storyobjectmodel.comfor newsrooms

What it is, if you run a newsroom.

A way of writing the story down once, so that every system in the building, and everyone in it, is working from the same thing.

Not a product, not a system you install, nothing to buy. An agreed format, the way a wire is a format, that the tools you already have can read and write. This page is the plain version. The rest of the site is the technical one, and you will not need it.

The problem it exists for

Today the most valuable context in your newsroom is fragmented across systems or lost in messaging apps. Journalists spend a real part of every day carrying it from one tool to the next by hand. What they do not carry never reaches your systems at all, and very little of any of it is logged.

The decisions that shaped a story, what is confirmed, what is held, who cleared what, live in a chat thread and somebody's head, and by the next shift they are gone. Six months later nobody can say what was known when.

As your tools become intelligent

Every tool you buy from now on will have AI in it, and every one of them will want the story. Today it gets the story by guessing from the rundown, by being told by hand, or through a custom point-to-point integration with whichever system happens to hold the piece it needs. Six tools means six of those, each built once, each maintained forever, and each producing a private version of the story that none of the others can see. With a single integration point, every tool reads the same story, and the story is yours rather than theirs.

SOM is not AI. It is data. It says nothing about which models you use, which vendors you trust, whether you build or buy, or how far you are prepared to let any of it go. That is your strategy, and this does not touch it.

What it does is take the most valuable data you own, the context of your stories, what is known, what is held, who decided and on what basis, and structure it so a machine can read it reliably: named fields, versions, attribution, the difference between a fact and a claim. Then it delivers that to every tool the moment it changes. It is the story, in the form a model or an agent can actually use.

The consequence is that the intelligence in your tools works from your editorial reality rather than reconstructing it, and anything it produces is attributable in the same terms as everything else on the desk. An agent joins under the same rules as a person: it reads the story, it proposes, and it never puts anything to air. The one thing this asks of your AI strategy is that whatever you buy or build should read the story rather than rebuild it.

What changes on the desk

Three things, and they are all things you already do.

You decide once. In the tool you already use. A hold placed on a figure by the duty editor holds in the graphics, the ticker, the web page and the social cut, without anyone being told. When it lifts, everything that was waiting is released together.

A change reaches everything built on it. A figure revised at source flags every script, graphic and post that was cut on the old one, so the recall is done for you and the judgement stays yours.

It is on the record by default. Who decided what, against which rule, when. Not a log somebody built, but a by-product of how the decisions are written down. When a complaint, a regulator or your own archive desk asks what you knew and when, the answer already exists.

It is wider than the desk

Compliance is the easiest thing to show on a stage, so it is what the demonstrations lean on. It is not the point. The point is that the story becomes something every system in the building can read, and that changes four things that have nothing to do with a standards desk.

Control. The gallery, automation and playout can read the story too. A control system that knows a graphic is held, that a package was re-cut after the fact changed, or that a script has been cleared for one route and not another, can act on its own rules without a person relaying it. Your vendors decide what their tools do with that; the story just makes it knowable.

Screens with context. Any monitor that subscribes can show the live state of every story in the building rather than pictures with a slug. What is confirmed, what is held, who is on it, what changed in the last ten minutes. The morning meeting, the gallery wall and the editor's phone are all reading the same thing.

Data about your own operation. Because every decision and every change is written down as it happens, you get, for the first time, a record of how the newsroom actually works rather than how the handbook says it does. How long holds really last. How often things are reversioned. Where corrections originate. It is your record, it exists as a by-product, and what you do with it is yours.

A different way to integrate. Today every pair of tools that needs to talk needs its own connection, and the number of connections grows faster than the number of tools. With a shared story, each tool integrates once, with the story, and inherits everything else that reads it. A new tool joins in days. An old one joins through a bridge without being touched. That is the part your technology team will care about most, and it is also what makes the rest of this affordable.

Down to the frame

Today a clip is a file, and what you know about it lives somewhere else. The confirmation came by phone, the hold is in a chat thread, the clearance is in an email, and none of it is attached to the pictures. When the clip is cut down, reversioned and sent to three outputs, what you knew about it stays behind.

SOM connects to a companion open standard from BBC R&D called TAMS, which does one thing: it makes every frame of every piece of media in the building individually addressable. That lets the story point not at a file but at the exact run of frames that matters, the eleven seconds in which the governor gives the figure. Those frames are then tied to everything the newsroom knows about that story: where they came from, who vouched for them, what was confirmed, what was held, who cleared them for which output, and what changed afterwards. At that moment, and over time, because the record keeps growing after the pictures have gone out.

It runs both ways. From a story you can find every place those frames were ever used. From a frame that went to air you can get back to exactly what was known when it did. A hold placed on a figure holds on the frames that carry it and nothing else, so the rest of the clip stays usable. Your media store does not have to change for this; the same connection works at the level it can address, and it gets sharper as the tools do.

What it does not do

It does not make decisions. Nothing goes to air because a machine decided it should; a person clears it, and if the words change afterwards it has to be cleared again. It does not replace anything you run: your rundown, your media store and your gallery stay exactly as they are. It does not replace journalists. It puts their judgement on the record, attributed to them, which is what a byline has always been for.

What it costs

Nothing to license, and nothing to sign. The standard is open and owned by nobody. The integration work sits with your vendors, one system is enough to start, and the first tier takes them days rather than months. Your part is a policy conversation, not a technical one: which of your editorial rules do you want written down so that every tool respects them. You already have those rules. They are in your compliance handbook.

What you can do this month

Ask your vendors. Whether they speak SOM, and which tier. It is one question, and the answer tells you a great deal about how they think about your newsroom.

Trace one story. The model came out of sitting with real stories after transmission and following how the context actually moved between people and systems. It takes an afternoon, it needs no technology, and what you learn about your own desk is worth it on its own. Ask and we will walk you through one.

Join the working group. It is open to any organisation, taking part costs nothing and never will, and the specification will be shaped by those who show up. That is the basis the standard was built on. It is where the standard changes, and newsroom voices are the ones it most needs.

Who built it

Four years of IBC Accelerator projects, led by the broadcasters and news agencies named on the home page, with the vendors you already buy from doing the engineering. It was shown running with real production tools at IBC in September 2026, and the two demonstrations are on the demos page.

If you want the technical version, everything on this site below this page is it. If you want to talk it through first, hello@storyobjectmodel.com.

what it looks like on the desk