Run, Fall, Run
There’s a familiar refrain echoing across agency boardrooms right now. Use AI to eliminate the repetitive, operational drag so teams can get back to thinking. It’s a clean idea. Almost too clean. At this year’s 4A’s Decisions 2026 in Boston, that idea started to feel less like a talking point and more like an operational mandate. The difference now is not belief. It’s proximity. The tools have caught up to the ambition.
According to Campfire’s Managing Director Pip Kolmar, agencies that treat this as a future-state conversation are already behind. “We’ve spent the last year saying we need to use AI to eliminate the repeatable work,” he said. “This was the first time it felt like there are no more excuses. The capability is here.” What emerged from the conference wasn’t a roadmap as much as a shift in posture, one that favors building over buying, speed over perfection, and systems thinking over siloed execution.
One of the more quietly disruptive ideas circulating the event floor was something being referred to as “vibe coding.” The premise is simple, but the implications are not. Instead of waiting on product roadmaps or external platforms, agencies describe the internal tools they need and use AI to build them. Under the hood, every advanced prompt is triggering layers of software development. The unlock is recognizing that and leaning into it. “There’s real development happening behind every complex AI query,” Kolmar said. “If you can clearly describe the tool you need, AI can help you build it. Not just as a prototype, but as something your team actually uses.” This doesn’t replace enterprise software. It fills the gaps those systems were never designed to solve, the hyper-specific workflows, the internal friction points, the small inefficiencies that compound across a team. For agencies that have historically relied on patchwork systems, this is less about innovation and more about control.
“The pace has completely changed. You can build something functional in a day, test it immediately, and know whether it works. Waiting is the risk now, not moving too fast.”
That shift in control pairs with an equally important change in pace. For years, agencies have been taught to roll out change cautiously, through pilot programs, phased adoption, and careful iteration. That model doesn’t hold up anymore. At Decisions, a new rhythm emerged: run, fall, run again. “The pace has completely changed,” Kolmar noted. “You can build something functional in a day, test it immediately, and know whether it works. Waiting is the risk now, not moving too fast.” This acceleration does not apply universally. There are still areas that require deliberation, like pricing models, talent structures, and governance. But when it comes to operational systems, speed is now the advantage. Agencies that cling to traditional rollout frameworks risk optimizing for stability in a moment that rewards momentum.
If speed is the new expectation, data is the new dependency. For years, centralized data infrastructure felt like something reserved for holding companies and scaled independents. That barrier is collapsing. Smaller agencies now have access to tools that allow them to unify performance data, planning inputs, contracts, audience insights, and internal documentation into a single environment. “Even small agencies are sitting on massive amounts of data,” Kolmar said. “If it’s fragmented, AI can’t help you. If it’s unified, you can start asking much more complex questions and getting meaningful answers back.” The implication is bigger than efficiency. It’s about decision quality. When large language models are querying across a unified data set, they move from being reactive tools to something closer to operational infrastructure, answering questions about best practices, supporting media planning decisions, and even automating pacing across channels.
Still, none of that works without discipline. For all the excitement, one theme grounded nearly every conversation: AI doesn’t fix bad systems. It amplifies them. “Your AI agents are designed to be confident,” Kolmar said. “If your data is messy or outdated, they won’t just give you the wrong answer. They’ll give it to you with conviction.” That raises the bar on internal operations in a way many agencies have historically deprioritized. Data hygiene, version control, taxonomy alignment, and documentation standards are no longer back-office concerns. They are prerequisites for making AI usable. Agencies that invest in clean systems will unlock exponential value. Those that don’t will automate confusion at scale.
“Your AI agents are designed to be confident. If your data is messy or outdated, they won’t just give you the wrong answer. They’ll give it to you with conviction.”
What ultimately came through from Boston was not a sense that AI will replace agency work, but that it will expose what parts of the work actually matter. “We have a deeply strategic team,” Kolmar said. “What holds us back isn’t thinking. It’s the systems around us. If we can offload the repeatable work, we get to spend more time doing what actually moves the needle for our clients.” That may be the clearest signal coming out of 4A’s Decisions 2026. The opportunity is not just efficiency. It’s focus. And for agencies willing to rebuild how they operate, that shift may be the most valuable outcome of all.
