Over the last couple of months, I’ve been experimenting with AICV.news, trying to answer a fairly straightforward question: How can AI provide real value to local journalism?
That’s particularly important to me because AI Coachella Valley received support through the Inland Empire Journalism Fund to explore that very question.
Initially, I wasn’t entirely sure where the project should go.
We could produce another newsletter. We could publish more AI news. We could use AI to research stories faster, summarize announcements and help write articles.
All useful things. But the Coachella Valley already has newspapers, television stations, magazines, independent reporters and institutions publishing information every day.
I wasn’t convinced the region needed another publication doing a variation of the same thing.
Then something interesting happened.
The newsroom became the experiment
We started building an AI-assisted research desk that looks for developments affecting the Coachella Valley and the wider Inland Empire.
The research goes well beyond conventional news sources. We’re examining city council agendas, university announcements, government documents, grant awards, institutional websites, business organizations, public events and the reporting produced by other local journalists.
Twice a day, the AI desk reviews what it finds.
And rather than simply handing me another collection of headlines, it tells me what it thinks deserves attention.
Publish this. Update that. Keep watching this other development. Don’t publish that claim because we haven’t been able to verify it.
It also tells me what it couldn’t find.
That’s the part I hadn’t fully appreciated when we started.
The research process itself was becoming as interesting and useful as the articles we were preparing.
What if we published the process?
Consider a recent example.
UC Riverside announced a $5 million federal AI education initiative involving six community colleges, including College of the Desert.
That’s newsworthy for our region.
But the more interesting questions begin after the announcement.
How much of that money will actually reach College of the Desert? When will classes become available? Who qualifies? What opportunities will there be for students, instructors or local businesses?
We have confirmed the award and the partnership. We haven’t established all those other answers.
Why shouldn’t readers know that?
Or consider Indio’s data-center moratorium. Local journalists reported the council’s decision. Our AI research desk found the city’s supporting documents but hadn’t independently retrieved the signed ordinance.
We should say so, credit the journalists who reported the decision, link to their work and explain what remains unverified.
That’s not unfinished journalism masquerading as a finished article.
It’s a dated account of what we know, how we know it, and what we still need to establish.
There’s another audience arriving
For decades, journalism has largely been organized around human readers visiting websites, subscribing to publications, opening newsletters and following links.
Those readers aren’t going away.
But AI assistants and agents are becoming another way people discover and use information.
I believe that audience will grow substantially.
I can’t tell you today what percentage of AICV.news traffic comes from AI agents. We haven’t established that. And I’m not particularly interested in claiming a number we can’t verify.
What interests me is building for that audience before we know how large it will become.
An AI assistant investigating business opportunities in Palm Desert, workforce training in Indio or government AI policy in Rancho Mirage needs something more useful than a page of promotional copy or disconnected headlines.
It needs sources, dates, evidence, context, relationships between organizations and a clear distinction between what was proposed, approved and actually delivered.
It also needs to know when information is missing.
That’s what I mean by agent-native journalism.
Not journalism written only by machines or only for machines. Journalism organized so that both humans and AI agents can discover, evaluate and use the underlying information.
Credit where credit is due
This experiment isn’t about replacing local journalism.
If KESQ, The Desert Sun, The Palm Springs Post, The Riverside Record or another journalist breaks a story, they deserve credit.
I want AICV.news to help people and agents discover that work, with attribution and direct links to the original reporting.
Likewise, if a university publishes research, a city releases a staff report or a public agency announces a program, we should preserve that source rather than make ourselves the center of the story.
Our contribution is connecting those scattered pieces of regional information, explaining their significance and keeping track of what changes.
Sometimes that will lead to original reporting. Other times, the most useful thing we can do is point to someone else’s work.
Both have value.
Building in public, including the mistakes
We’re still figuring this out.
Our agents sometimes encounter inaccessible documents. They occasionally discover that an earlier recommendation needs to change. They find announcements that sound impressive but lack evidence of implementation.
We have also had technical problems keeping the newsroom’s own research records synchronized.
I don’t see much value in pretending those things aren’t happening.
In fact, I think they belong in the experiment.
We should document the tools we use, the sources we inspect, the decisions the AI desk recommends and the limits of what it can establish.
That doesn’t mean publishing private information or abandoning editorial responsibility. I remain responsible for deciding what AICV.news publishes.
But it does mean making more of the process visible.
A different question for local journalism
The question I started with was: How can AI help us produce local journalism?
I’m beginning to think the more interesting question is:
Can AI help make a region’s knowledge more accessible, attributable, current and useful — not just to people, but to the agents increasingly working on their behalf?
We’re using the Coachella Valley as our starting point. The same approach could work across the Inland Empire, or in other regions.
I don’t know yet whether this becomes a new model for local journalism, a useful intelligence service or simply an experiment that teaches us what doesn’t work.
That’s why we’re doing it.
For now, we’re keeping AICV.news organized around News, Views and Intelligence Briefs. But the Intelligence Briefs are becoming something more distinctive: a public record of what our AI newsroom observes, what it recommends, what it can verify and what remains unresolved.
We aren’t claiming to have figured out the future of journalism.
We’re experimenting with what happens when journalism starts preparing for a future in which its readers may increasingly be AI agents.
And we’re doing that experiment in public.