How I Used CoPilot and ChatGPT To Improve My Media Lists (and why that matters) – Part 1

 

PR professionals and their firms have subscriptions to commercial media databases like MuckRack, Cision and Agility. Our Pando PR team prefers MuckRack and we find this database indispensable, however it wasn’t always like that.

If you are a sole proprietor, or work in a small niche agency, sometimes a database isn’t entirely necessary. For years, Pando functioned quite well building media lists by hand. It was manageable because first, our focus was so narrow. We worked only withK-12 and higher education trades and the beat reporters at national and metro media. I knew every editor and reporter in the education trades and if I didn’t know a reporter in a metro or national outlet, it was pretty easy to find the best reporter to pitch on the media outlet’s website.

Now that there are fewer dedicated beat reporters, high staff turnover at outlets, and new digital-only news sites launching every day, subscribing to a media database has become essential. Using an AI tool like CoPilot and ChatGPT along with your media database can make your media lists even more effective. Here’s how I did it.

Bigger lists are not more effective, nor are they ideal

First, it is important to stress that the goal should not be to build the biggest possible media list for pitching your story. Rather the goal should be to find a handful of reporters who are likely to be interested in your story and send them a terrific pitch. It’s a whole other article to explain how to make pitching more effective, but for this article, we’re looking at the media list itself.

If you know your beat well, you might be able to compile a list from your personal contacts. However, if you lack those, are new to the space, or are pitching in a new area, CoPilot or ChatGPT can help. First, build a media list in your preferred database like MuckRack and make sure to check each name and their beat to remove those who only landed there by chance.

How I used AI to make my list better

Once you have that larger list from the database, then use an AI engine to search the web. Don’t rely on AI search, like the one that shows up on Google, Bing or other search tools. Use an actual GPT platform like OpenAI or CoPilot so that you can prompt and refine the AI’s results.

For a recent project, I tried both OpenAI and CoPilot. Here’s the basic flow of that effort.

First, I put this query into MuckRack:

"international student" Near/10 "enrollment" NOT "harvard". 

That generated a great list but I did a lot more work to refine it. Adding the restriction of limiting for words that only appeared in the reporter’s own byline got it down to a very manageable group of reporters who had written about this topic before. Then I did more culling to get to 50 names.

Then I moved to CoPilot and gave it this prompt:

Build a table of 20 reporters in the U.S. or other countries who have written about trends in international students enrolling in U.S. based colleges and universities. Rank the contacts by the rank of the publication where the article appeared.

This list was okay but a lot of the outlets were independent organizations, not media. Also it gave me several “staff reporter” contacts, which clearly wasn’t of help. This search just wasn’t helpful so I abandoned it.

I decided to try ChatGPT using this prompt:

Give me a list of 20 reporters in the U.S. or other countries who have written about trends in international students enrolling in U.S. based colleges and universities. Rank the contacts by the rank of the publication where the article appeared. Exclude media that don't have an individual name, like those that say "Staff report".

ChatGPT took a long time to process the request (it was using the more sophisticated GPT-5 for reasoning capabilities) but it produced a few names that had not shown up in MuckRack. Cool right? More work was needed though.

I went through each of the reporters ChatGPT had produced, and discovered that many of their relevant articles were from 2023. I needed more current material so then I got another idea: to look for the relevant articles first and then backtrack to the reporters.

Here’s how I did it. My prompt was:

Find me the most significant news articles in the last 6 months about international student enrollment in US colleges. Don't pick stories that are only about Harvard or Columbia. I need stories that are broader in topic, looking at how dropping enrollment from international students might affect higher education.

This ended up being the most useful prompt because it let me quickly evaluate the list to find articles relating to my topic. Adding “last 6 months” to the query ensured the articles were current. Then, it was a simple matter to add the names to my list in MuckRack because there were several that MuckRack had not surfaced. I was really happy to have done the extra searches in ChatGPT.

The key is to filter for articles and then examine the reporters’ beats. –My last prompt to ChatGPT was the most helpful and it made me rethink how I’ll do my media list building in MuckRack, too. Seeing the articles in a list familiarized me with significant and recent news, and it simultaneously made it easier to review each reporter’s body of work. I could immediately see their beat just by clicking on their byline. This was a quicker and better way to find the small handful of reporters who were already very current on the topic and they weren’t mixed in with a hundred other mildly irrelevant reporters.

There ended up being so much in this article that I broke it into two parts. I’ll publish Part 2 on this article in two weeks and add the link here.

This was originally published in PR in EdTech on LinkedIn on September 4, 2025.