Something worth talking about

Keeping a company visible sounds simple until somebody has to keep finding things to say.

For a while, the obvious material is easy.

A new service launches. There is a promotion. A customer story appears. Something changes inside the business. There is a new member of the team, an event or a result worth sharing.

But most businesses are not launching something new every Tuesday.

Most of the time, people are simply doing the work.

That creates an awkward choice for whoever is responsible for the website or social media.

Post only when the company has its own news, and the channels can go quiet for weeks.

Or publish more often by filling the gaps with generic content — the familiar “five ways to…”, “ten things every business should know…” and AI-written advice that could have come from almost any company.

Neither option felt particularly useful.

There is plenty of genuinely interesting material out there. The problem is the scale of it.

Industry news, local stories, research, case studies, unusual business decisions, failures, experiments, process changes and real examples are being published constantly.

No one person can sensibly monitor all of it.

And ordinary keyword search does not solve the problem very well either.

A page can contain all the right words and still be painfully dull.

What matters is whether there is actually something there that could become a story — a person, a decision, a tension, an unexpected outcome, a practical lesson or simply a detail that makes somebody want to keep reading.

And even then, the same story will not be right for every company.

A useful story for an accountancy firm may be completely irrelevant to a car dealer. Two companies in the same industry may still have different audiences, personalities and things they are comfortable talking about.

So the real problem was not:

“How do we generate more content?”

It was:

“How do we keep finding things worth talking about — specifically for this business?”

• Built a Brand Profile for each company using the system: what the business does, who it speaks to, the subjects it cares about, geographic priorities, editorial boundaries, tone of voice and the kinds of stories that belong in its world.
• Made Discovery search from that context rather than relying on a static list of keywords. The aim is not simply to find relevant pages, but to find material with the potential to become a genuinely interesting story.
• Separated “relevant” from “worth telling”. A source can mention exactly the right subject and still be generic, predictable or too thin to build useful content around.
• Added a Storyworthiness layer that looks for things such as a real actor, action, consequence, novelty, freshness and tellability before a source is treated as a serious story candidate.
• Added a separate Editorial Value check for specificity, evidence, tension, practical insight and genericity.
• Scored promising material against the individual Brand Profile rather than assuming that one interesting story belongs to every company.
• Created a short Editorial Preview for each candidate. It does more than summarise what the source says: it helps explain what is interesting about it and why it may be worth developing.
• Kept Source and Story separate. A discovered article, report or webpage remains evidence; it does not automatically become the story itself. Several sources can support one stronger editorial idea.
• Added optional research and enrichment when the initial material needs more context, an original source, another example or an opposing view.
• Put an Editorial Brief between research and writing. The Brief keeps the central angle, useful facts, source trace, counterpoints and boundaries together before any publication text is created.
• Built separate outputs for Website, LinkedIn and Facebook. A social post is not simply a shortened article: each platform gets its own composition from the same underlying editorial idea.
• Used an AI Writer to create the first version as a story rather than a mechanical summary of the source material.
• Added an optional AI Editor when the first version needs another editorial pass, while preserving the previous draft rather than silently replacing it.
• Kept creative writing and factual checking separate. The system validates names, dates, figures, quotations and reported outcomes against the available evidence before the material moves forward.
• Created platform-specific image prompts after the final text exists, so the visual is based on the actual finished story rather than a generic topic.
• Connected the finished material to publication destinations such as websites, Facebook and LinkedIn.
• Deliberately kept a person inside the workflow. Discovery can find candidates, AI can score, research, draft, edit and check them — but the system does not automatically publish everything it finds.

The aim was never to find an article online and ask AI to rewrite it.

NewsMaker looks for the thing behind the source that is actually worth telling.

The source provides the facts.The Story decides what the piece is really about.

**How the writing works:**

Once a story is approved, NewsMaker does not simply turn the source into a tidy AI summary.

It first works from the editorial angle and the evidence selected for that story.

The Writer's job is to make it readable: find the shape of the story, build the opening, keep it moving and leave the reader with something useful.

If the result is nearly there but still needs work, the Editor can create another version — much like a writer and editor working on the same piece in a magazine or newspaper.

The original version is not lost.

And because Website, LinkedIn and Facebook behave differently, each format is created separately rather than chopping the same article into smaller pieces.

**Human control:**

It would have been possible to automate almost all of this.

Discovery could find a story, a model could score it, another could write it and the system could send it straight to the website or social feed.

We deliberately did not build it that way.

Because the last question is not technical:

“Can we publish this?”

It is:

“Would our audience actually find this interesting?”

And that is still a judgement worth making by a person.

So the system brings good possibilities forward instead of automatically turning every possibility into content.

The machine does the searching, sorting, drafting and checking.

The human chooses what deserves attention.

NewsMaker turns an open-ended content problem into a manageable editorial flow.

Instead of starting with a blank page and asking, “What on earth do we post this week?”, the user starts with a shortlist of real stories that already fit the company’s subject matter, audience and voice.

Each candidate arrives with context: what happened, what makes it interesting and why it may be worth telling.

Weak or generic material can be rejected before time is spent writing it.

Promising material can be developed with additional sources, turned into a proper editorial Brief and then written specifically for the channel where it will appear.

The same underlying system can support different companies, industries and voices because the Brand Profile changes what Discovery looks for and how the resulting story is developed.

The final output is not just more content.

It is a repeatable way of finding better raw material for content — without expecting somebody to spend their working day watching the entire internet.

And because a person remains involved at the key decisions, the company does not have to hand its public voice over to an automated content machine.

The content shortage was never really a shortage of things happening in the world.

It was a shortage of time to find the right things.

Once we separated those two problems, the role of AI became much clearer.

AI is very good at covering more ground than one person reasonably can: searching, comparing, filtering, organising evidence, drafting variations and checking details.

But volume is not the same as judgement.

A hundred relevant links are not a hundred good stories.

And a technically correct article is not automatically something anybody wants to read.

That is why the strongest part of NewsMaker is not simply the generation step.

It is the path that comes before it: understanding the company, searching through that lens and filtering for material with enough life in it to deserve the reader’s time.

**AI solves the problem of scale.**

**People still decide what is interesting.**