Good question. Here's our take👇 While most AI in marketing stays on the surface by polishing subject lines or churning out copy, the real transformation isn't just about automating tasks; it's about automating strategy. That's what Active Intelligence is built to do. Instead of guessing, it analyzes your performance data to tell you exactly what to do next. You won't get generic best practices or recycled strategies. Instead, you get precise recommendations based on what is actually moving the needle in your account. ➡️ It's why 83% of surveyed customers hit ROI before the end of their first year.

Every marketing executive I know has already been told to “go do AI”, and some of them are wondering how, or if they reliably can, given the state of their data. They’re not wrong. Most marketing data environments weren’t built to hold up under pressure. Different systems define the same thing in different ways. Campaign structures drift over time. Content can’t be consistently tied back to performance. Reporting looks clean, but doesn’t hold up when you actually try to make decisions. So the instinct is to work around it. Stand up a pilot. Create a clean dataset on the side. Test AI in a controlled environment where everything behaves. And it works! For a minute. Then the question becomes how to scale it. That’s where most teams stall, not because AI isn’t working, but because the system it needs to plug into doesn’t. This isn’t a tooling issue. It’s an operating model issue. If planning, content, activation, and measurement aren’t connected through a consistent data model, AI just amplifies the inconsistency faster. You don’t need another workaround. You need the core system that runs your marketing organization to actually hold. A clear system of record. Structured metadata that doesn’t change depending on who’s using it. Ownership that’s defined and enforced. Workflows that connect strategy all the way through to measurement. Until that exists, every AI investment is going to feel harder than it should, and scaling it will keep slipping. You don’t need perfect data, but you do need a system that can be trusted. Tell the truth. Is your marketing data really ready for reliable AI at scale?

Most marketing AI in 2026 is doing the easy part. The team is still doing the work. It scores leads. Someone routes them. It recommends a campaign. Someone launches it. It drafts a report nobody reads until someone rewrites the executive summary. Then this week, The Trade Desk announced Koa Agents. AI that plans, buys, optimizes, and measures media across the open internet, without the manual step-by-step setup that used to define a media buy. Stagwell is the first agency to roll it out. Jeff Green's framing was sharp: "We're increasingly focused on powering the advertising companies of the future (those who make things more efficient, not those who exploit inefficiencies)." That line is also a test for every AI tool already in your stack. PitchBook calls this workflow wrapper risk: software that facilitates a process without owning the outcome. The gap between AI that informs and AI that executes is the most important distinction in B2B marketing right now, and most buyers haven't named it yet. A quick test for anyone auditing their own stack this week: → Does this tool make the decision, or present one? → Does it execute the workflow, or just describe it? → What does my team still have to do after the "AI" part is done? If the answers leave a lot of manual work on the table, it's a wrapper. What's the one tool in your stack that actually does the work, and what would change if it didn't? #AIStrategy #AgenticAI #WorkflowAutomation

New research from Stensul surveyed 321 enterprise marketing teams. Here's the finding that stopped me: 53% of organizations lack comprehensive AI governance for marketing campaign creation. At the same time, AI is the number one organizational mandate for 38% of those same teams. Nearly 90% of the organizations without governance reported at least one campaign error in the past year. And the most common consequence? Heavier review processes and increased scrutiny. Which means the speed advantage they adopted AI to create is being canceled out by the errors AI is producing without guardrails. I've said it before in different ways but this data puts a hard number on it. AI accelerates whatever you put in front of it. If you put a governed, well-structured process in front of it, you get speed. If you put a disorganized one in front of it, you get faster errors and slower approvals. The teams winning with AI right now aren't the ones moving fastest. They're the ones who built the infrastructure to move fast safely. What does your AI governance actually look like today?

This is one of the best thing I've read in a while on AI. Here's a quick summary of what's going on. - We have a handful of horizontal AI OS (OpenAI, Anthropic etc.). They will own general consumer and business workflows - Some vertical AI operating systems will win in industries where context, regulation, and services matter. Think health, law, construction, finance etc. - The rest becomes "headless", which doesn't mean a bad business, but they are no longer the front door. Massive implications on their GTM and branding. - A new layer of AI-native service agencies basically absorbs the messy industries/use-cases/edges-cases where exceptions are the norm, where software alone doesn't work. You have more details on the 7 ingdoms in the article (which I strongly encourage you to read). Some initial thoughts on what this means for branding and GTM: - If your product becomes headless, you lose the customer interface, so basically the brand surface shrinks. How can marketing compound? - Distribution gets disintermediated by whoever owns the conversation above you, and you don't see it. - The new moat is not features (we've been saying this for 20 years), but it's industry trust, the regulatory depth, the specificity of your services, or a brand strong enough to be picked as the default operating system of a given vertical. - GTM shifts from a "category leader" framing to an "operating system" ("powered by"), with probably less product marketing and more industry presence. Interestingly I think the question for most B2B founders isn't about the features, but "which kingdom am I actually in" and "is my brand strong enough to be relevant and visible". This is a quick, imperfect summary of Luke Sophinos's article, pls check out his full map. Link is in the comments

Marketing's "gold rush" toward AI. Marketing departments are racing to "do AI," yet many are not sure what they're racing toward. Pilots are launched. Tools are licensed. Prompt experiments abound. Six months in, the CMO is asked what has changed — and the honest answer is that we're faster at the work we shouldn't have been doing in the first place. That's not an AI problem. It's a readiness problem. Here’s a three-point readiness approach that marketing leaders can use: how ready is our marketing function to use AI across how it makes decisions, how it operates, and what it produces? *How does the department decide? Are AI use cases mapped to the top strategic priorities, or is the team chasing efficiency without effectiveness? If this is weak, AI will accelerate the wrong work. Pause activation. Align first. *How the department works. Is the infrastructure provisioned but not adopted? Does the team have the literacy, the prompt skills, and the workflow design to operationalize AI in daily work? This is the most common gap. Tools exist. Adoption doesn't. *What the department produces. Can you control what AI puts into the market? In financial services and healthcare, AI-generated content that fails to meet FINRA 2210 or HIPAA isn't an efficiency problem. It's a regulatory incident. Governance, brand-voice guardrails, human-in-the-loop review, and audit trails must exist before scale, not after. The shape across those three helps you make informed decisions: *Strong on decisions, weak on workflow → you know what to do but can't execute it. *Strong on workflow, weak on decisions → you're doing the wrong work faster. *Weak on output control in a regulated industry → you're one piece of non-compliant content away from discrediting the entire initiative. A CMO can act on that. A CIO respects it. A CFO can fund a plan to fix it. The marketing departments that succeed with AI aren't necessarily the ones with the most tools or the largest budgets. Instead, they view AI as a comprehensive system. They honestly assess all three dimensions and prioritize fixing the weakest one first — not the easiest. #AIinMarketing #MarketingStrategy #FinancialServices #CMO #AIGovernance

Some decisions in an Agentic #Marketing system cannot belong to the AI. Not because AI cannot execute them. Because your organisation cannot afford for AI to own the outcome if it goes wrong. Over the past week, I named five sovereign decisions the AI Agentic CMO cannot delegate: 1- Who speaks first when it goes wrong? 2- What are you willing to trade for growth? 3- Who draws the line the model cannot see? 4- Who signs off when the rule is not enough? 5- Who has the authority to stop what works? None of these are technology questions. They are governance questions. Leadership questions. Accountability questions. The uncomfortable reality is this: Most organisations have already deployed systems making decisions at scale before defining who ultimately owns them. The AI is not waiting for your governance model to catch up. It is already acting. And when something eventually lands publicly, legally, regulatorily, or reputationally, the organisation will not ask what the AI decided. It will ask who allowed the system to decide it. That is the real mandate emerging inside modern Marketing leadership. ❌Not content production. ❌Not automation orchestration. ❌Not prompt engineering. Governance architecture. This conversation is becoming one of the defining Board-level discussions of the next three years. If this series raised questions inside your organisation, let’s discuss them properly. No pitch. No pressure. Just a serious conversation about what AI governance actually looks like when systems begin acting on behalf of brands at scale.

Everyone's talking about building a Gen AI marketing workbench. Very few are actually building one. After years in marketing transformation — and the last several deep in AI-led operations across automotive, pharma, telco, and retail — I've seen the pattern clearly now. Most enterprises are layering Gen AI onto fundamentally broken foundations. They're automating chaos. Here's what a truly AI-ready marketing workbench actually looks like in practice: 1. Data is unified before AI touches it. Not "good enough." Unified. Audience signals, content performance, channel attribution — all flowing into a single, governed layer. AI is only as intelligent as the data architecture beneath it. 2. The workflow is reimagined, not digitised. The biggest mistake I see? Replicating legacy approval chains inside a shiny new platform. Genuine AI readiness means rethinking who does what — and letting the machine handle the rest. 3. Personalisation operates at the variant level, not the campaign level. True AI-personalised content experiences don't produce one campaign with five versions. They produce hundreds of variants, dynamically assembled, continuously optimised. If your workbench can't do this in production, it's a pilot — not a capability. 4. Human oversight is built in, not bolted on. Governance, brand safety, and compliance can't be afterthoughts. The most mature workbenches I've seen embed review and control into the AI loop — not as a brake, but as a feature. 5. It's measured by business outcomes, not output volume. Content velocity is a vanity metric. The question is: did personalisation drive conversion? Did AI reduce time-to-market for priority markets? Tie every capability to a commercial signal. The organisations getting this right aren't the ones with the biggest AI budgets. They're the ones who started with the hardest question: what does marketing transformation actually need to deliver? That's where the work begins. What would you add to this list? I'd love to hear what's working — or not — in your organisation. #MarketingTransformation #GenerativeAI #AIWorkbench #IntelligentOperations #MarketingOperations #EnterpriseAI #ContentAtScale

I asked 160 CMOs why AI is not scaling Only 7% said tools, the rest lack use cases That's the gap no one’s talking about. Teams have the licences. They've done the all-hands. But six months in, a handful of people are achieving genuine gains Everyone else is still dabbling The decision paralysis is real: What do we automate? What will move revenue? What should we keep human? The teams pulling ahead aren't using more tools. They're decisive about one call: Which are the highest leverage use cases to automate? That decision is what separates execution from experimentation. Most marketing orgs haven't made it yet. So I built a decision engine to make it faster. 110 workflows mapped to revenue impact across 11 marketing functions. ↳ Pick your bottleneck. ↳ Choose your path: AI assisted or agentic. ↳ Instantly see the tools and the guardrails others are using to fix it. Which part of your marketing is stuck in AI experimentation right now? Comment USE CASE below and I'll DM you the playbook.

I was asked at the WWD CEO Beauty Summit where to start with AI in creator marketing. My answer: get ahold of your data and normalize it. Why? Your data is your future IP. Though most people and organizations think of AI as a way to automate and scale manual processes, unless your data allows you to tell apart the good, the bad and the ugly in your creator marketing practice, automation will just accelerate and scale mediocrity. By bringing together your data from spreadsheets and decks scattered across your agencies and subsidiaries, and structuring that data, you can let it tell the story of your business, extract best practices, and intelligently feed agents that will let you scale the impact of your program without scaling its cost. Even if your organization is not ready yet to jump on the AI rocket train, preserving your data now and organizing it will allow you to train your AI on it in the future.