I reworked HARO Emails - Less Spam, More Relevance

So the idea came when I was skimming through Twitter one day and bumped into a chat about HARO (Help a Reporter Out). A reply from a cool startup CEO hit home:

"The only problem I find with them is that now they send me 4 emails per day every day, and I sincerely read like 1-2 per week. Their deadlines usually are very tight as well. But def worth subscribing!"

I felt the same. HARO did get me some sweet media mentions before, but it's a huge time-sink. Sifting through those emails felt like hunting for a needle in a haystack.

Ironically, the folks most sought after for answers are often too swamped to sift through those emails.

Then I thought, why not redesign HARO emails and make it actually usable for busy people like us, with a dash of AI magic?

Keyword filtering, like the one HARO’s got, didn’t cut it. It’s like swinging between missing out or drowning in too much.

So, I played around with an idea. Gave a user profile and a HARO query to ChatGPT and asked it to play matchmaker, rating it from 0-100. The result is very interesting. GPT-3.5 was like, “Oh, you’re both into tech? Perfect match!” Where GPT-4 would understand a more nuance and be able to do a much better job at this.

I put the results into one graph, you can see how GPT-4 never scores anything higher than GPT-3.5, which I found super fascinating.
https://x.com/GlitchPhoton/status/1714434352023589308?s=20

With this approach, I filtered down the daily flood of 100 HARO spam to just 1-2 that actually make sense for you. Here’s a peek at what your new, breezy HARO email would look like.

Give it a whirl if you want: https://prpulse.xyz. I'm doing a 30 day publicity guarantee to make sure people actually find value in this service.

What's next? I'm currently exploring creating a vector embedding of the PR opportunity space, and map a profile to this embedding space, which can potentially reduce the cost of doing this.

Cheers,