What you need to know:

• Generative AI will fundamental shifts in how marketing departments innovate and operate as concepts like “instant software” and “service-as-software” emerge according to the Martech 2025 Report released by Scott Brinker and Frans Riemersma this morning.
• Instant software will enable marketing to instantly create custom applications, automations, and analyses quickly without technical expertise.
• Klarna’s announcement that it will replace Salesforce and Workday Saas with AI-powered custom solutions highlights the trend of tailored development and suggests the new era is coming at us faster than many anticipate.
• Service-as-a-software meanwhile is a paradigm-shifting business model inversion where marketing will move to buying commercial outcomes instead of renting software.
• AI is compressing innovation cycles, with capabilities moving from hype to productivity in 12-24 months.
• Governance challenges loom as AI apps and automations operate below the surface, complicating oversight.
• Custom development enabled by AI is reigniting the build-vs-buy debate for marketing technology teams.
• The bottom line: CMOs must prepare for agile, AI-native tech stacks to stay competitive in the rapidly evolving digital landscape.

Build, buy or AI on the fly

CMOs should prepare for some fundamental changes to the way marketing technology operates as generative AI introduces fundamentally new ways of working such as “instant software," and “services-as-a-software” You're not dyslexic, you read that right.

Those are two key take-outs from marketing technology doyen Scott Brinker from Chief Martech and his collaborator-in-chief from Martech Tribe, Frans Riemersma in their annual Martech 2025 report released this morning.

The report which delves into the impact of AI on martech begins with a cautionary reminder about Amara’s Law: we tend to overestimate the effect of a technology in the short run and underestimate its effect in the long run.

“Pets.com and Webvan, epically failed dot-com businesses from the late 1990s, overestimated e-commerce in the short run. At the peak of the dot-com boom in 1999, worldwide e-commerce transactions totalled ~$100 billion, a mere 0.3 per cent of the world’s GDP. But by the end of 2018, global e-commerce totalled ~$25.6 trillion — B2B and B2C combined — a whopping 29.7 per cent of ~$86.2 trillion global GDP at that point. Amazon went from being a $5 billion company in 2001, its future deeply discounted by Wall Street, to now being the fifth largest company in the world with a market cap of $1.8 trillion.”

The message: no matter how crazy and disruptive generative AI might seem today, you ain’t seen nothing yet.

Take two key trends they say are set to emerge from the messy fog of generative AI-driven change – what they call "instant software” and "service as software.”

According to Brinker and Riemersma, “True disruption often comes at you from an angle you aren’t expecting. While commercial incumbent and challenger platforms are battling it out, we believe one of AI’s biggest impacts will be the explosion of custom “software” in businesses of all sizes — not just enterprises.”

"Instant software" describes what they see as a revolutionary approach enabled by generative AI, where software programs are created on the fly, often without the explicit realisation of the user.

“These custom generative AI implementations can be run stand-alone or incorporated into larger software apps.”

For CMOs this means that instead of relying on pre-built software or extensive custom development, the brand should be able to quickly generate tailored applications, automation, and analyses using AI tools.

They outline that this approach will allow marketing teams to leverage AI for unique tasks like 'hyper-personalised' customer interactions or data analysis without needing deep technical expertise. It's a cost-effective, scalable way to enhance marketing operations and stay agile in a fast-evolving digital landscape​, per the duo.

It is also the latest iteration of the forever dance of buy versus build, something the authors tackle directly:

“What should you build vs. buy? Largely it’s a question of comparative advantage. Where commercial companies in the market have greater experience and expertise in building a particular kind of product or platform, it’s often most economically rational — in both direct costs and opportunity costs — to buy.”

However, they argue, that apps from which your brand can generate unique value due to its experience and expertise are the best candidates for building. “It’s where the expertise in your company — for those specialised needs — is greater than the more generalised expertise in the market at large with commercially packaged products.”

Instant software

Ahead of the report launch Brinker told Mi3: "I'll cover a good example of "instant software" in my presentation tomorrow. I asked ChatGPT to create a chart for me of stock market data. It complied in a matter of seconds. However, when I click on the little icon to see exactly how it created that chart, it shows me that it wrote a Python program and executed it on my behalf. I didn't ask it to write a software program. If I hadn't clicked to get more details on the chart, I wouldn't have even known that a Python program had been written or executed. But one was."

"The power — and challenge — of this era of agent-generated "instant software" is that much of it may happen below the visible waterline."

Brinker said that on one hand, this makes it very easy for companies to take advantage of it. "On the other hand, the governance and guardrails that will need to be placed on these capabilities have a way to go. It's the next evolution of tech stack management. I refer to it as a Big Ops mission, in a nod to the hurdles we once faced with Big Data, because the volume, velocity, and variety of apps, automations, and AI agents running in our environment is going to grow rapidly. Wrangling that will be a big challenge."

Service as software

Service as software is a business model flip where the focus shifts from buying tools and training teams to use them, to outsourcing outcomes directly from AI-powered services.

Put simply, you buy outcomes instead of renting software.

Traditionally, they say, “software has been an assist to labour.”

They predict the rise of what they call 'agentic' AI capabilities — basically AI systems that don’t just analyse data or provide insights but can also autonomously execute actions, making them more like proactive marketing assistants than mere tools.

As a result, according to the Martech 2025 report, software will increasingly serve as labour.

“That is the gateway to a much, much larger disruption than challenger platforms vs. incumbent platforms in the software industry.”

According to Brinker, "Service-as-a-software will be a little easier to pull into a company's digital operations, at least for tasks that have clear hand-off points, such as with first-line customer service issues or SDR activities in B2B sales. Will still need governance, especially around data sharing with these services. But many are designed to address that challenge with integrations to the major commercial martech platforms."

Agentic era

The report also leans into the work of Sequoia Capital’s Sonya Huang, and Pat Grady who earlier this year published a paper called “The Agentic Reasoning Era Begins.”

According to Huang and Grady, “The cloud transition was software-as-a-service. Software companies became cloud service providers. This was a $350bn opportunity.

"Thanks to agentic reasoning, the AI transition is service-as-a-software. Software companies turn labour into software. That means the addressable market is not the software market, but the services market measured in the trillions of dollars.”

For whom the algorithm tolls

Whether the transition to AI-based organisational operating systems presages doom for SaaS vendors, the Sequoia execs acknowledge that their views are evolving rapidly.

"We began with a strong default of 'no'. The classic battle between startups and incumbents is a horse race between startups building distribution and incumbents building product. Can the young companies with cool products get to a bunch of customers before the incumbents who own the customers come up with cool products? Given that so much of the magic in AI is coming from the foundation models, our default assumption has been no—the incumbents will do just fine, because those foundation models are just as accessible to them as they are to the startup universe, and they have the preexisting advantages of data and distribution."

But they note their initial thinking was that the primary opportunity for startups was not to replace incumbent software companies but instead to what they call "automatable pools of work."

Now, however, they are not so sure.

"There’s an enormous amount of engineering required to turn the raw capabilities of a model into a compelling, reliable, end-to-end business solution. What if we’re just dramatically underestimating what it means to be AI native?"

Indeed, it's already happening

Brinker and Riemersma raise the example of the giant Swedish fintech and buy now pay later company Klarna.

Its CEO Sebastian Siemiatkowski announced earlier this year that Klarna was ditching Salesforce (but not Slack) and Workday to build its own custom CRM and HCM applications using AI and composable cloud services.

The Martech 2025 report notes, "While there’s been scepticism about the wisdom of such a 'build it all yourself' strategy, the fact that such a move is even conceivable is a testament to both the improved economics of custom development and the perceived business advantage of more tailored digital operations in the AI era."

Siemiatkowski's comment attracted some controversy, and his PR team tried to walk back some of the comments when Mi3 contacted them earlier this year. But they only challenged the comments on Salesforce, the comments about swapping out Workday when uncontested.

After first trying to deny that Siemiatkowski said exactly what he said (it's on a video after all), the company spokesperson settled with, "We have a number of large internal initiatives that combine AI, standardisation, and simplification to enable us to shut down several software-as-a-service providers."

"This consolidation, supported by AI, will help create a much more lightweight tech stack, allowing us to run the business more effectively and with higher quality.”

Which is not a million miles away from what both Brinkworth and Riemersma, and Huang and Grady are arguing.

Copium

As to how well equipped companies are to cope with a dramatic acceleration in innovation, Brinker told Mi3, "This is the challenge of Martec's Law! Technology changes more quickly than organisations. There's no silver bullet to solve that challenge.

"However, companies can do two things to manage it. Firstly, be thoughtful and strategic about which changes they embrace. They can't do everything all at once. Next, develop technical and organisational agility, through open systems architectures and agile management practices. If you can adapt faster than your competitors, that's really all the edge you need to win.

"Empirically, the pace of change with AI is happening much faster than the SaaS revolution. Overall, we've been seeing a steady compression of the Gartner Hype Cycle — some AI capabilities are moving from the peak of hype to the trough of disillusionment, to the plateau of productivity in a matter of 12-24 months, in contrast to the 5-10 cycles of previous waves of innovation."

It's a reminder of our earlier reporting from 2023 about the potential disruption from generative AI: While GenAI offers breathtaking levels of time compression, turning months into minutes, there’s also the nagging risk that a dumpster fire of unfettered, unregulated, ungoverned disruption could derail organisations ill-equipped to handle massive acceleration from all angles, simultaneously.

That risk hasn't receded. It's accelerated.