Can You Automate Good Taste?
The Agentic Shift: Field Notes from a Creative at Web Summit Vancouver 2026
A 5-minute read for brand managers, marketers, and strategists.
I arrived at Web Summit Vancouver in May 2026 without the obvious credentials. I'm not an engineer, I don't raise venture rounds, and I don't have a repository to show. I come from the creative side. And still, walking into the Vancouver Convention Centre, I was clear about one thing: I wasn't in the wrong room. I was, perhaps, in the only right one.
This isn't the chronicle of someone who decided to "pivot into tech." It's the opposite. I went because I suspected the most important conversation about creativity was no longer happening in design studios or agencies, but in the hallways where the direction of artificial intelligence is being argued out. More than twenty thousand people from a hundred countries filled the convention center, and the theme threading through everything was AI's transition "from euphoria to deployment." I wanted a holistic view, not a change of guild. I came back with a notebook full of notes that, on re-reading, tell a single story: the biggest disruption of this era isn't the algorithm. It's how we choose to treat each other.
Why a creative needs to be in the room
For years we were sold a clean frontier: on one side the makers (code, product, infrastructure) and on the other the tellers (brands, stories, experiences). That wall is coming down. Tomorrow's software is, literally, today's prompt. Syntax stopped being the barrier to entry; code became fluid, rewritable across multiple languages, and the focus shifted from how something is built to what human impact it generates.
For someone in my craft, this isn't a threat — it's an invitation. If technical execution gets commoditized, what becomes scarce isn't the capacity to produce; it's the judgment to decide what's worth producing. I went to Vancouver to confirm that intuition. I confirmed it faster than I expected.
The agentic era is no longer theoretical
The concept that dominated every conversation was agentic AI. The difference from what we'd been using is substantive, not cosmetic. We've moved from passive tools — reactive copilots waiting for instructions — to systems that assume specific tasks, take control of segments of the workflow, and operate autonomously.
The architecture that came up most often was the multi-agent ecosystem: an "orchestrator" agent that receives an objective and decomposes it, distributing the parts among specialized agents that communicate with each other, correct themselves in feedback loops, and resolve the problem without asking permission at every step. Analysts describe this as moving from "information delivery" to "autonomous task execution" — with orchestration as the new control center of enterprise strategy.
Honesty is in order: not everyone is buying the enthusiasm. Serious voices — at IBM, for example — point out that "orchestration" is something programming has been doing for decades, and that part of the narrative is just slapping a buzzword on something old. The numbers invite humility too: in a 2025 Deloitte survey, only about a quarter of leaders felt they had mature capabilities with agents, against a large majority who had them with basic automation. We're at the beginning, not the summit. But the direction is unmistakable.
What interested me as a creative wasn't the technical plumbing. It was where the human sits in that diagram. And the answer, in the best presentations, was always the same: the human sits in strategic intervention. Not supervising every output, but directing the meaning of the whole.
Context Beats Math: Solving "The 80% Problem
A language model, by default, hands you a generic approximation—let's say 80% of a good answer. It's correct, it's competent, and it's essentially the exact same answer your competitor is getting.
The remaining 20%—the part that decides who wins—doesn't come from the model. It comes from the context you feed it.
That's why a small company with deep context about its market can beat a giant company with thin context. The competitive advantage is no longer the size of your computational muscle; it's the richness of your input.
Data Isn't Just Math—It's Humanity
Data is words, emotion, history, desire, and culture. It is multidimensional context. To train AI models that resonate in a human way, you have to feed them humanity, not just spreadsheets.
This completely reorders the map of who matters in enterprise tech. "Strange" data—the unfiltered pulse of a Reddit community, the unexpected crossovers between public databases, or GitHub's collaborative methodologies—becomes absolute gold. Not because it's clean, but because it's genuinely human.
Activating "Creative Data" Across Three Fronts
The question for modern brands isn't only what data you have, but how you activate it across your entire ecosystem. To bridge this gap, a business must orchestrate three distinct layers:
The Data Team: Detects the hidden pattern or anomalous consumer behavior.
The Creative Team: Translates that mathematical pattern into an emotional, visual, human narrative.
The Organization: Deploys AI to personalize and scale that story to every single customer in real time (think: a "Spotify Wrapped," but custom-built for your B2B clients).
The Blueprint: Data without narrative is noise. Narrative without data is decoration. The true corporate value lives in the bridge—and that bridge is built by a strategic profile that speaks both languages.
