AI Is Changing PR Outreach — But It Can’t Do the Job Alone

PR professionals share the real-world limits of AI tools, and the smart workarounds they’ve found.

Artificial intelligence has become a go-to tool for public relations professionals looking to work faster and pitch smarter. But ask anyone who has actually tried to run their outreach through AI, and they’ll tell you: it’s not as simple as typing a prompt and hitting send.

We asked PR pros and business owners who handle their own PR one question: What is one limitation you’ve run into when using AI tools for outreach? The responses — from agency founders and marketing leaders to startup CEOs and solo operators — painted a clear picture. AI can speed things up, but it can also send you off a cliff if you’re not careful.

Here is what they found, and how they fixed it.

 

1. AI Doesn’t Know Who Reporters Are Today

One of the most common problems PR professionals run into is simple but costly: AI tools give you outdated contact information. Reporter beats change. Journalists switch publications. Email addresses go dead. AI, trained on older data, doesn’t know any of that.

“AI was trained on data that knew none of those movements, and came up with pitches sent to reporters who had departed publications months ago or covering beats no one wrote about anymore.”

— Jimi Gibson, VP of Brand Communication, Thrive Internet Marketing Agency

Gibson said his team sent three pitches before catching the problem. The damage wasn’t just wasted effort — it was burned credibility before a relationship even started. His fix: never trust AI-generated contact details. Search the reporter’s name, check their latest bylines, and confirm their publication before sending anything. “AI doesn’t do well with anything that requires knowing up-to-date, person-specific information,” Gibson said.

 

2. Timing Is Everything — and AI Doesn’t Know What Day It Is

Even a well-crafted pitch can fall flat if it lands at the wrong moment. Saini Rhodes, a real estate expert at Clever Offers, learned this the hard way.

“I remember pitching about rates in such a way that the pitch landed in inboxes on the very day the FED made an unexpected move. There was nothing wrong with the pitch, except it was no longer relevant because it was outdated.”

— Saini Rhodes, Real Estate Expert, Clever Offers

Rhodes now checks financial news headlines before sending any pitch. If major news has just broken, the whole process stops until she can rethink the angle or wait for a better moment. “While the tools help me craft the pitch,” she said, “I need to evaluate its timeliness.”

 

3. Polish Without Relevance Is Just Fancy Spam

Several PR pros pointed out a subtle but dangerous trap: AI makes weak pitches sound good. That’s not a compliment. When a pitch is well-written but not relevant, it still gets ignored. It just wastes more of your time getting there.

“The biggest limitation with AI in PR outreach is that it can make a bad pitch sound polished instead of making it relevant.”

— Callum Gracie, Founder, Otto Media

Gracie’s solution was to bring in prompt engineers who focus on context, not just snappy phrasing. Every AI response now starts with the journalist’s brief, the source fit, specific proof points, and the exact claim the team can support. The results were striking: his Connectively account rank moved from over 3,000 to the top 85 in less than a year.

Nikita Baksheev, Head of Marketing at Ronas IT, saw the same issue from a software development company’s perspective. His team built a simple three-question check before any pitch goes out: Why this journalist? Why this topic now? What specific evidence can we add that a generic source can’t?

“AI makes PR outreach faster, but it also makes weak relevance look more professional. That’s dangerous.”

— Nikita Baksheev, Head of Marketing, Ronas IT

 

4. Generic Pitches Get Ignored — Every Time

The word that came up again and again in responses was “generic.” AI doesn’t know your client’s story. It doesn’t know what makes their company different from the 400 others in the same industry. And journalists notice immediately.

“AI has no understanding of what makes your client different from the other 400 personal injury firms in Los Angeles, and it’s going to be obvious right from the start.”

— Travis Hoechlin, CEO, RizeUp Media

Hoechlin said his team discovered that AI-generated pitches without firm-specific information got three to four times fewer replies from reporters. The turning point came when they stopped letting AI write the final pitch and instead used it to pull background and angles. Humans wrote the actual outreach. Response rates jumped, and reporters were replying within 24 hours.

Rafael Sarim Oezdemir, Head of Growth at EZContacts, described a similar pattern. AI would produce structurally sound pitches that defaulted to template language that could apply to any company. Journalists who cover specific beats spotted it right away.

“In case with pitching and media outreach, being credible alone is not enough. It needs to be combined with the ability to demonstrate that you actually understand what this specific journalist does.”

— Rafael Sarim Oezdemir, Head of Growth, EZContacts

His fix: use AI to edit, not to originate. His team now researches each journalist first, finds a human-generated angle, adds at least one concrete data point, then lets AI clean up the copy.

 

5. AI Personalization Can Feel Forced and Fake

One of AI’s big selling points is personalization at scale. In practice, it often misfires.

“When we tried letting the model write the actual pitch emails to reporters, it consistently hallucinated forced, awkward connections between a writer’s past articles and our software.”

— Kevin Lourd, Founder, GrowthAgency.dev

Lourd, who also builds AI outbound tools at distribute, pulled AI out of the drafting process entirely. Now his team uses it only for research: the AI scrapes a journalist’s recent work and drops bullet points into a dashboard about their beat and focus. Humans read those notes and write the actual pitch. “It keeps the research phase fast but entirely stops the tone-deaf emails,” he said.

 

6. Too Many Options Can Cause Paralysis

Not every AI problem is about bad output. Sometimes the problem is too much of everything.

“There’s so many options that it’s overwhelming. AI has made it too easy to do weak or useless outreach at scale.”

— Katie Steele, Founder & CEO, SmartFirm.io

Steele said she has sometimes been paralyzed by the sheer number of AI tools available, leading her to skip the process altogether. When she does use AI heavily for research and writing, she often ends up spending as much time as she would without it. Her conclusion: the real leverage is not volume. It’s targeting the right outlets and having genuine, personalized conversations with fewer, better-fit contacts.

 

7. The Fix Most Pros Landed On: Use AI to Research, Not to Write

Across all responses, one theme emerged loudly: AI belongs in the research and prep phase, not in the final email.

“The best outreach still sounds like a person sat down for two minutes and actually thought about who they were writing to.”

— Nemanja Marinkov, VP of Marketing, PressWhizz

Marinkov put it bluntly. Unless your strategy is spray-and-pray — which he said is a strategy problem, not an AI problem — you should be writing your own outreach. “People respond to people, not perfectly optimized prompts,” he said.

Scott Brown, Founder of Focus Group Placement, agreed. He uses AI for a first draft or an outline to get past the blank page, then rewrites heavily in his own voice and adds specific details about the journalist’s recent work.

“AI is a solid starting point for overcoming the blank-page problem, but the human layer of personalization is still what actually gets responses.”

— Scott Brown, Founder, Focus Group Placement

Milos Eric, Co-Founder of OysterLink, said his team learned the same lesson after noticing that their AI-generated media pitches sounded identical to ones produced by other companies using the same tools. Their fix was to move AI into a supporting role: data collection, insight organization, and concept validation. The media angle, the viewpoint, and the final writing stayed human.

“AI improved our development process. However, differentiation was still derived from having something valuable to communicate.”

— Milos Eric, Co-Founder, OysterLink

Christopher Helm, CEO of Helm & Nagel GmbH, made a similar point from the editorial side. He uses AI to automate large-scale research he couldn’t do on his own, not to generate the outreach copy itself. His concern about the broader trend: “I fear many use AI to generate more AI-slop and spam journalists. Offering just a good narrative of the same topic will not succeed anymore.”

 

A Smarter Starting Point: Let AI Draft, Then Take Over

There is a middle path between letting AI write everything and refusing to use it at all. It starts with giving AI something real to work from — and that is where most generic tools fall short.

On Connectively, journalists and editors post their source requests directly: what they are working on, what kind of expert they need, and what angle they are pursuing. That query context alone already solves one of the biggest problems the PR pros above identified. AI is no longer guessing at relevance. It is starting from a specific brief.

For paying users, Connectively AI takes that a step further. The platform lets you build a knowledge hub from your own body of work — past pitches, published articles, podcast appearances, YouTube videos, and more. Connectively AI uses that library to identify queries that are a strong match for your actual expertise, then draws on your own content to generate a pitch draft or suggest an angle that is grounded in what you have already said and proven. The result is a first draft that sounds like you, because it is built from you.

That is exactly what the experts in this piece said was missing from off-the-shelf AI output: the founding story, the specific data point, the client detail that only exists inside your knowledge of your own work. A knowledge hub does not fix AI’s generic tendencies — it bypasses them entirely.

Free plan users can take a similar approach manually. When a relevant query appears, feed the journalist’s brief into an AI tool alongside your own background — a past pitch, a relevant article you wrote, a case study, specific results from a campaign. That context gives the AI something concrete to build from instead of reaching for template language. The draft you get back will be closer to the finish line and require far less rewriting.

Either way, the workflow matches what every pro above eventually landed on: AI handles the starting point, you own the final pitch.

 

The Bottom Line

AI has a real role in PR outreach. It can scan topics, summarize journalist beats, organize research, and get you past a blank page faster than any other tool available. The professionals who are getting results with it are not the ones using it to automate everything. They are the ones who know exactly where to stop.

The clearest takeaway from all expert responses: AI can write a pitch that sounds like it was meant for anyone. The job of the PR pro is to make it sound like it was meant for one specific person. That part, for now, still requires a human.

 

—

Quotes were collected via Connectively.us from PR professionals and business owners who handle their own PR.