Publishing has an AI problem — but not the one you might think. 

The problem isn’t a lack of AI. Every platform in the market has some version of it: headline suggestions, auto-tagging, writing assistants. The problem is that most of it was designed as an afterthought. A layer added on top of workflows that were never built for it. Something you click when you remember to, configure when you have time, and gradually stop using because it never quite fits. 

At Atex, we think that’s the wrong starting point. 

The direction we’re moving with our solutions isn’t about adding AI features to an existing platform. It’s about rethinking what an application should become when AI is treated as a native part of the workflow, not a button, but a collaborator. 

From features to agents

The concept we’re building toward is what we call the Agentic Newsroom: a publishing environment  where AI doesn’t wait to be activated, but operates continuously alongside the team, handling specific tasks with a defined scope and reporting back when there’s something worth a human’s attention. 

The distinction matters. A feature responds when you ask. An agent works whether you ask or not. 

We’re developing this vision across Atex’s three business units: Content, Advertising, and Subscription. In each area, the same principle applies. Identify the tasks that are mechanical, repetitive, and time-consuming, and design agents that can own them reliably, so that journalists, commercial teams, and audience managers can focus on what requires human judgment. 

Content: intelligence inside the editorial workflow 

Inside MyType, the agents we’re working toward are designed with full awareness of editorial context. They’re not generic AI tools plugged into a CMS. They understand publishing priorities, content logic, and the rhythm of a newsroom. 

The directions we’re actively developing include a News Scout that monitors sources continuously and delivers structured briefs to the right journalist before they’ve had to search; a Fact Checker that scans every article before publication and flags unverified claims with source references; and a Newsletter Curator that assembles personalised editions for each subscriber segment based on real engagement data rather than manual selection. 

The common thread is that none of these agents interrupt the workflow. They belong inside it. 

Advertising: removing friction from the commercial operation

Commercial teams lose time in predictable places. Asset production starts from a blank page when it could start from a brief. Compliance checks happen at the end of the process, when catching a problem is most expensive. Prospecting relies on experience and instinct when the data to do it systematically already exists somewhere in the system. 

The agents we’re designing for the Advertising business unit are oriented around these three pressure points: accelerating the path from brief to finished asset, making compliance a continuous background process rather than a gate at the end, and turning historical campaign and audience data into structured sales intelligence before a conversation even begins. 

The goal is not to automate the commercial relationship. It’s to ensure that the time sellers spend with clients is preparation and strategy, not administration. 

Subscription: from monitoring to intervention

Audience teams rarely lack data. What they lack is the capacity to act on it quickly enough to make a difference, and that’s where the agentic approach has the most to offer for the Subscription business unit. 

The agents we’re designing in this area work across the full subscriber lifecycle. Early in the relationship, the focus is on identifying which anonymous and registered users are most likely to convert, and reaching them with the right offer at the right content moment, without requiring a team to manually define rules for every possible scenario. Later in the lifecycle, the focus shifts to retention: detecting the behavioural signals that precede cancellation early enough to intervene, and deepening engagement with high-value readers through more personalised experiences. 

Underlying both is a shift in how audience intelligence gets used. Rather than producing dashboards that describe what happened last week, the agents we’re building are designed to answer questions in real time and surface the actions worth taking now. 

Why this matters now

Publishers today are being asked to do more with the same, or fewer, people. The pressure is structural, not cyclical. Agentic workflows are one of the most credible responses to that pressure, because they address the right problem: not replacing editorial judgment, but removing the mechanical work that crowds it out. 

This is the platform MyType is becoming. Not all of this exists today in its final form. We’re building it deliberately, with real newsrooms and real workflows as the reference point. But the direction is clear, and the foundations are already in place. 

If you want to understand where we’re taking this and what it could mean for your organisation, we would love to have that conversation. Let’s connect now!