VentureBeat seeks an enterprise AI reporter

Large companies are spending heavily on AI agents, on inference, and on the infrastructure underneath both. They make those calls the way they always have: on vendor reputation, their own requirements, price, and what their pilots show them. What they cannot get anywhere is reliable data on what comparable companies actually chose, what it cost, and what broke.

The vendors are not slowing down to help. OpenAI, Anthropic and Google, the cloud platforms, and dozens of smaller providers are all making performance and cost claims faster than any buyer can test them.

That is the gap this job exists to fill. Every month you would see VentureBeat’s own survey data before anyone outside the company, decide what in it is genuinely new, call the executives and analysts who can confirm or refute it, and publish what it means for someone about to spend the money. Your name goes on the work, and the beat is yours to build.

The data is yours first

Every month VentureBeat surveys enterprise AI buyers across five trackers: agentic orchestration, agent reliability and evals, agentic security and identity, AI infrastructure and compute, and context layers and RAG, meaning retrieval-augmented generation. Respondents are director level and above at organizations with more than a hundred employees.

You see the results before anyone outside the company. You work with Rob Strechay, our lead analyst, and you decide what the story is, including where the numbers are thin and the reader deserves to be told so. Almost no publication our size can hand a writer this, and it is the single strongest reason to take the job.

Then you report it out. You find the result that contradicts what the market believes, form a view about why, and take it to people who can test that view: an executive running the AI platform or the security program at a company our readers have heard of, an investor with an original read on where the money is moving, a vendor who knows the territory cold and will talk about the hard parts. You come back with the piece that says what the data means and what a buyer should do about it on Monday.

And the week’s news, one level down

A release lands, everyone writes the same thing by lunchtime, and it is dead by the next morning. The story worth having is underneath it: what it costs, what it breaks, who is quietly not doing it, what a buyer should change as a result. That takes reporting rather than a faster rewrite, and you would have the room to do it.

Judgment is the job

When a Fortune 100 chief information officer, chief security officer or head of AI platform agrees to an interview, it is yours to run. You also decide what is worth running at all. Testing a vendor’s claim and telling the reader which parts of it hold up, rather than repeating the claim, is why a VP of platform engineering keeps reading us. Our readers are about to spend real money, so what they need from you is a view you will put your name to.

Charts are part of the writing

You make your own graphics. Not request them, make them. A good chart is often the fastest honest way to show what the data says, and VentureBeat has nobody doing this today. If you think in charts as naturally as in paragraphs, you will have an unusual amount of room here.

Our stages, and your name on them

VentureBeat runs the AI Impact event series for exactly these readers. You would moderate on stage, interviewing the same executives you write about, in front of the people who buy. Our research gets presented at those events too. A writer who uses that builds a name in this field faster than one who only files copy.

Pace

Roughly a couple of shorter pieces and one substantial one in a typical week, and the mix is yours to argue for. We measure impact rather than count filings: one story that gets cited, quoted and argued with beats four that were merely published on time. A piece that needs three weeks of sourcing gets three weeks, as long as we know it is coming.

Who does well here

Someone with a point of view who publishes it. Our readers are directors and vice presidents of technical functions building and deploying AI inside large companies, and they come to us to decide something. They want to be told which of the week’s claims to believe, and why.

Five or more years covering technology for a serious outlet, with clips that show reporting you originated rather than write-ups of announcements. Comfort with the technical material, meaning how a model gets served, what identity means for an agent, what an eval actually measures, or the appetite to get there fast. Having the data behind you closes some of that gap. It does not close all of it.