What you need to know:

• Chiefmartec's Scott Brinker says AI represents the most profound structural shift in marketing since the web’s arrival, driving not just tool proliferation but fundamental changes in how marketers operate.
• Despite long-held beliefs in consolidation, the martech ecosystem has blown past 15,000 tools, with generative and agentic AI accelerating a new wave of complexity and tool sprawl (though it’s a more nuanced story beneath the surface).
• Execution will fall further behind capability. Brinker’s "Martech’s Law" warns that while technology evolves exponentially, organisations... don't. Which leaves many marketing teams with powerful tools they can’t fully leverage.
• Universal data layers are becoming mission-critical and cloud-based data architectures are now central to AI-powered marketing. Shared data environments break down silos, enabling real-time customer visibility across departments.
• AI is upending traditional pricing models as seat-based SaaS pricing (i.e. dollars per number of users dialled in) gives way to consumption and outcome-based billing (what the customer gets out of said SaaS, regardless of licensed user numbers). Which means marketers must now forecast usage like cloud ops teams or risk bill shock.
• Like the first wave of generative AI on which it is built, Agentic AI is compressing the lead-to-revenue cycle by autonomously executing tasks like lead research, email sequencing, and deck creation: Tell the agents what to do, set the parameters and inputs, plug them in, save serious time.
• The early use cases are becoming clearer. For instance, B2B sales handoffs are being reinvented as intelligent agents stitch together marketing and sales workflows, automate personalised outreach and prep materials the moment leads are qualified, driving efficiency, improving consistency, and lifting conversion potential.
• Governance, not just tech, is the new battleground: With AI agents crossing systems, marketing bosses must build frameworks for authentication, permissions, and cost control before complexity turns into chaos.
• Beware the hype – standards are forming, but security gaps remain: Emerging protocols like MCP and A2A show promise for multi-agent ecosystems, but  without strong governance, AI integration could repeat past martech failures.
• Bottom line: Without data alignment, operational discipline, and cultural change, marketers risk being left behind the biggest upheaval in martech since the emergence of the web.

Agentic AI will structurally rebuild the marketing technology landscape as profoundly as the internet did two decades ago, reckons Chiefmartec's Scott Brinker, who, by and large has been on the money over the last two decades and more.

In this Mi3 deep dive, Brinker pulls back the curtain on the Cambrian explosion of AI-powered tools, the widening gap between capability and execution, and why marketers who fail to get their data and governance houses in order may soon find themselves obsolete. Or at least outcompeted and really, really stretched.

The martech truth-teller, whose day job at Hubspot sees him overseeing the vendor’s partner ecosystem, is happy to punch when challenged about the tech sector’s enduring habit of over-promising, and failing to clean up messes surges forward: martech integration, cybersecurity anyone.

Fifteen years ago, Scott Brinker mapped a few hundred marketing technology solutions. This year, the number blew past 15,000. And while the sheer scale of that explosion is remarkable, Brinker believes the real upheaval is still to come.

Brinker has long tracked the rise, churn, and mutation of martech vendors. The data shows a sharp uptick in generative AI-driven innovation, but it’s not just about shiny new interfaces. Rather, the ability of AI to act autonomously, and increasingly with minimal human prompts, is creating a new category of software that reshapes how marketers work.

Long anticipated, the consolidation story has finally begun, Brinker said. But it’s nuanced. “Back in 2023 to 2024, we saw one of the sharpest rises in solutions in years – something like 3,000 or so new ones. Many were inspired by ChatGPT technologies, generative AI, and new use cases. It became easier to build software, which led to an explosion of new products. But coming into 2024 to 2025, we started to see a number of companies disappear. Our first guess was that it was probably some of those generative AI startups that had launched during that earlier wave."

Per Brinker: "They never really found product-market fit, so they left as quickly as they arrived. But that wasn’t actually the case. Most of the companies that came in during 2023 to 2024 were still going – and growing – in 2025. It was more about the older set of martech tools from the previous decade ... the ones founded between 2014 and 2018. They raised a lot of money and had ambitions to reach multiple billions in value. Some did, but a lot didn’t. I think what we’re seeing now is that their runway finally ran out."

"You had all these companies competing in the same categories with very similar features. Everyone expected a shakeout eventually, but it took a lot longer than most people anticipated. Last year felt like the moment reality finally caught up with a lot of the also-rans."

Far from simplification, this wave brings greater complexity. Brinker sees marketing stacks expanding, not shrinking, However he does not necessarily believe that this will add lots of new complexity to any one buyer.

"For an individual buyer, there’s no way they’re looking at hundreds – let alone thousands – of solutions. What it really comes down to is that buyers tend to focus on specific categories and usually on the top leaders in those spaces," says Brinker. "The only caveat is that when we talk about 'top leaders,' some are truly global and horizontal in nature. But there are also cases where certain leaders dominate a specific region or vertical market. So while there’s still a fair number of solutions overall, the selection set any one buyer is actually considering has gotten smaller over the years."

Still, with 150 new tools a month entering the market, there are plenty of baubles to distract buyers as a time when governance, evaluation, and strategic alignment remain critical.

Martech’s Law in action

Brinker has a metaphor for the dissonance between the speed of technology and the speed of human institutions. He calls it Martech's Law.

"Technology changes exponentially. Organisations change logarithmically," he explained. I.e. tech moves really fast while businesses (mostly) are still blinking and looking around.

It’s a formula for strain. 

"When you put these two curves against each other, we’re in a world where technology is changing exponentially, but the pace of change within organisations is much slower. Those curves keep drifting further apart. I always picture it like having one foot on the dock and one foot on a ferry as it pulls away. Welcome to the 21st-century. Welcome to the quintessential challenge of management, " he says.

That pressure is only intensifying. Brinker points to the rise of large language models and the maturing of low-code/no-code interfaces as a turning point.

"One of the reasons I think we’ve seen a drop in martech utilisation rates over the past five years is that these products have become so large. They have so many capabilities, but a human can only hold a limited subset in their head, just enough to know what to work with."

AI changes that, he suggests.

"What’s really interesting about AI is that it’s not only expanding the set of possible capabilities, it’s also, for the first time, providing a new kind of interface that can actually make it easier for us to tap into those features."

User interfaces are evolving rapidly from menus to prompting agents that make powerful tools more accessible. But Brinker is careful not to oversell ease of change. While technology implementation may improve, cultural and structural alignment still lags far behind.

"I don’t know of a shortcut for that."

According to Brinker, the widening gap between capability and adoption doesn’t just create inefficiencies; it reshapes organisational risk. Companies that fail to adapt fast enough risk losing the advantage of early AI acceleration, while still absorbing its cost.

This is where operational design becomes strategic. Brinker suggests that successful firms are already investing in the connective tissue of governance, process automation, and interdepartmental data flows necessary to turn capability into business outcomes.

But the gap remains real, and growing.

"We’re in a world where the technology is changing exponentially, but those of us running organisations find that the change of the organisation is much slower."

Universal data layers

What’s happening in martech right now is also happening in HR tech, finance, ERP, and cybersecurity. For hard pressed marketers, innovating in martech was hard enough – but how much harder does it get as innovation acceleration kicks in? To that end, Mi3 asked Brinker how much harder is it going to be to achieve a true single view of the customer, and to activate it, if every part of the enterprise is adding new solutions during a time of potentially extraordinary acceleration.

"There are two things that give me hope. First, we’ve been steadily moving toward the idea of a more universal data layer across the business, using tools like cloud data warehouses or 'lakehouses'. Almost every department in an organisation of any size is now feeding data into this common layer. There’s still plenty of work to do in managing and organising that data, but when it comes to getting data flowing across the organiaation, we have better capabilities now than ever before.

"And the second, as we talked about earlier, AI is proving to be very helpful, especially with natural language interfaces that let us access more capabilities without having to learn all the different configuration options."

These changes matter because marketers can no longer operate in isolation. Data silos don’t just hurt campaigns; they cripple AI.

"The customer does not stop at the boundary of the marketing department," Brinker underlined.

AI’s potential hinges on context. That means drawing from service interactions, digital product usage, and post-sale engagement, not just campaigns. With a shared data layer, marketers finally gain visibility across that full spectrum.

"This is an enormous gift to marketing," Brinker said. "Not only can we understand the customer better, but we can also activate more of the organisation as a channel."

Brinker sees this shift as not just tactical, but strategic. It’s the difference between a marketing view and a company-wide view. With a universal data foundation, AI can deliver not just more campaigns, but more relevant experiences.

In a July 2024 article on his Chiefmartech titled The New Data Layer in Martech Has Taken Hold, Brinker elaborated further, writing: "The martech data layer has matured from a patchwork of integrations into a more coherent foundation built on cloud data platforms and universal identifiers."

He argued that what was once the exclusive domain of IT and analytics teams is now central to marketing strategy. “This isn’t about abstract architecture. It’s about business performance,” Brinker wrote.

Critically, he sees the marketing data layer not as a single product but as an architecture that blends data centralisation with operational agility. In the same article, Brinker added: "A universal data layer allows marketers to operate with agility while maintaining consistency and governance across systems."

In the Mi3 interview, Brinker reiterated this viewpoint: The universal data layer is what will enable agentic AI to operate with full enterprise awareness, integrating signals and automating decisions across CX, commerce, service, and loyalty.

It’s no longer just about knowing your customer. It’s about orchestrating that knowledge in real time, across systems.

"We’ve got all this data, from how customers engage with our digital products to their interactions with Customer Service and Support. For marketing, the first step is simply gaining visibility into that data, understanding customer insights, identifying the right moments to reach out, and knowing what message to deliver.

Beyond that, he says, it’s about starting to view those other departments and teams as channels too. "Channels that can help manage customer engagement and move a marketing objective forward, even outside of traditional marketing touchpoints. Yes – please embrace that." he said.

Martech budget shakeup

The surge in AI capabilities is not just reshaping how marketers work, it will change how brands pay for the stack.

For years, SaaS pricing was predictable. Marketers paid by the seat, regardless of usage. That model, Brinker says, is now under pressure.

"What’s been really interesting with AI is the emergence of more consumption-based or outcome-based pricing models."

But, he cautioned, it’s a double-edged sword.

"On one hand, it’s a fantastic benefit for buyers, they’re starting to think, 'Wait a second, maybe we just pay for what we’re actually using or the outcomes we’re actually getting.' It holds the promise of helping optimise spend far better than we could in a purely seat-based model.

According to Brinker, "Now, the flip side is that seat-based pricing has been extremely predictable for martech buyers. It’s like, 'Okay, I’m going to pay X. I might not be fully utilising it the way I’d like, but at least I know my annual budget, my bill is always going to be X.' ”

With consumption  or outcome-based models, it may be economically advantageous, but now brands will have to predict and plan:

"'What is my actual usage going to be? That uncertainty introduces a new layer of complexity for budgeting and forecasting."

The risk? Bill shock, something that Mi3 reported earlier this year is unfortunately common in some parts of the martech space, such as CDPs where many buyers miscalculated their consumption and faced unexpected costs.

Buyer beware, warns Brinker. "I do think a big part of the responsibility in consumption or outcome-based models falls on the buyer. They need to be able to realistically estimate what level of consumption or capabilities they’re going to need."

Still, Brinker sees potential in the shift, and he suggests a Keep Calm and Carry On attitude. "When you actually dig into it, most of the time it’s not that difficult. For example, one common outcome-based pricing model is around using an AI agent to resolve customer service tickets without needing a human to step in."

"If I end up paying $5 for every resolved ticket, yes, there’s some variability, but generally, a business knows. 'Okay, we get about 10,000 support tickets a year.' If we expect AI to resolve, say, 67 per cent of them, then we can estimate the cost pretty reliably and add a buffer."

It’s a different notion than just buying seats for your Customer Service or Customer Success team, but it’s not a crazy model to figure out.

But marketers accustomed to fixed-license costs will face a learning curve. Help is close by, he says. "IT has a lot of experience with this through working with hyperscalers, so it’s definitely something that can be figured out. But we have to acknowledge, it’s going to be a bit of a new notion and a new model for marketing and marketing operations to learn and adapt to.

Brinker’s point is clear: the math isn’t hard, but the mindset shift is. Martech leaders need to adopt financial operations discipline typically found in cloud-native IT.

The result, he suggests, is not just cost control, but precision. And a budget aligned with actual value creation.

It's also a message that's starting to land in the agency channel. As Jen Davidson, Managing Partner at marketing advisory firm Tumbleturn wrote last week in Mi3 Australia, "When pricing is time-based and AI reduces time, a race to the bottom becomes inevitable. Unless agencies dramatically grow new business, revenues will decline. Which is what is now happening for some of the holding company majors."

Marketing’s internet-level shift

Don't think of AI as just another tech trend, Brinker insists. It’s more like the internet itself.

"What’s fascinating about AI is that it’s probably at least as big of a revolution as the internet in how it’s changing the way businesses operate, how customers buy, and how we live day to day. But unlike the internet, which in retrospect felt like a slow burn over a decade AI is moving much faster."

"But with AI, the speed of adoption and advancement is just unprecedented. So yes, I think this is absolutely the biggest wave any of us will have seen in our careers in marketing and martech."

Brinker also draws a sharp distinction between traditional workflows and what agentic AI enables. "There are usually two components to something being agentic. One is its ability to take action. For a while, people made a distinction between a large language model you could talk to and a large action model that could actually do things in the world. But that distinction has mostly disappeared. Now we recognise that LLMs can use tools, they have the ability to take action."

The second component of agency is the ability to work in a more autonomous fashion.  "A lot of our interactions with tools like ChatGPT or Claude are still 100 per cent human-in-the-loop. But as we move toward true AI agents, we see a range, from systems where a human is just monitoring, to ones where a human gets looped in at specific points, all the way to fully automated agents that run entirely on their own."

Brinker says it's the combination action and autonomy that makes something truly agentic. And he is optimistic about what that means for marketers.

"And when it comes to use cases, oh boy, in marketing and marketing operations, it’s a cornucopia of possibilities. Pretty much anything we’ve thought of as a workflow, especially those that used to require manual steps to move from one stage to another, is now a candidate for agent-based automation."

The gain isn’t simply one of efficiency. It’s strategic acceleration. AI gives marketers a new lever to compress the time between lead qualification and revenue.

He likens the current phase to Clayton Christensen’s model of disruptive innovation. "These tools start by handling the stuff nobody had time to do, such as simple, under-served tasks. But they keep improving."

Today, agents can write cold emails and spin up basic ROI calculators. Tomorrow? Brinker believes they’ll take on deeper processes.

That, he said, is when the real transformation begins.

Fly, meet ointment

As it gins up the hype, the tech sector is pushing hard on the vision of a world of millions, and potentially billions of autonomous agents, self modifying at times, and independently working together, safety to the betterment of brands and customers alike.

But there's a problem. Actually, there's a lot of problems at least based on what Mi3 learned during a month long agentic study tour of the US.

For starters, if agents are going to work across systems, they’ll need standards. Most of those standards either don't exist, or are very immature.Then there is the usual and yet to come competing standards stupidity. Brinker, however, thinks that may yet be avoided – and sees early progress.

"Anthropic introduced something called the Model Context Protocol (MCP) and within months, OpenAI, Microsoft, and Google jumped on board," he said.

MCP is a simple framework that allows agents to interact with external data sources and APIs. “It doesn’t solve everything, but it removes friction,” Brinker said. Meanwhile, Google has proposed an agent-to-agent protocol, A2A, but it hasn’t achieved the same adoption. “We’re not there yet,” Brinker acknowledged. But progress is faster than expected.

There's a long way to go. Discovery, authentication, permissions, evaluation, monitoring, and observability are just a few of the areas where technology needs to emerge or evolve if the industry's vision of agentic is to be fulfilled. Even the semantic layer at which LLMs currently operate needs a lot of work. 

Yet even as he acknowledged the momentum, we asked Brinker to tackle a harder truth: the technology industry has an imperfect track record when it comes to keeping promises, particularly around integration, interoperability and cybersecurity (as any Optus, Medibank Private, or Qantas customer will attest).

Brinker didn’t dismiss the criticism. Instead, he offered a clear distinction. “It’s not that MCP has an inherently flawed security or authentication model – it’s that it doesn’t even purport to provide that capability. It’s entirely reliant on a higher layer to handle those concerns. So just keep in mind, MCP is really about enabling these components to plug into each other, it’s more of a connectivity framework than a security layer.”

He added: “The mechanisms for security, governance, and authentication have to be built on top of that. And you're right, that part is still far from standardised. We do have things like OAuth as an authentication mechanism, but even that is just one piece. There still needs to be a layer above it to handle broader governance and control."

But Brinker acknowledged there's plenty of work to be done particularly around registries, authentication protocols, and standards for permissions. "But over time, people do tend to tackle those other things."

Mi3 put it to Brinker – by way of example – that the ‘fix it later,' crowd are not even close to resolving application integration issues in the stack to the extent marketers would prefer. Why should they trust the tech sector to get it right this time?

He offered a blunt assessment of the industry’s occasional inertia: "Yes, the tech industry has been saying 'we'll fix it later' for a long time. But in this case (agentic), we actually are starting to fix it now."

Plus, there is a bot for that. Or soon will be.

Early use cases

Brinker is clear about where most agentic AI activity is starting: practical, narrow tasks that require relatively low risk and high operational burden. "These tools start by handling the stuff nobody had time to do such as simple, under-served tasks," he said. This could include things like cold email sequences, internal research briefings, and ROI calculators for instance.

It’s a long list that includes things that weren’t being done consistently before because the manual effort wasn’t justifiable, he suggests.

While these tasks may seem basic, Brinker sees them as the thin end of the wedge. "You're not going to create your entire ERP system with a prompt," he said. "But something like a quick ROI calculator or a lead research tool? You can do that today, and surprisingly, it works."

He views this wave of task-level automation not as trivial but as foundational. Because these small, repeatable jobs are ripe for disruption, they're also where agentic AI can deliver fast, measurable returns.

While much of the AI conversation is focused on automation and productivity, Brinker offers up B2B sales enablement as an area where strong early results are being found, specifically the hand-off between marketing and sales.

"Previously, a salesperson had to start from scratch. [That meant] research the lead, build decks, write emails," he said. "Now, AI can prep all of that in the background."

Brinker outlines how intelligent agents are now stitching together marketing and sales operations, automatically assembling personalised outreach sequences, briefing documents, and pitch decks the moment a lead reaches a qualified threshold.

"You’ve got a lead in a B2B environment that crosses the threshold and is ready for sales. We can actually start to put together a very personalised introductory sequence over email," he said. "We can dynamically pull together a briefing book for the rep, even create a custom slide presentation."

It’s not just a time saver, it’s a structural shift. Where once marketing handed off a spreadsheet and hoped for the best, agents now orchestrate the transition with precision. Sales reps start their engagement already armed with context, insight, and assets tailored to the opportunity.

"We've cut out a week or so of time and manual labour," Brinker explained. "Now, sales can go directly to what they actually do – human engagement with the prospect."

These tactical improvements are already unlocking strategic momentum. Compressed cycles, greater consistency, and more relevant engagements are giving B2B organisations new levers to accelerate revenue. While Brinker concedes that many of these use cases remain on the tactical end of the spectrum, they are opening the door to broader change.

From hype to operational edge

For all the noise, Brinker's message is equal parts pragmatic and urgent: agentic AI is already reshaping how marketers work, but only for those prepared to rethink how they govern, budget, and build.

He doesn’t romanticise the pace or promise of the martech industry. The hype is real. So is the occasional mess. Yet Brinker sees a clear path forward: anchor AI strategy in operational clarity, universal data access, and a bias for action.

Per Brinker, you're not going to transform your entire organisation with a prompt. But you can start solving the right problems now, and that's where the momentum builds

In a space littered with empty claims and technical dead ends, Brinker’s advice is refreshingly grounded. Start small. Align cross-functionally. Rethink procurement. Don’t try to boil the ocean. But do start now, because the ferry is already pulling away.