Newsrooms worldwide are currently grappling with a fundamental strategic crossroads: is the future of journalism about serving autonomous agents with high-velocity intelligence, or about fostering deep human connections and community?
At Media Party New York 2026, Joanna Geary, Head of Content and Audience at Bloomberg, addressed this tension through the lens of her recent work on the front lines of agentic AI. In her keynote “The Big Split,” Geary moved beyond theoretical predictions to analyze how these two paths—intelligence for machines and community for people—are reshaping the value of news in real-time.
The Shift to Agentic Workloads
For Geary, the “agentic” future is an operational reality. She defines agentic AI as systems that are not just reactive, but goal-oriented. As Bloomberg CTO Shawn Edwards puts it, these are systems that can be given a goal, build a plan, and carry it out autonomously.
With Bloomberg’s launch of its primary agentic AI product this week, Geary argues that this technology will become the dominant way customers interact with information. This transition forces newsrooms to reconsider their core job: it is no longer just about storytelling, but about providing the precise information required for autonomous systems to make smart decisions in complex environments.
Deconstructing the "Human Compression Protocol"
In a striking analysis of the medium, Geary described the traditional news story as a “human compression and registration protocol.” While the article format has been refined over centuries as a perfect vehicle for human consumption, Geary argues it is “completely unnecessary” for an AI agent.
- The Great Reversal: In 2014, the industry trend was “data-to-narrative” (automated reporting). In 2026, Geary’s focus is the inverse: turning narrative back into data.
- Bubbles of Data: Agents do not require the “inverted pyramid” or a catchy lead. They need “structured, operable knowledge” they can reason over. In this landscape, news becomes a “bubble of data” where the primary values are authority, signal, and specific utility.
- The Non-Human Reader: Geary noted that the first reader—and the first actor—on informational news is often no longer human. Success is no longer measured by “time on site,” but by whether the data helped a client or an agent make the correct decision.
The "Archipelago of High-Trust Communities"
On the other side of the split, Geary examined the human element through a concept articulated by Tony Haile: the emergence of an “archipelago of high-trust private communities.” This refers to a specific phase of our digital lives where human connection is moving into smaller, more specific, and higher-trust spaces built for collaboration. Geary emphasized Haile’s bottom line: this “archipelago” is exactly where journalism has the best shot at building a sustainable future.
While some technologists suggest this niche could become a “minority of the web” where humans coalesce as a reaction to pervasive AI agents, Geary sees it as a profound opportunity. It is in these spaces that journalists can act as conveners, tapping into a community of readers who possess intelligence and information that no algorithm—and no reporter working alone—could ever access.
That relationship—the post-publication relationship between journalist and reader, and the pre-publication relationship between reporter and source—is the information no one else can get. That is the hard work.”
The Synthesis: Community as the Intelligence Pipeline
The most critical takeaway from Geary’s keynote is that the “Big Split” is not a binary choice. Instead, intelligence delivery and community are inextricably connected.
For agentic systems to remain useful and competitive, they require unique, high-fidelity information that cannot be scraped from the open web. This “information edge” is found within the high-trust communities where humans collaborate and share specialist intelligence. In this model, the community acts as the intelligence pipeline for the agent.
Conclusion
The news organizations that survive the agentic revolution will be those that master both poles of the split. They must sharpen the technical tools required to serve autonomous agents with structured data, while simultaneously deepening the human skills required to convene the communities that provide the world’s most valuable, non-synthetic information.
In 2026, the target is no longer just the “audience,” but the community that fuels the intelligence.
Don’t miss our full Media Party New York 2026 recap: three days of conversation, collaboration, and real action. All in one read!
