There’s a dangerous tendency to think of AI as magic instead of infrastructure, with Matthew Rance

BY HAZEL BROADLEY, BEELER.TECH

Hosted by Rob Beeler and featuring Immediate Media’s Head of Commercial Data & Analytics, Matthew Rance, our inaugural publisher AI automation show-and-tell avoided the usual ‘AI will change everything’ keynote theater. Instead, we got to see a publisher sharing actual workflows.

This conversation is for our exclusive publishers and partners only community. Please ask in the community for the password to watch this video. You can also become a member of our community here.

AI strategy is no longer theoretical

For the past two years, publishers have largely talked about AI in abstract terms: opportunities, risks, legal concerns, SEO disruption, licensing deals, existential dread. Yes, they are all important. But they’re increasingly disconnected from the practical question every operations, revenue, and editorial team now faces: what are we actually doing with this stuff? Matthew’s demo answered that question with refreshing specificity.

Using the automation platform n8n, he built a fully automated AI-focused newsletter workflow that sources research papers, articles, and videos; scores and enriches them using LLMs; stores them in a database; and surfaces them through a lightweight editorial interface for final human review and publishing.

Plenty of publishers can already auto-generate newsletters. But what’s interesting about this is the architecture behind it: AI not as a replacement for editorial judgment, but as a system for scaling curation, enrichment, and operational efficiency.

That distinction came up repeatedly throughout the discussion. “I wanted to see if I could create a system which sustains a newsletter function with minimal input from me,” Matthew explained. “Theoretically, it only takes me five or ten minutes to sustain this flow… although I usually like to add a POV, which takes a bit more time.”

And that last sentiment may be the entire future of publishing condensed into one sentence. Because despite all the panic about AI-generated content flooding the web, the real opportunity for publishers is not to replace humans, but to reduce the amount of mechanical, repetitive, soul-draining work surrounding human expertise.

The internet today can do without another fully AI-written thought leadership article. What it does need is fewer editors manually copying URLs into spreadsheets.

The next competitive advantage is operational

Notably, AI discussions are shifting away from content generation and toward systems design. Publishers are discovering the challenge isn’t simply choosing an LLM, but orchestrating workflows between tools, databases, APIs, CMSs, analytics systems, editorial pipelines, and internal processes that were never designed to work together.

Or, as one attendee put it: “Everybody does one thing well. We haven’t found anybody that does everything.” That may become the defining operational challenge, and opportunity, of the AI era.

Most publisher stacks already resemble archaeological sites layered with years of acquisitions, workarounds, vendor migrations, and ‘temporary’ fixes that somehow became permanent. AI is now being dropped into that environment at high speed, often under executive mandates that sound suspiciously like: “Can you just AI this by next quarter?”

Rob certainly pushed back on that mentality, saying “the CEO who tells you to put AI in your processes should actually be helping from a corporate level setting out what our company is about. Otherwise, everyone’s just kind of going in their own direction.”

That’s a more important point than it may initially sound. Right now, many organizations are experimenting in isolation. Editorial teams are testing copilots. Revenue teams are automating reports. Product teams are building agents. Analysts are wiring together workflows in n8n, Zapier, or Make. Somewhere, inevitably, somebody has created a Slack bot nobody else knows exists.

Experimentation is good, but fragmentation is not. Without a broader operational strategy, publishers risk creating what Rob called a “Frankenstein monster of little AI things.” And he’s right.

AI literacy is becoming infrastructure

One of the most revealing aspects of the discussion centered not on prompts, but on data structures, transformations, APIs, schemas, orchestration, and process logic. In other words: the boring stuff. Which is precisely why it matters. The publishers gaining early advantages from AI are not necessarily the ones with the flashiest generative tools. They’re often the ones quietly developing organizational AI literacy around workflows and systems thinking.

Matthew repeatedly emphasized that n8n itself isn’t ‘the AI.’ It’s a workflow automation layer that becomes more powerful when connected to AI models. This is important because there’s currently a dangerous tendency across the industry to think of AI as magic instead of infrastructure. But infrastructure is exactly what this moment requires: reliable workflows, structured data, defined outputs, human oversight and operational guardrails.

At one point, Matthew described the importance of forcing AI outputs into strict JSON schemas because otherwise models “can go rogue.” That sentence should probably be printed out and taped to several thousand conference booth walls.

The publishers who win will be the ones who use AI best, not fastest

Matthew talked extensively about trying to codify his own ‘taste and discernment’ into scoring systems that evaluated research papers based on relevance, novelty, and clarity. But even after building the automation, he still manually reviews and selects final outputs.

That hybrid model feels important, and not because humans are inherently better at every task, because they aren’t (AI is already outperforming people in plenty of narrow operational functions). But because publishing value has never really come from production alone. It comes from judgment.

The problem is that judgment doesn’t scale neatly. That’s why so much of the current AI conversation feels unstable. The industry is trying to figure out which parts of publishing are fundamentally computational and which parts remain stubbornly human.

The publishers likely to emerge strongest from this period will not necessarily be the ones that automate the most aggressively, but those who understand where AI genuinely improves workflow efficiency, where human expertise still creates differentiation, and how to combine the two without losing either. Or put another way: the future belongs to publishers that use AI to remove friction, not personality.

👉Want to stay in the loop with what’s happening in the Beeler.Tech community? Subscribe to our newsletter. If you’re interested in attending or sponsoring a future event, you can explore our upcoming events here.