Last month I wrote about the company that raised 150 million dollars to run marketing with AI agents, and what it means for businesses like yours. The short version: the technology is coming downstream, and the businesses that win will be the ones with clean, connected data ready for it.
This month the industry put numbers behind that argument. The State of Martech 2026 report, from Scott Brinker and Frans Riemersma, is the most complete read on where marketing technology actually is. It is free and ungated, worth the download. Buried under the AI headlines is a warning almost nobody is talking about. It is worth your ten minutes.
Everyone is adopting AI. That is not the interesting part.
AI adoption is now everywhere. In the report, AI use rose across every single category of marketing work. Nothing declined. Commerce and Sales nearly doubled, from 28% to 49%. Data-related AI use climbed from 61% to 75%. Two years ago most of this was experimentation. Now it is embedded in how the work gets done.
So AI is no longer the differentiator. Everyone has it. The question that decides who benefits and who gets burned is what sits underneath it.
The warning under the noise
Here is the line from the report that should stop you. When an AI agent runs a customer interaction, it can only work with the data, content and tools it can actually reach. There is no human in the middle to quietly fix the handoff between two systems that do not talk, to guess what the customer really meant, or to cover for the product information that has not been updated since last quarter.
This is the same point I keep making, now confirmed by the industry data. AI amplifies what it sees. Feed it clarity and it compounds your growth. Feed it a Frankenstack of disconnected tools and it amplifies the fragmentation. You do not get efficiency. You get chaos, faster.
The layer everyone forgot to build
The report is blunt about where the real constraint has moved. It is no longer “can we connect these systems?” The tools to connect exist. The constraint is now orchestration and coherence. Which connections matter, who governs what flows through them, and, in their words, does this new thing fit into anything larger or have we just added another ungoverned node to the stack?
You can see it in what is quietly growing. The data integration category, described in the report as now as much an AI orchestration layer as an integration tool, grew 8% in a year. Their phrase is worth keeping: the connective tissue holding it together matters more when agents need to traverse it autonomously. That connective tissue is the orchestration layer. It is the thing AI sits on top of, and it is the thing most businesses have never actually built.
The split that proves the point
The most telling finding is where AI has taken over and where it has not. AI copy production sits at around 91% adoption. Writing is easy to hand over, because it needs little more than a prompt and some brand context. But the jobs that actually drive revenue, lead scoring and routing, email deliverability, personalisation, are still run by connected platforms, not standalone AI tools.
The report explains why in two words: data gravity. You cannot score a lead in a clever AI tool that does not know your pipeline. The valuable work depends on connected, structured data, which is exactly what a fragmented stack does not have. So the businesses racing to bolt AI onto the easy jobs are missing the point. The advantage is in the jobs AI can only do well when the layer beneath it is sorted.
And the part almost no one admits
One more number. 73% of organisations now have a formal generative AI policy, up from 52% two years ago. Progress, on the face of it. But separate research in the same report found only 8% are fully confident in their broader AI governance readiness. A policy is a starting point, not a finish line.
That gap is the drift I keep warning about. AI starts well anchored and useful, then drifts, quietly, until what it produces has stopped matching what your business needs. No alarm goes off. You just slowly stop getting what you wanted. The discipline that prevents it is not a policy document. It is keeping the AI re-bedded to the fundamentals of your business, on a foundation that is actually connected.
What this means for you, in plain terms
You do not need AI agents yet. You need the layer they will eventually run on, and that layer earns its money long before any agent is switched on. In practice it is one connected system doing five jobs: capture every lead in one place, profile each contact properly, engage based on behaviour, nurture the ones who are not ready, and convert when the signals say so, with sales seeing exactly what marketing knows.
This is the work we call getting AI-ready, and it is context engineering: making your orchestration layer fit, connected and true to how your business actually runs. It can be built on the budget and team you already have. Our clients typically see a return inside six months, because better follow-up and better timing convert leads that were already being paid for and then wasted.
The industry just told you where the game is heading. The smart move is not to add more AI. It is to fix the layer underneath it, before the drift, and your competitors, catch you out.
Get ready while it still counts as early. If you want a clear-eyed view of where your gaps are, start with our Marketing and Sales Assessment, or book a free call with me and we will talk it through. No jargon, no pressure.
Source: State of Martech 2026, Scott Brinker and Frans Riemersma, chiefmartec and MartechTribe. Free and ungated: content.martechday.com/state-of-martech-2026.pdf