:45 from. Um in an agentic buy, there's much
26:47 more kind of curation. the the
26:48 probability of being successful in an
26:51 agentic media buy is exponentially
26:52 higher than if you created a
26:54 programmatic campaign and you just spent
26:57 some budget on some random requests. Um
26:59 I think that's part of the beauty of the
27:01 agents is that they're working to
27:03 fulfill the outcome of the advertiser.
27:05 Um and they're doing that in an
27:06 orchestration by both combining
27:08 intelligence on the buy side and the
27:09 sell side.
27:12 >> Yeah, it's awesome. Okay, so now we've
27:13 create found out what products are
27:15 available for us. Let's actually go
27:19 create our media buy. So, I um didn't
27:20 want to type this, so I've got it copy
27:21 and pasteed. Um but we're going to
27:24 create a new buy on my um awareness
27:26 campaign. So, I remember I've got my
27:29 campaign up here. Um we're specifically
27:31 going to target the LG content network.
27:33 Um so, I wanted video content. This is,
27:36 you know, a guaranteed product format of
27:38 video. So, this is perfect for me. So,
27:40 I'm going to create a new buy. gonna do
27:43 a 10k budget, $20 CPM, and I want even
27:46 pacing. Um, and then the other thing
27:48 that I'm going to do at the same time is
27:50 I'm gonna create a tactic. Um, and a
27:53 tactic on our side is really allowing us
27:55 to create more specific targeting from
27:57 our campaign. So, we've got our
27:58 campaign. That's like generally what we
28:00 want to do. And then our tactic is going
28:02 to allow us to have specific targeting
28:05 like us, premium CTV, bring in all that
28:07 greatness in terms of our stories and
28:10 standards. So, we've got our tactic.
28:13 We've created our media buy. And this is
28:14 just in draft, so I could look at it
28:16 before I actually execute it against
28:18 Swivel. So, now I'm going to go and
28:21 execute uh this media buy against uh
28:24 Swivel to get approval. And so, awesome.
28:27 It's been submitted. Joe, why don't we
28:28 see what's going on in Swivel. All
28:32 right. So, we are back in
28:35 the seller agent UI. Uh don't see any
28:38 proposals. Let's refresh the screen. Ha,
28:41 we have a proposal. So, you can see we
28:43 have a buyer reference here. We have a
28:48 promoted offering. If I navigate here to
28:52 uh this proposal, you can see the brief
28:55 that was submitted by the buyer agent.
28:57 Um, and then importantly, you can see
29:00 the opportunity to approve the proposal
29:02 um in the bottom right. So, let's go
29:06 ahead and approve this media buy. Awesome.
29:12 >> So, approval sets into motion a series
29:14 of agents on the swivel side. Um, most
29:17 of those agents um are trapping agents
29:19 specifically for the purpose of creating
29:21 the downstream objects that will power
29:24 the media by. So the expectation here is
29:26 that inside the LG ad server spring
29:29 serve where we saw no objects created
29:32 earlier um we should begin to populate
29:35 um the the hierarchy that would execute
29:37 the media buy. Uh so depending on what
29:39 ad server that is you know that includes
29:42 things like campaigns line items um and
29:45 inevitably creative. Uh so I'm now going
29:48 to navigate back to the spring serve
29:51 instance. Uh this is LG Spring serve
29:53 instance. Right now it's empty. Let's
29:55 give it a refresh. And
29:56 And >> awesome.
29:56 >> awesome.
30:00 >> Voila. We uh we have objects. Um here is
30:02 the campaign. Um underneath the
30:06 campaign, it appears as though we have
30:09 line items. Um and on top of that, the
30:11 line items include the targeting. Um and
30:14 importantly, it includes the association
30:16 to the correct supply or the correct
30:19 inventory um to kind of close the loop
30:22 on the trafficking. Um so now that this
30:24 element is complete and the objects are
30:26 made in the downstream ad platform, you
30:29 know, theoretically we're we're in an
30:31 asynchronous process. The media buy has
30:33 been created yet we do not have
30:35 creative. Um frequently you see media
30:37 buys are executed and creatives are sent
30:40 after the media buy is complete. Um so
30:41 now we're going to fast forward towards
30:43 our start date um when we receive
30:46 creative from the agent buyer.
30:51 >> Cool. So let's go back into scope 3. Um,
30:53 and first thing I'm going to do is I'm
30:54 just going to get it. Um, so you can see
30:56 here before is that it was pending. Joe
30:59 went and approved it. And so now in our
31:01 system, we should have gotten that back.
31:03 And we can see that it was approved and
31:05 activated. So this is that async
31:07 interaction that's happening. Okay. So
31:09 let's take its advice. Let's add some
31:11 creatives to our media buys. So the
31:12 first thing that I'm going to do is I'm
31:14 going to create a creative. So in this
31:15 case, it's just going to be a very
31:17 simple creative. It's an MP4 with just a
31:20 media URL. But of course, you can do so
31:22 much more with creatives. Um Joe, you
31:24 want to give us some examples?
31:26 >> Sure. I mean, I think one of the
31:28 advantages of an agent agent transaction
31:31 is that you're not limited to what is or
31:34 is not supported. Um in in a bidstream,
31:36 you can execute on things like home
31:39 screens or sponsorships or pause ads. Uh
31:41 because everything's trafficked into the
31:44 downstream ad server. Um this the seller
31:46 agent knows what to do with the the
31:47 creative files. that knows what to tell
31:50 the buyer agent in order to uh correctly
31:53 execute and uh many many different
31:55 things that have never been available
31:58 commercially to uh a buyer without using
32:00 kind of a direct sales technique will
32:02 will now be enabled.
32:04 >> Yeah. And one thing to also keep in mind
32:06 is, you know, I wrote in this prompt to
32:08 create the creative, but in future this
32:10 might not be a human. This might be a,
32:12 you know, agent that's actually going to
32:13 go and create the creative because the
32:15 next thing that we do is we need to
32:17 assign it to our media buy. So, we've
32:19 assigned it and then we've gone and
32:21 updated the media buy that we created
32:23 before um and we showed you in swivel.
32:25 Um, and at this point, right, it still
32:27 then has to get um actually approved um
32:29 through the creative process. Um, but as
32:30 you can see, we've created the creative.
32:33 So, now let's uh jump over to Swivel
32:36 with Joe um to see what happened.
32:37 >> We're back to Swivel. Look what we have
32:39 in our creatives tab. We have some
32:41 immatini video creative. >> Um
32:43 >> Um
32:45 a human in the loop will approve this
32:48 video creative. Um I think human in the
32:50 loop is kind of a critical thing. I
32:52 think I think everybody that's watching
32:54 probably wants to know what the humans
32:56 continue to do. I think you can see
32:58 that, you know, prompting is still a
33:00 skill. Approval is still a skill. Um,
33:01 but more importantly, it's it's
33:03 important that um, especially for things
33:05 like premium video or things like a home
33:07 screen execution that that someone looks
33:10 at and approves a creative. Um, I'm now
33:14 navigating back to the LG Spring Server
33:16 instance. Um, and what I'm looking for now,
33:18 now,
33:21 um, is a creative in the creatives tab.
33:23 And there it is. The swivel agentic
33:26 crowd matini video creative is present.
33:30 Um, and thus completes a successful
33:33 agentic media buy. Um,
33:36 >> amazing. So, we did that from end to end
33:38 with our agents and let's go make some
33:40 money with this now. So, just to recap,
33:43 my agent working behalf of our buyer,
33:44 our brand was able to communicate with
33:46 the Swivel sales agent. He was able to
33:48 figure out what products inventory was
33:49 out there. He was able to go and create
33:52 that media buy and ultimately create and
33:54 add creatives to it. And the the seller
33:55 agent, you know, today is is
33:57 representative of one seller, but you
33:59 can imagine a future where an individual
34:01 seller could have hundreds, thousands,
34:03 millions of ad products. Um, as well as,
34:05 you know, a full array of different
34:08 sellers to choose from. Um, and this the
34:10 Sweller agent will will select ad
34:12 products based off the the media brief
34:14 such that even in an infinite world of
34:15 ad products, we're selecting ones that
34:16 have the highest probability of
34:19 producing an outcome. Um, so today, uh,
34:22 small scale, but you can see how in the
34:24 future there's a a scalable architecture here.
34:25 here.
34:27 >> Yeah, this is so cool. So cool to
34:30 actually do the first, um, Aentic media
34:33 buy with you. Um, so really awesome.
34:34 >> All right, Emma, really appreciate this
34:37 time. Look forward to many more of these
34:38 in the future. >> Fantastic.
34:39 >> Fantastic.
34:41 >> Thank you, Emma and Joseph. That was
34:44 incredible. And finally, please welcome
34:48 Kyle Dozeman, founding member of ADC CP
34:51 and CRO of PubMatic to round out the
34:53 launch with Next Steps before we're
34:55 joined by a special guest.
34:56 >> What you've seen today isn't
34:59 theoretical. It's not a vision for 2027
35:01 or 2030. This is happening now. Those of
35:03 us who have been in the industry long
35:04 enough have lived through the major
35:07 platform shifts from print to digital,
35:09 desktop to mobile, manual buying to
35:12 programmatic. But what makes this moment
35:14 different is the speed and the stakes.
35:17 If we look to history for context, the
35:19 world has seen about 20 technological
35:21 waves in the last 10,000 years. So a
35:23 wave is a set of technologies coming
35:25 together with profound societal
35:27 implications. Those that enable seismic
35:30 advances in what human beings can do. So
35:32 whether it was the combustion engine or
35:34 agriculture or the internet, these waves
35:37 transform everything. But the speed of
35:39 these waves is also accelerating. The
35:41 first seven waves occurred over 10,000
35:43 years. The next seven only took about
35:46 200 years and the most recent seven less
35:48 than a hundred years. The speed at which
35:51 AI will reshape our entire industry will
35:53 surprise all of us. Agentic advertising
35:56 isn't coming to disrupt us. It's coming
35:58 to empower us, but only if we build it
36:00 right and only if we build it together.
36:03 So what we're announcing today, ADCP and
36:05 this independent industry group is our
36:07 collective answer to one simple
36:09 question. Can we build the next phase of
36:11 the open internet better from the start?
36:13 Or will we make the same mistakes that
36:15 we made in the past? I believe that we
36:17 can get it right. And here's why I'm
36:20 confident. Because you are here. Over a
36:23 thousand of you said yes to this moment.
36:25 You recognize that being in the room
36:27 where standards are being set is more
36:28 valuable than watching from the
36:31 sidelines as proprietary walls go up
36:32 around you. So let's be direct about
36:35 what's at stake. So without a protocol
36:37 like ADCP, we're headed towards a world
36:39 where every platform builds its own
36:42 agentic solution. Every integration
36:44 requires custom development, the walled
36:46 gardens get higher, toll booths continue
36:49 to proliferate, and costs rise, and we
36:50 once again lose transparency and
36:52 effective working media dollars to drive
36:55 growth. or and this is the path that
36:57 we're choosing today. We create an open
36:59 standard where innovation accelerates
37:01 because everyone can plug in.
37:03 Competition thrives and the barriers
37:06 come down. Advertisers and publishers
37:08 can connect directly with true trust and
37:10 transparency and the entire ecosystem
37:12 grows stronger together. This is the
37:14 founding moment. This is where we decide
37:17 which future we're building. So, here's
37:19 what happens next. And there's a place
37:20 for everyone in this movement,
37:22 regardless of where you sit in the
37:24 ecosystem. First, join the community.
37:26 So, if you're a practitioner, a
37:28 strategist, someone who just wants to
37:30 understand how Agentic Advertising will
37:32 change your day-to-day work, join our
37:34 community. We have Agentic Advertising
37:36 meetups already scheduled. These are
37:38 where we'll roll up our sleeves and talk
37:40 real use cases, real challenges, and
37:42 opportunities. Join the LinkedIn
37:43 community. That's where we'll have
37:46 ongoing discussions and updates. This is
37:47 where you can stay ahead of the curve.
37:49 The link is live. Make sure you sign up
37:52 before you leave today. Second, start
37:54 building. So if you're a developer, an
37:56 engineer, a technical team, ready to
37:58 implement, it's time to build.
38:00 Publishers, the open source code is
38:02 available now. Access it directly or
38:05 contact us for implementation support.
38:06 Let's get you up and running as quickly
38:10 as possible. And for any team building a
38:12 buying agent or a platform, join the
38:14 ADCP Slack channel. Documentation,
38:16 testing tools, and a community of
38:19 builders are waiting for you. This is
38:21 how we prove the protocol works by
38:23 putting it into production together. And
38:25 third, let's shape the future through
38:27 governance. So if you want to influence
38:29 the direction of this protocol, the
38:30 standards that we're setting, the
38:32 priorities that we are pursuing, get
38:34 involved in governance. We're forming a
38:37 nonprofit entity to govern ADCP
38:39 industrywide. This ensures no single
38:41 company controls the protocol and that
38:43 it belongs to all of us. If you want a
38:45 seat at that table, just reach out. This
38:47 consortium is only as strong as the
38:49 diverse voices that help shape it. And
38:50 so let me leave you with this. Today we
38:53 have a major opportunity in front of us.
38:55 The foundational models are ready. The
38:57 infrastructure is built. And the market
38:59 momentum is undeniable. And most
39:01 importantly, we are ready. This room,
39:04 this community, this coalition. So what
39:06 happens next isn't up to any single
39:08 company. It's up to all of us. To the
39:10 publishers. You have a unique
39:12 opportunity with inventory and audience
39:14 insights that brands and their agents
39:17 need. Open your doors through AdCP. To
39:19 advertisers and agencies, push your
39:20 technology partners to adopt open
39:23 standards and get building and testing
39:26 your brand agents to be out in front. To
39:28 the builders and developers, ship code,
39:30 test protocols, break it, improve it,
39:33 make it stronger. And to everyone in
39:35 this room, tell your networks. This is
39:37 only effective if the industry moves
39:39 together. Share what you learned today.
39:40 Invite your colleagues to the next
39:42 meetup and help make this movement
39:45 contagious. The window for defining
39:47 standards is narrow, but it is open
39:49 right now. Let's not look back in 5
39:52 years and wished we'd acted. Let's look
39:54 back and say we were there. We built
39:56 this and we got it right. So, thank you
39:58 for being here, for building this future
40:01 with us. And now, let's get to work.
40:03 Before we close, all the links that you
40:05 need will be in a follow-up email that
40:07 hits your inboxes within the next hour.
40:08 the community sign up, the Slack
40:10 channel, the documentation, the
40:12 governance information, it's all there.
40:14 So, don't wait. Dig in and get started.
40:16 So, thank you again. This is just the beginning.
40:17 beginning.
40:19 >> Thanks, Kyle, for underscoring the next
40:21 steps for how to get involved. Before we
40:23 go, please welcome a hot take from the
40:26 industry's one and only Terry Kawaja.
40:28 >> AI and advertising disruption of
40:30 tectonic proportions. Cuz this is the
40:33 big one, folks. AI's effects will be
40:35 broad and deep. It will alter business
40:38 models and change the ecosystem at its
40:41 core. And this coming from someone not
40:43 afraid to call balls and strikes on
40:46 trends, unlike other shiny new object
40:48 trends that undisiplined observers
40:51 willingly went along with despite zero
40:53 evidence that they were nothing but a
40:55 steaming pile of horseshit. So AI has
40:58 been on Lumis agenda for over a decade,
41:00 calling it one of the top trends in
41:03 advertising in 2016. And look, we've
41:05 advised on all major developments in
41:08 adtech, take data, mobile app, and CTV
41:11 as examples, but now doing a lot of work
41:13 in AI. And while we've been able to
41:17 garner above market outcomes, high to
41:19 singledigit multiples of net revenue for
41:22 those sectors, AI is bigger, greater
41:24 valuations, and greater still. There's
41:26 only one problem with this green box.
41:27 It's not big enough. We're going to be
41:30 doing AI AI deals for quite some time.
41:32 Um, we like to think of Maslo's
41:35 hierarchy as a construct to take a look
41:39 at AI's application to advertising. And
41:41 as you go up the triangle, just like in
41:43 Maslo's original, uh, it gets, uh, the
41:45 disruption and the impact are greater.
41:46 So, at the base level, you've got
41:48 workflow, audience measurement, and
41:50 media buying, the meat and potatoes of
41:53 adtech. Next up is content, ad creative,
41:56 and communication. uh further up still
41:58 the big daddy internet navigation and
42:00 finally who knows powerful technology we
42:02 don't know what it can do. So to take
42:05 them one at a time at this core layer
42:09 essentially you know this right ad tech
42:12 is a largely manual largely sequential
42:14 largely linear sort of process from
42:17 planning targeting buying optimization
42:21 reporting uh and measurement and what AI
42:22 brings to the table is the ability to
42:24 bring all those together and do them in
42:26 real time and if you were noticing with
42:29 a lot less heads how many more heads I
42:32 don't know we'll see uh next up is
42:34 content, ad creative, and communication.
42:36 Here, I think of it like an iceberg.
42:38 What you see above the surface are the
42:41 no-brainer reduced costs of asset
42:43 production in the name of efficiency.
42:44 And that's all well and good. Everyone
42:46 should do that. But I think what moves
42:49 the needle is what is below the water
42:51 surface around data to determine which
42:55 ads uh creatives will work for driving
42:57 effectiveness. Think of the top as
43:00 tactical, the bottom strategic.
43:02 And we've had this 25-y year construct
43:04 of navigation where we try and find
43:06 people and potential customers and drive
43:08 them to our website to convert them.
43:10 Hell no. Go to where the people are.
43:12 Take your message to them. The benefits
43:14 are you go to them. It's a one-on-one
43:15 conversation. It's a shorter path to
43:17 conversion. All right. Uh internet
43:19 navigation. This is the big one. This is
43:21 search blue links going to an answers
43:24 economy. And what's driving this is the
43:26 more capable LLMs, the unprecedented
43:29 consumer adoption, emergence of new AI
43:33 interfaces uh from chat to assistance to
43:37 uh uh coming soon uh browsers and now a
43:41 paniply a spectrum of monetization uh
43:43 options ranging from crude data
43:44 licensing all the way through to
43:47 contextual ads and more that we don't
43:49 even know about. And a lot of companies
43:50 doing this, large companies, small
43:52 companies. We welcome Microsoft who just
43:54 announced they're getting in this game
43:57 as well. You can think of these as tech
44:01 traction services and models. uh and
44:03 based on the amount of dollars, the
44:05 massive amount of dollars invested uh in
44:09 LLMs to date and the necessary required
44:11 return, there's zero chance they're
44:13 going to derive all of that return from
44:15 a subscription business considering less
44:17 than 10% take up today, which is all
44:20 roads lead to advertising. So whether
44:22 it's new entrance getting into media or
44:25 LLMs, advertising will be it. Uh and we
44:27 don't know by the top, that is the
44:29 ultimate unknown unknown. Um, AI, if you
44:31 think about it, is like search intent
44:33 data on steroids, right? You type in
44:35 Cleveland divorce lawyer, we know three
44:37 things about you. You live in Cleveland,
44:38 you're getting a divorce, and you need a
44:40 lawyer, right? Think about that in an AI
44:43 context. Uh, the user starts with on
44:45 average a 25word prompt and has multiple
44:47 interactions. The reality is it's a
44:51 higher intent signal. Um, Zuckerberg
44:53 even thinks this is going to change the
44:56 ratio, the perspective of uh advertising
44:59 spent to global GDP. We'll see. So, one
45:01 thing is clear. Doing nothing is not a
45:04 viable strategy and but yet there are
45:05 significant consequences from corporate
45:08 development to the ecosystem revolution.
45:09 Let's try and get through these.
45:11 Discerning true ad tech is challenging,
45:13 right? We characterize these companies
45:15 as you're either AI first, a fast
45:17 adapter, or you're an AI poser. I know
45:19 it's a harsh term, but let's be real.
45:22 Uh, and based on the feedback I got from
45:24 my LinkedIn post, a lot of people are
45:25 focused on this. We all know that
45:27 outcomes are better than proxies and
45:28 we're living in the outcomes world.
45:30 We've been talking about this phenomena
45:33 for over a decade of how uh scale and
45:35 automation can bring uh operating
45:38 leverage and look we think of it as a
45:39 continued gestation as we said at the
45:43 ANA in 2018 that started with gaming and
45:45 is moving to all uh marketers. It's you
45:47 could think of gaming as canary in the
45:50 coal mine here. Historically performance
45:52 defaulted to deterministic measurement
45:53 and we know all the problems of last
45:55 click the perverse incentives the
45:56 shenanigans the lack of scale of
45:59 firstparty data airgo the probabilistic
46:02 world being better although complex slow
46:04 and expensive. So you bring AI into the
46:06 equation and all of a sudden you can
46:09 turn that probabilistic into predictive
46:11 measurement buying and ultimately
46:13 outcomes like the big guys do. We
46:15 believe that supply side decisioning is
46:18 the next evolution for optimization.
46:20 what started as waterfall uh and then
46:22 header bidding and then we had SPO and
46:24 curation and we believe supply side
46:27 decisioning is the next phase that
46:30 leverages AI to not take take a a view
46:33 of the entire spectrum of inventory not
46:35 just the signal coming from a QPS uh
46:38 throttled uh demand side view and we
46:41 believe this will help the open internet
46:44 compete against the walled gardens or as
46:46 in the words of my friend Snoop Dogg
46:48 outcomes get high with your own supply.
46:51 >> There you have it. Uh by the way, uh uh
46:53 uh yes, what do I mean by compete with
46:54 the wall gardens? Well, you know, we
46:56 know traffic is moving in the wrong
46:58 direction towards wall gardens. And how
47:00 do we preserve that in the open web and
47:02 the uh hedge gardens? Well, let's take a
47:04 look at the attributes of the wall
47:06 gardens. Here are the five. And we
47:09 believe that with uh uh third-party
47:11 inventory and supply side decisioning
47:13 and maybe a little bit of M&A in there,
47:15 we can replicate most of those
47:17 attributes and replicate those results.
47:19 Don't believe me? Apploven's already
47:22 done it. They have executed both organic
47:25 and inorganic strategies that have
47:27 allowed them to replicate all five of
47:29 these attributes. How do we know this in
47:32 such detail? We advise them. How did it
47:34 turn out by the way? Uh unbelievable.
47:37 Since launching Axon 2.0 companies on a
47:40 76% compounded growth rate on revenues
47:44 combine that with uh their 80% EBIT
47:46 margins, it's a rule of 150 plus company
47:49 with a market capitalization of $235
47:52 billion which is greater than every
47:55 other ade competitor that's public and
47:58 every single advertising holding company
48:00 period. full stop with enough leftover
48:03 to buy a few, you know, fun companies
48:06 like Pinterest and Sirius and Snap and
48:08 the New York Times and depending on the
48:12 day, Warner Brothers. We view this as
48:16 the best M&A case study ever. The other
48:18 thing that AI does is it reframes what
48:21 performance means. Connecting brand to
48:24 performance. Probabilistic AI connects
48:27 these two worlds so that you end up with
48:29 something combination of the two. We'll
48:31 call it brand performance that measures
48:32 impact on outcomes and assesses
48:35 predictive KPIs. The implication being
48:37 it allows you to buy on guarantees and
48:39 it shifts ad spend uh from a
48:41 discretionary expense to a cost of goods
48:43 sold. And the other impact is that it
48:45 effectively shortens the path to
48:47 purchase. Think about it. And when you
48:49 can show a tight connection between
48:53 spend X and revenue Y with LTV to CACM
48:55 math, you change the nature of
48:57 advertising from a discretionary
48:59 expansion up the income statement to a
49:01 cost of goods sold. And then there's no
49:04 off button on campaigns. This is why
49:06 Adam Fero, founder of Apploving, is
49:15 >> Again, canary in the coal mine. He saw
49:19 it first. AI is a gamecher for
49:21 advertising. I don't know about you, but
49:22 I'm sick and tired of this incremental
49:24 approach we have. Our product is 10%
49:26 better. No, I want X. I want our product
49:28 is 10x better. And when you think about
49:30 it, right, every order of magnitude
49:33 change in advertising has come from
49:35 native formats on native infrastructure.
49:37 Think Google Search, Facebook, Tik Tok,
49:41 AppLov, and coming soon, Open AI. That's
49:42 right. They've already announced they're
49:44 building out infrastructure, which begs
49:47 the question, will AI catalyze a long
49:51 overdue great reckoning in adtech? We
49:53 would submit that AI will separate the
49:55 wheat from the chaff and drive much
49:56 needed rationalization and
49:58 consolidation. The Lumaccape, after all,
50:01 is a fragmented ecosystem of largely
50:03 undifferiated companies built in the
50:05 Zerp era, which ek out their existence
50:08 on revshares and kickbacks. Harsh, but
50:11 true. The result is an inefficient
50:13 supply chain with economic opacity,
50:15 perverse incentives, shenanigans, and
50:18 outright fraud. What's needed are fewer
50:20 players doing more volume at lower take
50:22 rates with better quality. In other
50:25 words, like every other industry. This
50:29 one just refuses to grow up.
50:33 >> Some advocate for a revolution that
50:34 blows up the existing programmatic
50:38 ecosystem and starts over.
50:42 Yes, comrade O. Kelly driving charge
50:45 with ubiquitous post on LinkedIn about
50:48 how AI is going to reinvent supply
50:51 chain. Okay, enough of that. Point is,
50:54 he's not wrong. Uh we are due for a
50:55 great reckoning and AI may be the
50:58 catalyst that pushes us over the edge.
51:01 By the way, by the way, given the
51:03 ubiquitous ofveness of his posts, one
51:05 wonders what kind of automation
51:08 technology he's utilizing. ah explains
51:10 everything. And if you think about it,
51:12 we are entering what we call the fifth
51:16 phase of adtech, right? We had the yes
51:18 digital ad spend has grown from 13
51:20 billion to half a trillion over the
51:22 course of these 25 years. And Luma was
51:25 launched in 2010 to take advantage of
51:27 this very opportunity that we saw
51:28 coming. But it hasn't been a smooth
51:31 ride. Right? The first 10 years we had
51:34 the ad network era and what I call the
51:36 early programmatic era. We started with
51:38 contextual, we moved to audience. We
51:39 started with bootstrap companies, we
51:42 moved to venture financed. Then we had
51:44 the next 10 years, we hit strong
51:46 headwinds from wall gardens and a
51:48 variety of issues in the ecosystem. We
51:51 actually reached peak lumiscape uh uh
51:53 fragmentation and then we entered the
51:56 scale and PE era while you know CTV and
51:58 mobile app and commerce media continued
52:00 to grow but we had the cookie circle
52:02 jerk and we had all kinds of shenanigans
52:04 going on in the marketplaces. Boom and
52:07 bust. The key point being innovation
52:11 waned and here we are. Thank God finally
52:14 we are in phase five we're calling AI
52:16 reinvention to see a rapid innovation
52:19 cycle lot of disruption outcomes will
52:21 accelerate that and of course Google's
52:23 on their back feet because of antitrust.
52:25 We believe this will bring a muchneeded
52:28 return to strategic M&A imperatives and
52:31 finally sector consolidation or as my
52:33 partner Connor said to me the other day mortimer
52:35 mortimer
52:44 calls me a mortimer. So uh you get the
52:45 idea and hopefully you're on board for
52:48 this as well. But the AIdriven future
52:51 holds much promise of an exciting new
52:53 reality with emphasis on the word new.
52:56 Think about it. $500 billion market cap
52:58 of a private company makes them the most
53:00 valuable. They've got a new consumer
53:03 interface, new product, new pipes, new
53:05 value proposition, new business model,
53:07 and a new ecosystem. At the end of the
53:09 day though, we think this is a win for
53:12 consumers. We think this is a no-brainer
53:14 win for advertisers. agencies could go
53:16 either way depending on whether you move
53:18 fast or not, organically or
53:20 inorganically. Same thing with the ad
53:22 tech ecosystem. The winners will be the
53:24 ones that move fast. When it comes to
53:26 publishers, I think the large ones will
53:28 do fine uh and the small ones and the
53:31 news ones, I'm worried about them. Uh we
53:34 have to find a solution uh for them. So,
53:36 what are my key takeaways? Changes on
53:38 hyperdrive rethink everything in this
53:40 environment. Legacy assumptions, they're
53:43 out the window. Trust but verify has a
53:45 whole new meaning. Separate the wheat
53:48 from the chaff. Finally, thank God with
53:49 a better business model for the
53:51 surviving intermediaries. And
53:53 ultimately, we believe inorganic growth
53:55 key to winning. Now, I know what you're thinking.
53:55 thinking.
53:57 >> I'm shocked. Shocked to see Luma
54:00 pitching M&A is a strategic solution.
54:01 >> You're winning, sir.
54:03 >> Oh, thank you very much.
54:06 >> Thank you all and good luck on this very
