Are you still a publisher if you don't publish?

Adriana Whiteley, Director, FT Strategies, the consulting arm of the Financial Times, draws on the firm’s “Resilient Content Businesses in the Age of AI” series. She is a participant in The Definitive AI Forum for Media, Information & Events 2026 and here poses vital questions for traditional media as GenAI turns information into content – and content itself into utility.

**“Value migrates from the output to the underlying information”**

For two centuries, the publisher’s job was to stand between people who knew things and people who wanted to know them. Generative AI moves the value away from that middleman role and towards the two ends of the chain: the proprietary information and expertise from which answers are derived, and the interfaces that sit closest to the user when those answers are delivered.

Our resilience testing across the archives of B2B and B2C publishers shows that much of what a typical title produces – and sometimes most of it – can already be replicated by decent prompting and a fact-check.

AI drives competition from both supply and demand

On the supply side, AI reduces the cost and the skill needed to produce competent content at scale, enabling automation of much of what fills B2B publications today: update-type news from official sources, press-release reworkings, evergreen explainers.

That expands the competitive set. Every brand, consultancy, trade body and specialist AI application can now produce a passable explainer or industry briefing. For many, content is just a marketing cost. Publishers are therefore competing for attention and talent with organisations that do not need to monetise their outputs.

On the demand side, users can increasingly aggregate and summarise content themselves. Corporate users, or the agents acting on their behalf, can go straight to primary sources such as regulatory filings and corporate announcements, then combine, personalise and contextualise them against internal proprietary data. A job that used to be performed by an article can now be done by data feeding directly into an internal process, with no publication in between.

If AI can write the article, what’s the editor for?

AI can increasingly replace one of publishing’s traditional editorial functions: turning facts into readable text. But it does not remove the effort of establishing those facts through original reporting, validation and curation.

Some of this work will remain “artisanal”: cultivating relationships with human sources, deciding what is important, or finding creative ways to present complex information. Other functions will become more “industrial”, focused on defining rules to codify style and output formats, establishing processes for validating and normalising data, and setting the parameters within which AI systems operate.

A journalist might develop sources and uncover information about a company that is not available elsewhere. But once the facts have been verified and organised, AI can adapt the raw material to different needs and audiences. The same content can generate a news story, a company briefing, a chart, an audio summary, an explainer video or an alert, according to editorially defined rules. The editor still decides which information can be trusted, how it should be interpreted and how machines are allowed to use it.

This is likely to create more hybrid editorial, product and data roles. A few newsrooms are already writing job descriptions for “prompt editors”, responsible for defining how systems turn underlying data into different products and formats.

As production becomes automated, editorial accountability itself can become part of the product, especially when information feeds directly into corporate decisions, or where provenance has economic value – when a claim must be traceable to an authoritative source in a regulatory report, say.

Value migrates from the output to the information underneath it.

Content as a data derivative

Content providers often overvalue their finished products – the articles, the reports, the newsletters – and undervalue the information that enables them. Interview notes, archives, user comments, conference transcripts and even sales calls are potentially valuable data assets. This information is rarely captured systematically or structured. It is usually organised according to the teams that created it, and scattered across CMS and CRM platforms, event teams, research databases and individual employees’ notes.

AI can produce more content from these assets, but it can also combine them into products with greater utility and, potentially, predictive value. A specialist publisher might use them, alongside company announcements and usage patterns, to identify emerging themes or alert customers to developments in the markets they follow.

This changes what it means to build a valuable content asset. The finished article becomes one derivative of an underlying body of structured, proprietary and trustworthy information; the same body might also generate a benchmark, a dataset, an alert or a research tool. Publishers therefore need to become much more disciplined about capturing information systematically, validating it, structuring it consistently and documenting where it came from.

Getting closer to the decision

Generating proprietary information is only one side of the opportunity. AI is also enabling higher utility.

Publishers have traditionally delivered information and left the user to decide what to do with it. A regulatory newsletter tells a compliance officer what has changed; the compliance officer then has to determine how it matters to their company and if any action is needed.

AI allows information products to move further into that decision. Instead of summarising new regulation, a product might identify the changes relevant to a particular company, compare them with its existing policies, flag the documents that need updating and alert the employees responsible.

Not every publisher needs to build software or embed itself in every customer’s operations. But it is important to think about how information is delivered and how it can be combined with contextualising data, so that it evolves from reporting what happened to triggering action.

So: are you still a publisher if you don’t publish? Possibly not, and maybe it doesn’t matter. The businesses best positioned to succeed have stopped treating publication as the product. They look more like proprietary information infrastructure, with events, benchmarks, alerts and APIs sitting alongside the editorial that feeds them. Publishing becomes one of many outputs of these operations.

5 questions for your next board meeting

How much of our revenue depends on resilient content? If non-branded search referrals fell by half next year, which products would survive?
What is our competitive differentiation? What are we known for, and which audiences can we serve?
Which products are embedded in a customer’s workflow? How can they be integrated more deeply either through product development or partnerships?
What data do we generate that nobody has yet thought to collect, and what would it be worth to someone else?
What are we going to stop doing so we can focus on increasing our strategic resilience?