Your media database already knows who covers your client's category, when someone on the list was last contacted, which pitches went out and which ones ran. What it doesn't do is decide who should hear from you today, or notice that a reporter on a live list moved outlet last month. This guide covers how to connect that database to Claude or ChatGPT so the list work, the follow-up and the coverage record arrive drafted.

It goes deeper on the list side of our PR agency automation guide. Client approvals and status updates are covered in how to automate PR client admin; retainers and billable hours are in PR agency billing automation.

Connecting AI to your media database software (Muck Rack or Cision)

There are three ways in. Muck Rack and Cision expose an API on the right plan. Failing that, the exports you already pull — a list, a coverage report, a pitch log — can be read on a schedule. Failing both, a private machine reads the reports you download the way a person would.

The rule is the same whichever route you take: the AI reads your system and drafts the action. It does not write back into your media database, and it does not send a pitch to a reporter until you approve the message or decide a job has earned the right to send on its own.

That boundary matters more in PR than in most places. A list is a professional relationship record, and a wrong send burns a contact who took years to earn.

Using AI to build and refresh a PR media list

List building is the job worth handing over first, and it splits into two halves that behave differently.

Building from a brief. Give it the client, the category and the angle, and you get a first-pass list back: who covers that beat, at which outlets, what they have written recently, and whether anyone at your agency has talked to them before. That last column is the one a bought list can never give you, because it comes from your own pitch history.

Refreshing a list you already have. This is the quieter win. Reporters change beats, move outlets and leave the business, so a list built for a launch three months ago is doing less work than it looks like it is. A scheduled pass over an active list flags the contacts whose recent writing no longer matches the beat you filed them under, the ones who have not published in a while, and the ones you have pitched three times without a reply.

The output is a marked-up version of your own list, not a new one. You decide who comes off.

If list hygiene is the part that never reaches the top of the pile, Unprompted can set this up against the database you already pay for.

Using AI to streamline pitch follow-up

A pitch has a short useful life. The follow-up that matters is the one that goes out while the story is still live, and that timing competes with every client call in the day.

• The no-response list. Pitches past the window you set, grouped by client, each with a follow-up drafted that references the original angle rather than repeating it.
• The warm reply. When a reporter answers with interest or a question, the draft back is prepared with the client's materials attached and the question actually answered.
• The deadline flag. A reporter writing on a tight turnaround gets pulled to the top of your morning note instead of sitting in the queue behind routine mail.
• The angle that worked. When one pitch lands, the reporters on the same beat who have not yet heard it are surfaced, with the angle adjusted for each.

Follow-ups are drafted, never fired blind: a reporter who already passed, already covered it, or asked not to be contacted drops off the list before you see it.

Getting insights from your coverage history

Your agency has years of coverage recorded and almost no time to read it as a dataset. Asked in plain language, it answers questions that change what you pitch next.

Which outlets have actually covered this client, and which have been pitched repeatedly without ever running anything. Which angles landed and which ones never did. Which reporters cover your client's competitors but not your client. Whether a tier-one placement last quarter came from a relationship or a single well-timed pitch.

The same records make the monthly client recap a review rather than a build: the clips that ran, the outlets, the reach and the pitches still in flight, assembled from your own logs for you to check and send.

Why you should layer AI on your existing PR software

Your media database stays where it is, and it keeps doing the job it is good at: holding the contact record, the outlet detail and the coverage archive in one searchable place that your whole team trusts. Nothing is migrated.

What sits on top is the reading and drafting layer — the pass over the list, the follow-up that should go today, the recap built from what ran. The judgment stays with your people: whether a relationship can take another ask this month, whether an angle is right for a particular reporter, whether a story is worth calling in a favor for. Those were never list management in the first place.

If list hygiene and pitch follow-up are the parts eating your account leads' evenings, talk to us and we'll show you the weekly pass running on your own list.