By Editorial TeamIn PR, we’re always looking for an edge, a way to work smarter. The arrival of AI like ChatGPT is completely changing how agencies handle content and media outreach, giving us tools that can draft a press release, spitball pitch angles, and personalize a thousand emails at once. This is about more than just raw speed. It’s about giving human experts a massive strategic boost. So where do you actually start if you want to get real results from ChatGPT in your daily PR grind?
Key Takeaways
- Set up ChatGPT’s custom instructions with your brand voice and tone guidelines. Doing this upfront saves a ton of time on revisions later.
- Build a library of go-to prompts for routine PR jobs, like drafting press releases or media Q&As, to keep the output quality high.
- Let AI do the grunt work of initial research and brainstorming, but every single fact and sentence must be checked by a human before it goes out the door.
- You can personalize media pitches at a scale that was impossible before by feeding ChatGPT a reporter’s bio and recent work to generate custom opening lines.
- Hooking AI up to your CRM lets you see what’s actually working, so you can use that engagement data to make your next AI-assisted campaign even better.
“Earned media that appears in AI results now carries more influence than traditional coverage alone. The PR teams pulling ahead are the ones treating AI visibility as a measurable outcome of their work, not something that happens downstream of traditional metrics.”
1. Define Your Brand Voice and Messaging with Custom Instructions
Before you ask ChatGPT to write a single word, you need to set up its “Custom Instructions” (available in Plus and Enterprise). This is your first and most important step. I get really specific here, inputting guidelines for tone (like authoritative vs. empathetic), words we always use, and words we never use. For a tech client, I might put in something like: “Write with a professional, forward-thinking tone. If there’s a simpler way to say it, use it instead of jargon. Always use the phrase ‘innovative solutions’ and never ‘modern tech.’ No exclamation points in formal comms.” Getting this right at the start means you’re not constantly rewriting everything the AI gives you, which would defeat the whole point.
Pro Tip: Add a section describing your target audience. For instance, “We’re talking to B2B SaaS decision-makers in finance. They care about ROI and operational efficiency, so make sure the content hits those points hard.” The AI will then frame its answers for that specific reader.
Common Mistake: Giving vague instructions. Telling it to “be professional” is useless. You have to spell it out: “Professional means we use formal language, no contractions, and every claim is backed by data.”
2. Generate Draft Press Releases and Media Alerts
With your custom instructions locked in, drafting press releases gets way faster. You have to start with a solid prompt. For a product launch, I’ll use a detailed template: “Draft a 500-word press release in a journalistic style announcing the launch of [Product Name], a new [Product Category] from [Company Name]. The main feature is [Feature 1], which solves [Problem 1]. Also mention [Feature 2] and [Benefit 2]. The target audience is [Audience]. Give me a quote from [Spokesperson Name, Title] about the product’s impact.”
ChatGPT will spit out a full draft. Now, the human work begins. You have to comb through it for factual accuracy, subtle messaging, and to make sure it actually sounds like your brand. I find it’s great for the basic structure and a few headline ideas, but the quotes it generates are almost always terrible and need a complete rewrite to sound like a real person. There’s a reason a 2024 PRSA survey found that while 70% of PR pros are using AI for content, only 15% would ever trust it to produce a final draft without heavy human oversight.
Pro Tip: Make it generate three different headlines and three different opening paragraphs. You’ll get more options to play with, and it’s a better way to get the creative juices flowing than staring at a blank page.
Common Mistake: Just taking the first draft and running with it. AI can “hallucinate” facts and churn out generic phrases. Check everything. Then inject your own human insight to make it good.
3. Develop Tailored Media Pitches and Outreach Emails
Personalizing media pitches is where ChatGPT really starts to feel like a superpower. Instead of writing every single email from scratch, a process that takes forever, you can feed the AI info on a specific journalist. For example: “I’m pitching [Journalist Name] at [Publication Name] about [Story Angle]. They just wrote about [Related Topic – provide URL if possible]. Draft a short, personalized email pitch under 150 words that connects our story to their recent work and gives them a fresh angle.”
The AI can scan their past articles and find the right hook, tailoring the first few lines to what they actually care about. This is how you scale personalization without sounding like a robot. I’ve found pitches we’ve developed this way get much higher open rates because they don’t read like a generic blast, they show we’ve done our homework. The goal is to make that first touchpoint more effective, not to automate the relationship out of existence.
Pro Tip: Build a few prompt templates for the different types of journalists you contact, like one for tech reporters, another for business editors, and a third for lifestyle bloggers. It just makes the process that much faster.
Common Mistake: Firing off an AI-generated pitch without reading it first. A bad pitch, whether a human or a robot wrote it, is a great way to burn a bridge with a reporter. Make sure the tone is right and the references are spot on.
4. Brainstorm Story Angles and Campaign Ideas
When you’re completely out of ideas, ChatGPT can be a great brainstorming partner. Just feed it the basics: your company’s message, who you’re trying to reach, and what’s happening in the industry. For example: “Our company, [Company Name], sells [Service/Product]. We need some fresh media coverage around the [Theme, e.g., the future of remote work]. Give me 5 unique story angles we could pitch to national business publications, keeping current economic trends in mind.”
The AI will come back with a list of angles, and while some will be duds, there are usually a few you hadn’t thought of. It’s also great for mapping out thought leadership topics or figuring out what data you’d need to build a really compelling story. I’ve used this to find weirdly specific angles that ended up landing us placements in niche trade publications we would have otherwise overlooked in our PR strategy.
Pro Tip: After it gives you angles, ask it to suggest which types of media outlets or journalists might be a good fit. It won’t have their email address, but it’s pretty good at suggesting beats or publication categories.
Common Mistake: Thinking every idea it generates is a winner. Many of them will be generic or just won’t work. You have to use the list as a starting point and apply your own professional judgment.
5. Draft Q&A Documents and Talking Points
Prepping a spokesperson for an interview is one of the most important things we do. ChatGPT can help speed this up by drafting a list of likely questions and answers. Just give it the press release and key messages. Prompt: “Based on this release about [Topic], give me 10 tough questions a journalist might ask. Then, draft on-message answers for each, making sure they hit our key talking points: [Key Point 1], [Key Point 2].”
Getting this first draft done in minutes lets your team spend their time on the more valuable work of refining the answers and running mock interviews. It’s also incredibly helpful for war-gaming the really difficult questions and having pre-approved responses ready to go. I always tell my clients to use these Q&As to internalize the message, not to memorize a script, ensuring they sound like themselves, not a robot.
Pro Tip: Push it further by asking for a follow-up question for each of the main questions. That’s how you prepare a spokesperson for when a reporter really starts digging in.
Common Mistake: Letting a spokesperson use the AI-generated answers word-for-word. They’ll sound stiff and unnatural. The AI’s job is to build the foundation. The human’s job is to deliver it convincingly.
6. Analyze Media Coverage and Sentiment (with Limitations)
Sure, dedicated media monitoring tools like Cision or Meltwater are built for deep analysis, but you can use ChatGPT for quick, qualitative reads. Just copy and paste a chunk of text from an article and ask: “What’s the sentiment of this article about [Company Name]? Positive, negative, or neutral? What are the main themes?”
For a quick check on a couple of articles, it’s surprisingly good at pulling out recurring words and summarizing the overall tone. But you have to know its limits. It can’t track coverage volume over time, calculate share of voice, or tell you who the most influential writers are across the entire internet. Think of it as a smart assistant for a first-pass review, not a replacement for a real analytics platform.
Pro Tip: Paste in the text from several articles on the same topic and ask it to summarize the media’s consensus. It’s a fast way to get the gist of the current narrative.
Common Mistake: Expecting it to be a full-blown media monitoring suite. It’s a language model designed to process text, not a data platform built to aggregate and analyze huge volumes of information.
Integrating ChatGPT into a PR workflow isn’t about firing your strategists or relationship-builders. It’s about augmenting their skills. When you automate the repetitive stuff and speed up content creation, you free up your people to spend more time on what they’re best at: high-level planning, building media relationships, and handling a crisis. The trick is to treat AI like a very smart, very fast assistant who always needs a human expert to check their work.
Can ChatGPT write a press release that’s ready to send?
No. It can draft a surprisingly good press release, but it should never be sent without a human reviewing and editing it heavily. AI models are known to “hallucinate” facts, use clichéd language, and miss the kind of nuance that makes PR effective. You have to fact-check everything, punch up the tone, and make sure the message is perfectly aligned with your goals.
How do I get ChatGPT to sound like my brand?
Use the “Custom Instructions” feature. Be extremely specific. Feed it detailed guidelines on your brand’s tone, style, words to use, and words to avoid. Giving it examples of your best-performing content also helps it learn your voice and apply it to everything it generates.
Is it ethical to use AI for media outreach?
Yes, as long as you’re being responsible. The ethical line is crossed when you use it to spam people or be deceptive. The right way to use it is to create more relevant, better-researched pitches that respect a journalist’s time. A human should always review and send the final message to keep the relationship authentic.
What are the biggest limits of using ChatGPT for PR?
It has several big limitations. Its knowledge has a cutoff date, so it’s not aware of breaking news. It can be confidently wrong, making up facts. It also lacks real strategic judgment and can’t read the room or understand complex human dynamics. It’s a tool for execution, not a replacement for a seasoned PR mind.
Can ChatGPT help in a crisis?
It can definitely help, but with extreme caution. In a crisis, speed matters, and ChatGPT can draft initial holding statements or potential Q&A docs in seconds. But given the high stakes, every single word it generates must be scrutinized and approved by experienced crisis communication professionals before it sees the light of day.
