The Human in the Room

Every enterprise AI content strategy conversation I've had in the last year starts the same way - companies want AI everywhere. And they trust it almost nowhere.

I spent a decade building Adobe Experience Manager Assets - the DAM at the foundation of how some of the world's largest brands manage, govern, and distribute content at scale. I watched the AI conversation shift from "can we auto-tag assets with Sensei?" to "can we have AI agents orchestrate our entire content supply chain?" And I watched enterprise customers get excited about the second question while quietly panicking about what it actually means.

That panic is one of the most legitimate signals I encountered in 10 years. It's worth taking seriously.

The Model I'd Defend

The model I'd defend in any room -

AI as enforcer of human-defined guardrails. Not author of brand expression.

This isn't a limitation to apologize for. It's a design principle. AI's job in the content supply chain is to make it as hard as possible to ship off-brand content - to flag, block, and reroute at a speed no human team can match. The creative expression, the brand judgment, the "does this feel like us?" question - that stays with people.

What agentic AI changes is what those guardrails can actually do. An AI agent that monitors every asset across every channel and flags brand violations in real time is extraordinarily valuable. An AI agent that decides what the brand should say next quarter is a liability.

The Real Unlock

The companies that will win the next phase of the agentic content supply chain aren't the ones deploying the most agents. They're the ones that do something much harder first - translate their brand from a human-readable document into something a machine can actually enforce.

That means going beyond "warm and joyful" and defining - precisely, specifically, measurably - what on-brand looks like. Approved visual signatures. Tone patterns derived from your best-performing content. Rules specific enough that a system can flag a violation without anyone having to review it. No brand has fully cracked this at scale yet. The ones that do it first will have a real, lasting advantage, because the AI layer only works as well as the human layer beneath it.

This isn't a technology problem. It's an organizational one. MIT research found that 95% of enterprise AI pilots fail - not because the models underperform, but because of poor integration into how teams actually work, and the hard internal effort of defining what success means. Brand AI is no different. The model is the easy part. The brief is the hard part.

The Forrester Wave named AEM Assets a Leader again in Q1 2026. I'm proud of that. But the more interesting race over the next 18 months isn't which DAM wins analyst recognition - it's which brands do the organizational work to define themselves precisely enough that an AI can honor it.

That's the unlock. And most companies haven't started yet.

From I Spent 10 Years Building Adobe's DAM. Here's What the AI Content Supply Chain Hype Gets Wrong. (opens in a new tab), June 24, 2026.