AI Agents Are Coming for PR and Comms - And That's a Good Thing

PR has an execution problem. Not a creativity problem, not a talent problem, but an execution problem.

Every comms pro knows the drill. You spend half a day building a media list. Another few hours crafting pitches. You toggle between six browser tabs, three spreadsheets, and a CRM that was designed for salespeople, not publicists. By the time you actually hit that send button, the news cycle has moved on.

Meanwhile, sales teams have had AI agents booking meetings for them since 2023. Customer support runs on autonomous bots. Marketing has programmatic everything. But what about PR? PR is still copy-pasting journalist emails from editorial mastheads, outlet author pages, and WhatsApp groups.

That's starting to change. And the shift isn't coming from the tools you'd expect.

The numbers tell the story. According to a 2026 survey of more than 500 PR professionals, 76% of comms pros now use generative AI in some form, and 90% of PR teams have integrated AI into their workflows. But only 13% describe their operations as “highly integrated.” And when it comes to AI agents specifically — the systems that actually do the work rather than just suggest it — only 12% of PR pros have adopted them. Another 28% haven’t even heard the term. That gap, between AI awareness and AI execution, is the most important story in communications right now.

General-purpose AI is a half-measure for PR

When ChatGPT launched, PR teams were among the first to experiment. And for good reason — writing is the job. But two years in, the honeymoon is fading. Here's why.

Tools like ChatGPT, Claude, and Gemini are quite extraordinary at generating text. They can draft a press release in thirty seconds or brainstorm campaign angles on demand. But drafting text is maybe 15% of what PR pros actually do. The other 85%, that is, identifying the right journalists, understanding their coverage history, personalizing outreach at scale, tracking what lands and what doesn't, none of that lives inside a general-purpose language model.

Ask ChatGPT to build you a media list of tech reporters who cover enterprise SaaS and have written favorably about your competitor in the last six months. It can't. It doesn't have access to journalist databases, coverage archives, or pitch history. It'll give you a plausible-sounding list but half the names will be wrong, outdated, or irrelevant.

This is the copilot vs. autopilot distinction that matters. A copilot helps you write faster. An autopilot researches, plans, builds, personalizes, sends, and measures, and then tells you what worked. PR doesn't need another copilot. It needs an agent.

What actually makes an AI agent different

The word "agent" gets thrown around loosely in tech, so let's be specific.

An AI agent isn't just a chatbot that remembers your last message. It's a system that can take a goal, break it into steps, use specialized tools to execute those steps, and adapt based on what it learns along the way. Think of it as the difference between asking someone to edit a paragraph for you and hiring someone to run your entire media campaign.

For PR, that means an agent needs to do several things at once: understand your brand's narrative and positioning, access real journalist data (not hallucinated contacts), assess which reporters are likely to engage positively, generate outreach that's genuinely tailored to each recipient, distribute through real email infrastructure, and track outcomes back to business metrics.

That's a tough ask. It requires proprietary data, domain-specific training, and purpose-built tooling that no general-purpose model can replicate out of the box.

Think about what an agentic AI workflow has to handle on any given media campaign. Strategy: understanding your launch narrative, your competitive positioning, your audience priorities. Research: scanning thousands of journalists, beat changes, recent articles, and historical coverage patterns. Targeting: scoring each reporter not just for relevance but for engagement likelihood and brand-risk fit. Personalization: writing pitches that reference each journalist’s last three stories without sounding like a mail merge. Distribution: sending from a real inbox that journalists actually open, not a marketing platform their spam filters already know. Monitoring: tracking opens, replies, and coverage in real time. Attribution: connecting every placement back to traffic, leads, and revenue. That’s seven distinct capabilities, and a general-purpose model can do roughly one of them well.

PropeLLM: what purpose-built actually looks like

This is where Propel's AI agent, which we’re calling PropeLLM, enters the picture. And the reason it's worth paying attention to isn't the branding. It's the data.

PropeLLM is trained on the largest proprietary PR intelligence dataset ever assembled, a foundation that no general-purpose AI has access to: more than 25 million real pitches that have actually been sent to journalists, more than 500,000 verified journalist contacts, 150,000+ media outlets, and a corpus of over 2 billion articles spanning the open web and paywalled publications. That's not a fine-tuned wrapper on GPT. That's a model that understands how PR actually works, what gets opened, what gets covered, what falls flat. No general-purpose model can replicate this, because no general-purpose model has the underlying data. ChatGPT has read the internet. PropeLLM has read the inbox.

In practice, this means a PR professional can type a single prompt, let’s say, "Launch a media campaign for our Series B announcement targeting fintech and enterprise reporters in North America", and PropeLLM will execute the entire workflow. It ideates the campaign angle. It builds a media list using AI-powered journalist matching based on coverage patterns, beat relevance, and risk scoring. It writes hyper-personalized pitches that reference each journalist's recent articles and email tone. And now you can send from your native inbox with tracking baked in. Once the campaign is live, PropeLLM keeps working. One of the features customers reach for most is campaign analysis: ask it how a launch performed and it surfaces what worked, which messaging angles drove the strongest replies, which journalists engaged, where coverage landed, and what to do differently next time. It’s the kind of post-mortem that used to take a junior associate two days in a spreadsheet, delivered in seconds. And it reports back on engagement, coverage, and downstream business outcomes. The bigger point is that all of this happens in one place. Ideation, media list building, pitching, sending, monitoring, analysis: PropeLLM is a one-stop shop for the entire PR workflow. That’s something general-purpose AI fundamentally can’t offer. ChatGPT can help you draft a pitch, but you’re still juggling a database tool for contacts, an email platform for outreach, a monitoring service for coverage, and a spreadsheet to tie it all together. PropeLLM collapses that stack into a single agent that owns the campaign from prompt to performance review.

That's not a writing tool. That's a PR execution engine.

The hidden cost of a fragmented PR tech stack

Most PR teams don’t have an AI problem. They have an integration problem. The average comms team runs on a stack of disconnected tools: a media database for journalist contacts, a separate email platform for outreach, a monitoring service for coverage tracking, a social listening tool for sentiment, an analytics dashboard for traffic, and a spreadsheet to stitch the story together at the end of the month. Each tool was bought to solve a real problem. Together, they create a bigger one.

The data doesn’t flow between them. Your media list doesn’t know what your monitoring tool saw last week. Your pitch tracker doesn’t know which journalists drove the most engagement on the last launch. Your campaign report is a manual collage of screenshots. Industry research from shows that more than 70% of agency teams cite media fragmentation as a major operational hurdle — and only 13% of PR organizations describe their AI use as truly integrated. The tools exist. The connections don’t.

This is where a purpose-built AI agent earns its keep. PropeLLM doesn’t sit alongside your stack — it replaces the stitching. Journalist research, media list building, pitch writing, distribution, response tracking, coverage monitoring, sentiment analysis, and ROI reporting all live inside the same agent, sharing context. When a journalist replies to a pitch, the agent knows. When coverage lands, the agent knows. When that coverage drives traffic to your site, the agent ties it to the pitch that started the conversation. The result isn’t just less software. It’s a feedback loop that gets smarter with every campaign you run.

Finally measuring PR like a growth channel

The other persistent frustration in communications is measurement. PR teams have always struggled to connect earned media to business results. Impressions and AVE (advertising value equivalency) are vanity metrics. Executives want to know what PR contributed to pipeline, traffic, and revenue.

This is another area where an integrated AI agent has a structural advantage over bolted-on tools. Because PropeLLM handles the full campaign lifecycle right from ideation to distribution to tracking, it can close the attribution loop. It ties earned media coverage to downstream outcomes: website traffic from backlinks, demo requests, conversions, and revenue. It tracks both direct attribution and modeled impact across channels.

For PR leaders who have spent years defending their budget, this is a meaningful shift. It moves the conversation from "what did PR cost this quarter" to "what did PR generate."

The deeper shift is in what counts as evidence. For decades, PR reporting leaned on impressions, reach, and AVE because nothing better was available. Those metrics don’t survive a serious CFO conversation, and PR teams know it: industry data shows that nearly half of PR professionals still struggle to prove campaign value, even as 89% say demonstrating impact is their top measurement goal. An integrated AI agent changes the math. Multi-touch attribution becomes possible because every pitch, every placement, and every backlink lives in the same system. Sentiment analysis is automatic, not a weekly export. Coverage gets weighted by quality and audience match, not raw volume. The reports your team produces stop describing what happened and start explaining why it mattered.

There’s a newer layer of measurement that matters even more in 2026: AI search visibility. As ChatGPT, Gemini, Perplexity, and Google’s AI Overviews become primary discovery channels, earned media has quietly become the single biggest driver of how brands show up in LLM answers. The articles that get cited in AI summaries are the same articles your PR team is working to land. That means a placement in a tier-one publication isn’t just a moment in a news cycle anymore — it’s training data for the models your future customers will ask about you. A purpose-built PR agent can track that footprint: which outlets are getting cited, which narratives are being picked up, and how your share of AI-generated answers is moving over time. General-purpose AI can’t measure its own influence on itself. A purpose-built agent can.

The teams that go agentic will pull ahead

Every industry transformation follows the same pattern. Early adopters gain a compounding advantage. Late movers spend twice as much to catch up, and some never do.

PR is at that inflection point right now. The teams that adopt AI agents purpose-built for communications won't just save time on media lists and pitch drafts. They'll operate with better intelligence, lower reputational risk, and measurable business impact. They'll run in days what used to take weeks. And they'll have data to prove it.

General-purpose AI was the wake-up call. Purpose-built AI agents are the actual alarm clock.

The question isn't whether PR will become AI-native. It's whether your team will be leading that transition or scrambling to keep up with the ones who are.

Frequently asked questions about AI agents in PR

What is an AI agent in PR?

An AI agent in PR is a system that takes a goal, like launching a media campaign, breaks it into steps, and executes them autonomously across the entire workflow. Unlike a chatbot that only responds to single prompts, an AI agent for public relations can research journalists, build media lists, write personalized pitches, send outreach, track responses, and measure business impact. It does the work, rather than just helping you draft text faster.

How is a purpose-built PR AI like PropeLLM different from ChatGPT?

General-purpose models like ChatGPT, Claude, and Gemini are trained on the public internet and excel at drafting text. PropeLLM is trained on PR-specific data that general models simply don’t have access to: over 25 million real pitches that have been sent to journalists, more than 500,000 verified journalist contacts, 150,000+ media outlets, and 2 billion+ articles. It also has direct access to live PR infrastructure — journalist databases, email tools, monitoring services — so it can execute, not just suggest.

Can AI agents really measure PR ROI?

Yes, when the agent handles the full campaign lifecycle in one place. Because pitching, sending, monitoring, and tracking all happen inside the same system, an AI agent like PropeLLM can connect earned media coverage to downstream business outcomes — referral traffic, demo requests, conversions, and revenue. This replaces vanity metrics like AVE and impressions with attribution that finance teams actually accept.

Will AI agents replace PR professionals?

No. AI agents handle the repetitive, time-consuming parts of PR work — list building, first-draft pitches, monitoring, and reporting — so practitioners can focus on strategy, relationship-building, creative storytelling, and judgment calls that require human context. The teams pulling ahead aren’t the ones replacing humans with AI. They’re the ones equipping experienced PR pros with agentic tools that let them operate at far greater scale.

Key Takeaways

• PR's bottleneck is execution, not creativity. Comms pros spend most of their day stitching together media lists, CRMs, and inboxes — while sales, support, and marketing have been automating with AI agents for years.
• Adoption is wide but shallow. 90% of PR teams have integrated AI in some form, yet only 13% call their operations "highly integrated" and just 12% have actually adopted AI agents. Awareness is far ahead of execution.
• General-purpose AI (ChatGPT, Claude, Gemini) only solves ~15% of the job. It drafts text well but can't access journalist databases, score reporter fit, personalize at scale, send from real inboxes, or measure outcomes.
• Copilot vs. autopilot is the key distinction. A copilot helps you write faster; an agent researches, plans, builds, personalizes, sends, monitors, and reports back. PR needs the latter.
• A real PR agent has to handle seven capabilities at once: strategy, research, targeting, personalization, distribution, monitoring, and attribution. General-purpose models do roughly one of them well.
• Purpose-built beats general-purpose because of the data. PropeLLM is trained on 25M+ real pitches, 500K+ verified journalist contacts, 150K+ outlets, and 2B+ articles — proprietary data no general model can replicate.
• Tool fragmentation is the real problem. Most teams run a disconnected stack of media databases, email platforms, monitoring services, and spreadsheets. An integrated agent replaces the stitching and creates a feedback loop that compounds with every campaign.
• Attribution finally becomes possible. When ideation, distribution, and tracking live in one system, PR can be measured like a growth channel — tying coverage to traffic, demos, conversions, and revenue instead of impressions and AVE.
• AI search visibility is the new measurement frontier. Earned media is now training data for ChatGPT, Gemini, Perplexity, and AI Overviews. A purpose-built agent can track which placements feed LLM answers; a general-purpose one can't measure its own influence.
• Humans aren't being replaced — they're being leveraged. Agents take the repetitive work (lists, first drafts, monitoring, reporting) so PR pros can focus on strategy, relationships, and judgment.

The window is now. Early adopters will compound an advantage; late movers will pay twice to catch up. General-purpose AI was the wake-up call — purpose-built agents are the alarm clock.