For the last several years, “growth stack” has meant a shelf of platforms: a CMS, a personalization engine, a CRM, an automation tool, maybe a CDP if the budget stretched that far. Buy enough of them, integrate what you can, and hope the whole thing adds up to something coherent for the customer on the other end.
That model isn't wrong, exactly. It's incomplete. The platforms you already have — Adobe Experience Manager, Adobe Target, your CRM — were built to do specific things well. What's changed is that an AI orchestration layer can now sit across those platforms, connecting what they know and acting on it, instead of leaving that integration work to whichever team has the bandwidth this quarter.
That's a real shift in what a marketing stack can do. It's also one of the most consistently oversold ideas in the market right now, so it's worth being precise about what's actually new and what's still marketing.
AI's role in a marketing stack used to mean recommendation engines and predictive scoring bolted onto existing platforms — genuinely useful, but bounded to whatever that one tool was built to do. What's changed is the emergence of a governed AI orchestration layer that sits across your systems: AI that can read from your CRM, act inside your CMS, and route information between them, using plain-language instructions instead of custom point-to-point integrations for every connection.
Concretely, that looks like: a lead arrives on your site, an AI agent evaluates it against your criteria, and a governed connection updates your CRM and routes it to the right rep — without a person manually touching any step in between. Or: content that used to require someone logging into AEM to publish, schedule, and wire up now happens through a single instruction, because the orchestration layer handles the mechanical steps a person used to do by hand.
That's meaningfully different from “add a chatbot” or “turn on an AI feature inside your existing platform.” It's a layer that makes the systems in your marketing stack talk to each other more intelligently than they were individually designed to.
None of this replaces the platforms doing the actual work. AI orchestration doesn't personalize your website — Adobe Target does that, and it still needs to be properly configured and fed real data to do it well. AI doesn't manage your content — AEM does, and it still needs sound information architecture underneath it. The orchestration layer makes the handoffs between these systems faster and more automatic. It doesn't substitute for the platforms being set up correctly in the first place.
This matters because it's the most common place stack conversations go wrong. Adding an AI orchestration layer on top of a personalization platform that isn't properly wired into your CRM doesn't fix the underlying disconnection — it just automates around it, faster, in more places, with less visibility into what's actually happening. Orchestration is a force multiplier for a healthy stack. It's not a substitute for one.
The organizations getting genuine value from AI orchestration tend to start in the same place: a specific, high-friction handoff between systems that currently requires a person to bridge manually.
In each case, the value isn't a new capability the platforms didn't have. It's removing the manual bridge between platforms that were never talking to each other automatically.
The version of this that lasts isn't a system you depend on a vendor to run forever. The connections between your systems should be governed and secure, but the capability to use them — the “skills” your team runs on top of those connections — should be something your own people operate, in plain language, without needing to become engineers to do it.
That's also, per recent research on AI implementation approaches, the version that's more likely to work: AI capability built with an experienced partner succeeds roughly twice as often as the same effort attempted entirely from scratch internally, because the patterns that work have already been learned elsewhere. But the goal of bringing in that expertise should be to leave your team more capable, not more dependent — reusable connections and trained people are the actual deliverable, not a system only we can maintain.
The right starting point isn't a stack-wide AI orchestration rollout. It's the single workflow where the disconnection between your platforms is costing you the most time right now — usually somewhere between your CMS, your personalization platform, and your CRM, wherever a person is currently the one making systems talk to each other.