Why Medialyst's AI Agent Media Database Changes PR

Hey there,

Alright something crazy just happened:

I went mega viral on X(twitter) with 4.7M views last week.

TLDR is I shared an essay about how I run Medialyst with an army of AI agents—how they work, why they matter, what's coming next. Posted and went to bed.

Then woke up to millions of views and thousands of follower, including multiple tier-1 AI reporters, a16z partner, the "AI godfather" and Michael Arrington—the guy who founded TechCrunch. My inbox flooded with VCs, consulting, influencer gigs, job offers and founders asking for advice on building agent systems—and we also got 300+ people on Medialyst's waitlist.

Apparently I'm now "the AI agent guy" in the tech bubble.

Which is funny, because I've been building AI agents for the past year—just quietly, inside Medialyst.

AI Agents? Why PR?

Ok I know you probably don't care, but let me explain what this actually means for you:

Traditional media databases (e.g. Muckrack, Cision) charge $10,000+ per year for access to a spreadsheet. A spreadsheet of journalist names, emails, and "beats" that were accurate sometime in the past. Maybe last quarter. Maybe 2023. Who knows.

The problem? Humans can't scale. There are 800,000+ working journalists. They change jobs, shift beats, switch publications. A team of 500 researchers can't keep up. The math doesn't work.

So you pay $10k/year for data that decays the moment you access it. It's absurd.

What If the Database Could Research Itself?

This is the question we've been obsessing over.

What if, instead of humans manually updating records, you had AI agents that could automatically look up the latest data?

We built exactly that.

Now every time you pull a journalist record in Medialyst, here's what happens:

• We check our database first — if we have fresh, validated data, you get it instantly
• If it's stale or incomplete, we enrich live — AI agents research in real-time, pulling from author pages, social profiles, and 10+ data providers
• Everything gets written back — AI agents use these as "evidence" to improve the database, so the next lookup gets that improved data instantly

The database improves at a rate proportional to how much it runs—not how much we work on it. The longer it runs, the fresher it stays.

So this is what we've been building. And it's finally ready.

What's New in Medialyst

All of this AI-agent buzzwords lay the foundation for a much improved journalist lookup many of you have requested.

Here's what shipped today:

🏷️ Better publication filtering You told us you were getting results from brand blogs, corporate newsrooms, and newswires. Fixed.

We now classify every publication: Out of 65k publications we've indexed so far, 73.31% are real editorial outlets, 25.12% are brand content, and 1.57% are wire services. That means 1 in 4 "pitchable" looking results were actually unpitchable—now they're filtered out automatically.

🔗 Duplicate journalists merged Same person showing up multiple times? Now with the database we've built conflict resolution that merges duplicates into one canonical profile with all their verified contact info.

⚡ Faster lookups We rebuilt the enrichment pipeline on top of this new database. Best case: your lookup returns instantly from validated cache. We're still tuning the system—goal is all lookups at least 5x faster in the coming weeks.

📊 Richer journalist data Beats, recent coverage, social links, publication history—profiles are significantly more complete. We've seen significant improvement in LinkedIn find rates for tier 1 reporters, and 55% improvement overall. And profiles keep getting better because the agents never stop learning.

✅ Email verification built in Every email is validated before you see it. You'll see when it was last verified, a confidence score, and exactly where the email came from—so you know whether it's from an author page, a data provider, or somewhere else. No more guessing.

Go run a search. You'll feel the difference.

— Elvis

P.S. If you're still wondering whether $10k/year is still a fair price for a media database in 2026, read our full blog post on how we built world's first AI-agent powered media database. Spoiler: it's not.

P.P.S. If you want to geek out about AI agents and see how they can be applied for your company (not just PR workflows, anything), reply to this email I'll share some ideas.

— Elvis Sun, Founder at Medialyst

P.S. Need help getting started? Reply and or book a 1:1 with me.