The message has to survive the filter

There has never been a shortage of news available to me. There are apps, newsletters, websites, alerts, social feeds and an effectively endless supply of people explaining what just happened.

The scarce resource is not information. It is attention.

Opening five apps every morning did not necessarily leave me better informed. It often meant scanning the same story in several places, moving past dozens of things I did not need to know and still wondering whether I had missed the one development that genuinely mattered.

So I changed the interface. Instead of going out to find the news, I asked AI to prepare a daily briefing and bring it to me on a schedule.

That sounds like a small convenience. It has changed my relationship with information more than I expected.

My briefing does not try to recreate a newspaper or squeeze an entire news feed into a chat window. It selects roughly ten to twelve newsworthy items, gives me enough context to understand each one and distinguishes a genuine development from another day of commentary on the same event.

The value is not simply that the writing is shorter. A poor summary can be concise and still waste my time. The system has to make editorial judgements against the priorities I have given it.

For every candidate story, I want it to answer four practical questions.

• What changed
• Why it matters
• What is genuinely new
• What I can safely ignore

Fifty years ago, the newspaper was the interface. A publisher selected the stories, arranged their importance and delivered a finished package. Twenty-five years ago, much of that content moved online. The screen changed; the basic behaviour often did not. We still visited destinations and chose what to open.

Social feeds introduced another intermediary: an algorithm ranked what was likely to hold our attention. But the relationship remained mostly implicit. We did not usually tell the feed what a useful morning briefing should contain or what we already knew.

An AI briefing is different. I can state my interests, preferred depth, geography, tolerance for repetition and the kinds of developments that deserve interruption. The system does not only rank links. It interprets my intent, reads across sources and assembles a new interface around me.

The LLM is becoming a personalized layer between the information available in the world and the fraction of it I actually consume.

AI is becoming the intermediary between content and consumer.

This creates a problem for anyone who communicates for a living.

A publisher can deliver an article without the reader visiting the page. A company can send an email that an AI assistant reads before the person does. A brand can earn an impression inside a model-generated briefing even when its original format, headline and carefully designed journey are never seen.

That makes familiar signals less complete. An email may be technically opened without being read in the traditional sense. A page may influence an answer without receiving a click. A message may be accurately summarized, badly compressed or omitted entirely.

The important question starts shifting from ‘Was the content delivered?’ to ‘What, if anything, was actually received?’

This does not mean people should start writing for robots. It means the first reader of a message may increasingly be a system deciding whether and how to represent it to a person.

If AI becomes part of the distribution layer, producing more content will not guarantee more attention. Neither will making every headline louder.

The messages most likely to survive should be the ones with a clear point, specific language, credible evidence and enough context to be represented faithfully. Original information matters. So does structure. If the central claim is buried beneath repetition or depends entirely on visual treatment, an intermediary has more room to misunderstand—or ignore—it.

This does not make craft irrelevant. It raises the standard for it. Strong ideas still need a human voice, memorable framing and the depth to reward someone who continues to the source. But the content must first make its value legible to the layer deciding whether it deserves that attention.

The future may not belong to whoever publishes the most or declares that they have made the ‘best’ content. It may belong to whoever can communicate something distinct enough, useful enough and trustworthy enough to survive the filter.

The battle for attention may begin before a person ever sees the message.

There is an obvious danger in making information consumption this comfortable.

A briefing trained too narrowly on my existing interests can keep confirming what I already believe. An imperfect summary can remove nuance. A weak source can be made to sound authoritative. Serendipity—the valuable story I did not know to ask for—can disappear when personalization becomes too efficient.

That is why I do not treat the briefing as an oracle. Source links matter. Multiple perspectives matter. The instructions need room for developments outside my usual interests, and the system should be challenged when its choices become repetitive or too convenient.

Even the phrase ‘what I can safely ignore’ should be understood as a provisional editorial judgement, not an objective truth.

Delegating the first pass through the news does not remove my responsibility for what I believe. It changes where that responsibility begins.

The most useful part of this experiment is not any single briefing. It is that the interface can improve through an ongoing conversation.

I can tell it that a story was too shallow, that a topic is receiving too much attention or that a development deserves to be followed tomorrow. The briefing can remember the thread, arrive at the right time and adjust what it surfaces as my priorities change.

That makes it less like opening a publication and more like working with an editor who knows what I am trying to understand.

For consumers, the opportunity is to understand more while consuming less. For publishers and brands, the challenge is to create messages that retain their meaning when another intelligence reads them first. And for all of us, the responsibility is to examine the filter rather than forgetting that it exists.

The interface between people and information is changing again. This time, it is not only changing where we read. It is changing what reaches us at all.