143 editors wrote back after a digital PR desk run by AI agents pitched for Flip Education

Why I trusted agents with PR

I'd say what trusted sites say about a company counts more now, not less. Search engines look at it, and so do the AI tools people ask for answers, so being present on third-party sites matters more than ever. The slow part of PR is the research and the writing for each publication in its own language, and that's work AI agents can now do. So a desk of agents did the job end to end. Checks ran before anything was sent, and three rules of mine sat on top. Pitch only where the readers can use the product today, never pay for a post, never trade links. The agents gave the desk its reach. The checks and my rules decided what could go out, and the calls that could cost money or the site's standing in search stayed with me.

The challenge

We had no PR team and no agency. Bought links were never an option anyway, because they can hurt a site in search and they don't build trust with editors. Agencies price per market, and with seven languages across Europe and Latin America this would have needed several agencies or one very large retainer.

How the desk worked

The agents found publications by motive, such as articles quoting out-of-date statistics, pages with dead links, resource pages, newsletters and outlets ranking for our topics. Sites that charge for posts were blocked here. So each pitch started from something that publication could use, a fresh statistic for an old one, a working link for a dead one.

They researched each publication still online, its readers, its voice, its recent themes and the gap a pitch could fill, then wrote one pitch in its own language.

Before sending, scripts and an AI model checked each pitch for topic fit and for a publication already pitched. A pitch that failed any check wasn't sent.

The first batch sat as drafts in my email. I caught that the drafts described the product too narrowly, so I had every one redone before any were sent, then sent the batch myself. Later waves went through the checks, then out through an email platform loaded by the agents. How the product was described was my call before anything went out in volume.

The agents tracked every thread to a placement or a close. Where a reply asked for payment, or for a trade of links, the call was mine.

What happened

The desk started work one evening in February, when I opened my first working session with the AI agents. Within two hours the first pitches were out. The desk ran in waves after that. It pitched 1,774 publications in seven languages. 143 editors wrote back. The picture below shows two of the acceptance emails.

What it cost

My own time was about 1 to 2 hours a week. Published agency prices put a single-market retainer at about $5,000 to $15,000 a month. One agency says a six-month minimum is typical. Each extra market adds about $2,000 to $5,000 a month by one agency's guide.

What I would keep and change

I'd keep the audience rule and the duplicate check, because pitching the same editor twice makes a reply less likely. I'd write the workflow as fixed steps in software from the start, instead of the agent running it from memory and a written description.

When a team shouldn't do this

A competent digital PR team is still better for the average case, above all in a niche where their connections matter, and for placements in high-visibility publications, which is hard to reproduce even with a system like this. And for a single country, because this desk's advantage was many countries at once, where you would need several agencies. For one country, one agency covers the job.