Brand Intelligence is Adobe’s most ambitious content governance play yet, and after sitting down with the product lead at Adobe Summit, I think it’s worth paying attention to.
There’s a problem sitting quietly at the centre of most large marketing operations. It’s not a shortage of content ideas, or even production capacity. It’s the review and approval loop: the bottleneck of senior creative directors, brand managers and legal teams checking that everything going out the door actually looks like the brand it’s supposed to represent.
Adobe is betting that Brand Intelligence, announced at Adobe Summit in Las Vegas last month, can take a meaningful chunk of that burden off human shoulders. I was at Summit as a guest of Adobe, and got the chance to sit down with Aaron Finegold, Head of Product Marketing for Firefly Enterprise, to go deeper than the keynote presentations.
From guidelines document to AI learning system
Every brand has a guidelines document. Most are a PDF document somewhere in a shared drive, get updated annually if you’re lucky, and are selectively interpreted by whoever happens to be producing content that week. They are static artefacts trying to govern a process made increasingly automated by the use of AI.
Brand Intelligence is Adobe’s attempt to replace that document with something that learns. Rather than rules, it ingests signals: approved assets, rejected assets, review cycle feedback, annotations, editorial corrections. Over time it builds an evolving model of what the brand actually is in practice, not just what the guidelines say it should be. This is similar to how AI chatbots are trained to be accurate and personable, using a technique called RLHF (Reinforcement Learning from Human Feedback).
Finegold described to me the core problem: “A brand is so much more than a logo. But if the logo is wrong, the whole thing’s wrong. But also if the logo is right, the whole thing can still be wrong. That’s the whole thesis of Brand Intelligence.”
What he points out is that brand compliance should be more than a checklist. Moving from static rules to a continuously learning system is a major shift that makes sense given the volume of content marketing teams are now expected to produce.
Brand Intelligence inside a bigger platform
Before getting into the Skills themselves, it is worth understanding where Brand Intelligence sits. It is one piece of a significantly larger launch: Adobe CX Enterprise, a new suite spanning brand visibility, customer engagement, and content creation, all underpinned by the Adobe AI Platform.
Within CX Enterprise, teams can build agentic workflows that connect across their martech stack. A demo for Dick’s Sporting Goods showed a Catalog Agent optimising product metadata for visibility inside AI interfaces like ChatGPT, running alongside an agentic CMS and a Brand Concierge chatbot combining real-time pricing with individual customer profiles.
What surprised me from Adobe was how composable they have made it. Adobe knows customers do not want multiple agent platforms, so they have made CX Enterprise agents deployable anywhere. If your organisation runs Microsoft Copilot as its central AI home, your Adobe marketing agents can surface there. There is also a dedicated CX Enterprise Coworker interface for teams who want one, similar in concept to OpenAI Codex but built specifically for marketing workflows.
Three skills, and where they show up
Brand Intelligence itself is structured around three capabilities Adobe calls Skills.
The first Skill is Instruct to Assemble: a system that pulls approved elements (e.g. a hero asset for a campaign) from a digital asset library and composes them into new assets (such as localisation), guided by the brand model. The first place Adobe is making this available is in Figma, which is a great demonstration of the composibility approach they have embraced.
The second Skill is Validate, and this is where the composability story gets interesting. Validate checks assets for brand compliance, but it is not locked inside a single product or workflow. It runs automatically within Instruct to Assemble, catching and fixing the system’s own errors. It can operate as a node inside the Firefly Creative Production Workflow Builder. It surfaces in Workfront and Frame for review and approval workflows. And it can be accessed headlessly, meaning it can be called from essentially any AI-friendly app a team is already using.
It also means Validate can check assets the AI system has assembled, and also check assets that humans have created independently.
The third Skill, Predict Engagement, generated a lot of excitement in the room. In the demo I saw, it ran simulations against synthetic audiences to forecast how a campaign would perform – before any media spend was committed. It is currently in private beta though, so how well those synthetic audiences reflect real-world behaviour remains to be proven.
Robots with an eye for design
The headline case study at launch is Xfinity, Comcast’s consumer brand, which has been working with Adobe to embed Brand Intelligence into its marketing workflows. Xfinity’s Chief Growth Officer Jon Gieselman described the impact as allowing teams to “spend less time managing work and more time crafting the standout storytelling that defines the Xfinity brand.”
The position that Adobe is taking is not that AI replaces creative thinking. It’s that AI absorbs the compliance and production overhead so that human creative energy can go where it actually counts.
When I asked Finegold what customers were most excited about, his answer was: “Relieving the bottleneck they’ve had in human review and approval. The number of enterprises that tell me they’re putting a full time executive creative director simply on checking co-branded partner marketing.”
But there is still a significant distance between announcing an agentic platform and an organisation actually operating one well. That is a people and process challenge as much as a technology one, and in my experience it is where most implementations succeed or fall short.
The technology here is genuinely interesting. The organisations that get the most from it will be the ones who are clearest on what still needs a human hand.
Tim O’Neill is co-founder of Time Under Tension, Australia’s first generative AI experience agency. He attended Adobe Summit 2026 as a guest of Adobe.
