Opening Keynote: The Age of Agents | AirOps & Alex Halliday

A massive, massive welcome to you all. As Ali mentioned, this is our first ever dedicated Airops conference, and we really brought together our favorite customers, partners, thinkers in the space to try and do a couple of important things today. Obviously, talk about Airops and some of the amazing things the team's been cooking on, which we're very excited to show you, but also hopefully share some of our research, some of the playbooks that we're seeing working in the wild, and give you some things that you can go back to your team next week and start actioning. I would like to start with a little bit of an acknowledgement before we dive in today. And that's that these are actually quite challenging times for the practice of marketing and the channel that we spend our time thinking about. On a couple, on a couple of fronts. First of all, the way that consumers are finding and buying their next favorite product is shifting as people increasingly turn to generative search experiences. They start to consult agents more on what purchases they should be making. And then when it comes to the work of marketing, after a few false starts, this looks to be the year that we will finally be welcoming agents onto teams to do real valuable work. And part of the problem is that the people in this room are expected to have all the answers. And the reality is that most of the playbook simply doesn't exist yet. So before I go into some hopefully useful content for everyone, we have made a short video that I would love to show you.

The way people buy is changing, but this is nothing new. Print, television, internet, mobile, social— each one rewrote the rules. Each time we figured it out, leveled up, and did better work than ever before. This shift is no different, just bigger. Your customers are still searching, just not where you can see them. They go to the prompt now and agents do their research for them. And somewhere in that conversation, a recommendation gets made about your brand and without you involved. Some might tell you to feel scared, like this is an extinction event and the only way to act is to panic first. But we know you. Fear isn't your fuel. It never has been. You don't need convincing that AI search is here. You need a way to win it. We built the platform and the team to help you win. This is the biggest and most exhilarating opening marketers have ever had. The marketers who move now get to write the playbook. With AirOps, you see exactly how your brand shows up in the world of AI search. You multiply what's working, you fix what's broken, and you deploy an agent team that runs your strategy at a scale your org chart art never could. No matter what kind of brand you're building, you can now stop reacting to a narrative you had no input on and start shaping it instead. The marketers who move first will define this era. With AirOps, you're one of them. Take charge of your growth, own your narrative, and own what's next.

So the theme of today and something we're gonna keep coming back to is that we're betting on the people in this room to figure this out together. And events like our AEO Conf, AROps Next, our education programs are all designed to slowly but surely build our collective understanding of how to win this next chapter. So with that, let's dive in. I'd like to start today talking a little bit about how discovery is changing. Around us. And discovery has always been, uh, something that's shifted, but the rate of change that we're experiencing right now can make things feel a little bit disorienting. We know that buyers are increasingly shifting to generative search experiences. ChatGPT itself is knocking on the door of a billion weekly active users. People are turning to personal assistants like Claude, uh, to really change the way they think about the consideration and discovery phase of the buyer journey. And Google is increasingly becoming generative through first AI overviews and now the increasing rollout of AI mode. And probabilistic answers are now your job as marketers to influence. And this transition really shows no sign of slowing down. So as Google rolls out things like Personal Intelligence, as ChatGPT accumulates more context about you as a user, those answers are gonna get higher quality and more proactive. We also have these generative surfaces moving further down funnel with things like the apps ecosystem, things like the Universal Commerce Protocol from Google, and the e-commerce ecosystem evolving around ChatGPT. And we know that our, our, our long-term friend, the 10 blue links, is slowly being chipped away at with an increasing portion of the search results being generative. And agents being our companion for work and life is a trend that is really accelerating as they get better and better and become more front and center in how we decide. And this ladders up to just a fundamentally different buyer journey. So in the old world of head terms where people would be typing in 4 words to get a page, we would be fighting over those head terms. Now in AI search, the average query length is 20 to 60 words, which pushes so much of that demand into the long tail. The intent specificity goes up, not just because those questions are longer, but because there's more individual context attached to that request. We start to really think about the long tail as being the new battleground, um, where previously we would be thinking about maybe 1,000 head terms that we really focus our efforts on. And now the long tail becomes this overweight influencer of those responses that drives so much of our top and mid funnel responses. And then of course, when we think about the unit of work, we used to really focus on the page, and now there is a shift towards the chunk, that citable passage with the heading, first two sentences that really nails the response and gets cited and puts your brand front and center. And this is probably why this quote that we, we like to talk about, um, is so relevant. And that is that, um, content on-site and off-site is becoming the new fuel for agent influence. And that, that content has to be high quality and it has to be targeted to the specific needs of that buyer at that specific part of the funnel. And a lot of the themes of AI search are familiar from the world of SEO, but as we like to think about it, the contrast on some of those has simply been turned up. And as we are having abundance of content creation tools at our disposal, The bar is really getting raised all the time. I'm kind of here to remind you that content is a competitive sport and it's our job to make sure you win it. And with that, I would love to start to now talk a little bit about the emerging playbook and our observations as we look at the data available to us at AirOps. So we're not here to say SEO is dead. SEO is still front and center to the practice of AI search. But the way we like to talk about it is that SEO solved one face of the Rubik's Cube, and that cube has now rotated. Just to zoom out a little bit, AI search is fundamentally a game of winning consensus. A buyer question gets typed into a generative search experience. That model will make a real-time determination as to whether it needs to go and fetch new information or not. Some pages will be retrieved, and then, uh, those sources will be evaluated. Snippets of those will then be returned to the user, and there will be an answer, and you will be mentioned, you may be cited, and there may be an opinion that the model has about your brand. And it is now our job to understand how to influence that and give you the tools to do it. So this room mastered SEO. As the first act. But there are now another 5 surfaces to consider that we'll talk about today. First of all, what are the preferences of the models when it comes to citation? What things do they look for in the content that they're reaching out to get that will help you understand how you should construct your content, how you should think about headings, how you should think about, um, different quality signals that we'll go into. The second is brand voice. In a world where AI is abundant, it's really easy to take the storied brand that you've all spent years building and completely flatten it. And it's our job to make sure that doesn't happen. Offsite has always been important in SEO, but it's, it's incredibly important in a world of AI search where as much as 85% of top-of-funnel influence is driven by offsite recommendations and opinion. The freshness burden. This will come up multiple times today. Models reach out of their training corpus because they want fresh content. And so as a result, It's never been more important to make sure the content you create is up to date. And then of course, the question of attribution. So AirOps is many things. We are at our core a platform for doing the best work of your careers, but also we are a data platform as well. And over the last 7 or 8 months, we've collected nearly 2 billion citations across many different AI search experiences. And I wanted to just share a couple of the observations we have. That may be useful to you as you go back to your teams tomorrow. The first is thinking about the construction of a page and this idea of a chunk. I wish we had a better word for it. We're going to roll with this. A chunk is a piece of the page, usually paired with a heading, that is optimized for the way the models are currently behaving today. And again, this is a moving target, and we will continue to update as we observe different behaviors. But as of today, Typically, pages need to have headings, H2s and H3s, that really match the buyer question that's being asked. That is one of the strongest quality factors between citation chunks and the question being asked. And within the first two sentences of that response, comparison language, freshness signals, and real references all really support, uh, the model in seeing that as a trusted piece of text that adds to the quality of the response they're giving to their end user. And the semantic distance between the heading and the question being asked is a cheat code in many ways for, uh, making sure that you're, you're moving up the, the likelihood of being in the response. The second is content freshness. And, um, this will come up again in, in other people's talks today, but the models reach out of their training corpus because they want fresh content. Therefore, If you have a current year reference, you're signaling to the model that this has been updated recently. The content has been refreshed. That works both for EAT and Google's algorithm updates, but also the models absolutely love it as well. And the third is SERP ranking. So we are moving on from the doom and gloom of SEO. Your SERP ranking, and this is data from Kevin. Thank you, Kevin. He'll talk about it more today. But SERP is still the biggest factor in driving the likelihood that your page will be mentioned in an AI search response. So between position 1 and 10, your likelihood of citation goes from 58.4% all the way down to 14.2%. So when we think about how to evaluate the content we are creating, we offer this as just a snapshot of some of the signals that we observe across the models. That's super important. And we ranked over 100 different factors, and we'll make the research available to everybody. But what this comes back to is fresh, high-quality content, original opinion, many of the things that worked incredibly well in SEO but are all the more important in the world of AI search. This is a fun one actually, which is complementary clusters. So, Quite often we think about a single page tackling a single basket of intent. What we actually see is that different pages that are complementary and non-cannibalizing can actually help you get at more of the picture. So in this Instantly example, we have a comparison guide, a best sales tools article, and a guide for AI agents. And this allows you to get more of the intent surface area in a way that's complementary and helps boost your overall visibility. And then of course, review presence is incredibly important, particularly in B2B. Across our dataset, we actually saw in Q1 the importance of review sites growing from, uh, by as much as 7 points in the last quarter. So the models are increasingly looking for that frontier opinion about your, about your product, about your service, so that they can make sure that those buyer questions are being answered with, um, first-party experience an authoritative opinion about your service. And across our customers, we see this playbook working today. So whether it's Webflow boosting AI attributed signups from 2% to 10%, Chime 3x-ing their citation rate, or Ramp really starting to think about being thoughtful on how they show up in Reddit, we're seeing the playbook that we at AirOps have been developing with our customers delivering real results. So now I want to transition a little bit and talk about some of the amazing work that our team has been doing on the platform. And the big shift that we are excited to share today is the transition from building workflows, which we know everybody loves, uh, to starting to think in agents. Now, workflows were the first act of, of content engineering. And we were incredibly proud to be the first platform to push this at scale to the market. We launched our first content engineering course anchored around the construction of workflows. And we were the first to really start building very thoughtful, deep, precise workflows for the best brands in the world before other people decided that sounded like a good idea and started shipping their own. Um, and to be honest, workflows taught us a lot. They taught us that we can take LLMs these probabilistic kind of super machines and turn them into really valuable work that reflected our brand tone and voice that contributed to the conversation. And they taught us how to have an interplay between humans on the team and these LLMs in order to do work that, that brands could be incredibly proud of. And it got us to where we are today, but it will not get us to where we want to go next. And so I wanna talk about agents for a second. You are going to be inundated with talks of agents from every single vendor in the world. Your LinkedIn feed is probably overrun with them, but in our view, agents are often talked about in marketing, but actually rarely seen. And I think just to quickly talk about how we think about the definition of agent, which hopefully will arm you in your own procurement processes, and as you think about bringing agents onto your team, there are a few really important criteria that I would just ask you to keep in mind. The first is a real agent can be proactive. So it not only is it responding to you pushing a button, it's also coming to you with ideas. The second is that it can act with incredible precision. Your brands need precise work in order to be able to, you know, put things out into the world. And, uh, so a lot of the agent products out there simply do not hit that bar. The third is that full strategies can be codified into agents, which we'll talk a little bit about. The fourth is that they include your team as they work. These are not autonomous processes that just operate without any human oversight. The fifth, and a really important one, is that the more they work, they get better. So they get better over time. And the sixth is that they can prove their impact and ROI. So this is, this is the criteria I ask you to judge all the agents you onboard. Against. And so we're going to announce 4 things today with a 5th bonus. The first is that we have specialist agents coming to AirOps as of today. These are agents that have been shaped by the billions of citations we've observed. They understand your brand, they learn the channel, and they understand over time your tastes and preferences. And we are replacing the workflow with something we're really excited about called Playbooks. Playbooks are a natural language canvas. They look like a document and they allow your team to collaborate over incredibly precise strategies, leveraging all of the data we have on the channel, on your brand, on your team. You can loop in people at any stage you want. You can leverage any number of third-party services. This is an incredibly exciting surface for your next big ideas to come to life. Oh, it's a nice little thing. The second thing I would like to talk about is our inbox. And so a lot of the talk around agents is we are handing work off, right? So we're thinking about humans no longer being in the loop. We actually think that the best agents are the ones that include more people in the content process. The best content your team will ever produce, we think, is that which has a little bit of a creative tension in it. So you can ask other people what they think. You can ping the right subject matter expert. You can include them in the content process if they are the best person in your organization to contribute. And so we're rolling out an inbox today that members of your team can be included in. They will be able to be pinged. They can answer questions. They can provide their perspective so that you can get more of that valuable knowledge that's sitting inside your company outside into the world and be rewarded with it, with, with increased visibility. The third is Quill. So Quill is the answer to one of the biggest questions we've ever had on AirOps, which is, how do I learn this thing? Um, and Quill is an agent that lives across the entirety of the AirOps surface. It can answer questions on how to use the product. It understands all of your AI search data. It can help you improve the quality of your brand context, and the team will be showing it today. And as marketers, we also have sometimes in the history of AROPS created too high of a hurdle to building with the platform. And the goal here is that Quill can lower that, make it more fun, and hopefully will become a welcome addition to your team. Quill also runs all of your playbooks for you. So you can think of this as an agent captain that will be onboarded onto your team and will become a really valuable member of your team. The crew. And then the fourth is impact tracking. So much of the work that we have seen happen in this space over the last few years has felt, let's just say, a little bit like spaghetti throwing, where you do a bunch of things, you hope the numbers go up. We really want to make sure you will get the recognition for the work you're doing that you deserve. And so we are excited to be rolling out as part of the Playbooks project and Quill impact tracking of everything you do. So you get to the end of the quarter, you can see the impact you've had as a team. You can show that to the rest of the organization, but you can also importantly understand where to place your chips next. And we really believe at our core that the organizations that can learn the fastest are the ones that are gonna race ahead. And so being able to understand the impact of the work you're doing is critical to that learning loop. And if we think about what's happening and changing around us and why this is so important, We do believe that content is getting more competitive. We think work is changing quickly, and the access to AI models in every surface we interact with is just becoming ubiquitous. And so there is a need for us to dig deeper to make sure you can continue to get the win, stand out, and do the best work of your career. And then the bonus fifth one is that we're really excited to be relaunching Airopps University. Airopps' education programs have been at the core of what we've been working on since we when we first got started. And we think it's really important to help folks turn what honestly often feels like a lot of anxiety about the speed of change into highest, high-quality work and progress and being in the craft. So we're working with the AROps team and a number of people in this room to bring about a new day for AROps University, which will be going live in the next couple of weeks, and we can't wait for you to see it. And our wonderful students. You may have seen our wonderful students across many public transit bus stops and trains across New York City and San Francisco. We're excited that our students will continue to be part of that program with the AROPS Champions Program. And just to kind of wrap this up a little bit, we really think about the content engineer as being the MVP of the marketing org through this transition. Content engineering started as the discipline that took the big ambitious ideas with the previous generation of AI models and turned them into workflows that did productive work. The next stage of this is really thinking about content engineers as the agent shaper, bringing the taste, bringing the strategy, understanding the stakeholders in the organization, and really understanding the customer and working with Quill, our agent captain, to bring those ideas to life, um, and have even more leverage and become just incredibly valuable. Part of the organization. And if we do our job right, we will earn the right to say, well, for you to say that AROPS is how you hit your number every quarter. We want to make you incredibly successful in this next act. That is our commitment to you, and that is why we do events like this. That is why we invest in education and webinars and all the great work that we try to put out into the world to make sure you can thrive in the previous era and you continue to thrive in the next.

Era.

And so to wrap up, just to say, our bet is that you are absolutely going to crush it, and we hope that we are by your side as you do it.