Inside Our Live Propel LM Demo: Running an Entire PR Campaign in One Chat
This week we hosted a live demo of Propel LM, and the room filled up fast. Within minutes, the chat was scrolling with hellos from Los Angeles, Austin, Charlotte, Baltimore, Philadelphia, Boise, London, India, and even Macedonia- a genuinely global crowd of PR and communications professionals who all share the same curiosity: what does AI actually look like when it's built for our job, not bolted on as an afterthought?
If you couldn't make it, this is the full recap of what we covered. And if you'd rather watch it unfold yourself, you can watch the complete recording for free here.
Why we built a model just for PR
Most AI tools today are general-purpose. You can ask them to write a poem, debug some code, or plan a vacation, which is precisely why they're rarely excellent at any single thing. They're generalists by design. For a discipline as specific as public relations, that generalism shows up in all the wrong places: media lists that don't reflect who's actually covering a beat, pitches that read like they were written by someone who has never pitched, and zero understanding of how journalists really behave.
Propel LM takes the opposite approach. It's the only vertical LLM built specifically for public relations and communications. Our founder and CEO, Zach Cutler, spent his entire career in PR, running an agency and feeling, day after day, the friction that everyone in this industry knows intimately. Hours lost to media research. Lists rebuilt from scratch for every campaign. Personalized pitches written one painstaking paragraph at a time. That frustration is what led to building the first CRM for PR years ago, and it's the same instinct that drove the team to build the most advanced AI in the industry.
What makes a vertical model possible is data, the right data, at a scale that general tools simply don't have. Propel LM is trained on more than 25 million real email pitches that have been sent through Propel's Gmail and Outlook plugins over the years, used in an aggregated and anonymized way. On top of that sits a media database of half a million journalist profiles, more than 100,000 outlet profiles, and 2 billion articles. As Zach put it during the session, the LLM is powerful because the data behind it is powerful. You can't prompt your way to that kind of domain knowledge; you have to train on it.
The practical upshot is that Propel LM doesn't just generate text that sounds like PR. It understands the workflow, discovery, list-building, pitching, distribution, monitoring, and reporting, and it brings every one of those steps into a single place.
One chat, the whole campaign
The centerpiece of the demo was watching an entire campaign come together end to end inside one chat experience. Not five tools stitched together with copy-paste in between, but one continuous conversation. Here's how it played out.
Finding the right journalists
We started with the simplest possible prompt: find journalists who cover AI for healthcare. The agent offers quick-start buttons for the most common tasks: launch a full campaign, find journalists, build a media list, and you can either click or simply type what you want. There's a microphone, too, so you can talk to Propel LM directly, and a full archive of your past chats on the left, exactly like you'd expect from any modern LLM.
What's different is where the results come from. Because Propel LM pulls from Propel's own media database rather than the open web, the journalists it surfaces aren't a best-guess scrape of public pages. Each result comes with verified contact information, email, often a phone number, and social profiles like LinkedIn and X. These are results you genuinely won't find in a general-purpose chatbot, because those tools don't have a proprietary media database underneath them.
Journalist intelligence
Click into any journalist's name, and you land on their full profile. You'll see their recent articles, their bio, and the media lists they already appear on. But the part that drew the strongest reaction was the pitching preferences panel.
Drawn from aggregated, anonymized data across the entire Propel user base, this panel tells you how a journalist actually behaves: how many pitches they've received, their open rate, their response rate, the topics that have earned the most engagement, and the best day and time to reach them. In the demo we looked at a real profile and could immediately see, for instance, a strong open rate paired with a more modest response rate, exactly the kind of nuance that changes how you'd approach someone.
This is the difference between hoping a pitch works and understanding why it will. Instead of guessing at the right angle and the right moment, you're working from evidence.
Refining the list with a strategy partner
Here's where Propel LM stops behaving like a database and starts behaving like a collaborator. After returning an initial list of targets, it doesn't just stop. It asks how you'd like to refine or expand the list. Working in AI for healthcare? It might suggest narrowing your focus to clinical AI, revenue cycle management, or medical imaging, and if you choose imaging, it goes a level deeper still, asking whether you want to zero in on radiology workflows or diagnostics performance.
It works in the other direction, too, offering to expand a list with another 30 to 50 relevant contacts when you want broader reach. Any traditional media database leaves this thinking entirely to you: you'd have to sit there and reason out which adjacent beats and outlets to add. Propel LM makes the suggestions, follows up with new ideas, and prompts you to think critically about your own strategy. It's the kind of interactive partner Zach said he wished he'd had when he was running his agency.
Checking your pitch history
Before anything goes out, the agent reviews your past activity. Ask whether you've pitched these journalists before, and it scans your historical pitching to give you a clear green light or a red flag. In the demo, it confirmed none of the new targets had been pitched yet, but flagged a handful who appeared on other lists, which is exactly the kind of context that prevents the awkward, credibility-damaging mistake of double-pitching the same reporter.
Personalizing at scale
This was the moment that earned the most live reaction. Propel LM reads each journalist's recent work, pulls out a relevant detail or two, and drafts a tailored introduction for every single contact on the list. You decide whether to draft for select contacts or for everyone, and the agent generates personalized intros across the board, for example, opening with a specific note about a journalist's recent coverage of a company acquisition and the angle they took on it, before flowing naturally into your pitch.
The goal is to show each journalist that you actually did your homework, that you read their work and aren't just blasting a template. And before any drafts are created, a safety screen lets you confirm that every first name looks correct. It's a small touch, but anyone who has ever sent "Hi [FIRST NAME]" to a journalist knows why it matters.
AI as a partner, not a replacement
A clear philosophy ran through the entire session: the human stays in control.
Propel LM never sends pitches automatically. Every personalized draft lands in your drafts folder for you to review, edit, and approve. This is a deliberate design choice, and Zach was direct about why. At his agency, the rule was simple and absolute. Every pitch went out with a personalized intro, no copy-paste blasts. The trouble was the math. A good personalized intro took 10 to 12 minutes per journalist. Multiply that across a full media list and a single campaign could consume the better part of a day before one email left the building.
Propel LM does that same groundwork in a couple of minutes for an entire list. But the practitioner still brings the judgment, makes the edits, and clicks send. As Zach framed it, the AI can get you 85, 90, even 95% of the way there, and that alone saves hours every week, but the final stretch belongs to you.
This isn't false modesty about what the technology can do. It's a genuine point of view about what the technology is for. The repetitive, redundant busywork, the list-building, the research, the first-draft personalization, is exactly what should be automated. The substantive, strategic, human parts of the job are exactly what shouldn't be. Done right, AI doesn't remove you from the process; it gives you more room to do the work only you can do.
Measuring what actually matters
PR has always struggled with proof. The demo showed how Propel LM helps close that gap.
Ask the agent to analyze your campaigns, and it returns a clean summary of performance, pitches sent, opens, responses, articles published, unique monthly visitors, and backlink stats, and it can break those numbers down by account or by campaign. It's the reporting layer that turns activity into a story you can tell internally or to a client.
Beyond campaign metrics, Propel LM connects to business analytics, letting you track both the direct attribution and the broader correlation between your PR efforts and real business outcomes. A Google Analytics integration ties it all together, so the impact of communications work is easier to demonstrate to the people who hold the budget. For an industry that's spent decades fighting to prove its value, that's a meaningful shift.
Questions from the room
The live Q&A surfaced some of the most practical questions, and they're worth sharing.
How does it pull the personalized intro? Because Propel maintains a profile and a set of recent articles for essentially every journalist you can think of, the agent reads those recent pieces, pulls out a nugget or two, and weaves it into the opening of your pitch before transitioning into your actual message. The effect is an intro that proves you read their work rather than one that reveals you didn't.
How receptive are journalists to AI-assisted pitches? Propel has seen a meaningful rise in open and response rates among users who send personalized intros through Propel LM. Part of that comes from humanization built into the drafts so they don't read as automated or obviously AI-generated, and part of it comes from the human review step, where practitioners add their own voice before sending.
Is there an affordable option for small brands, agencies, and freelancers? Yes. There's a range of plans. The standard plan includes a limited allotment of Propel LM credits, which is enough to run some campaigns and get real value, and additional pricing is available for those who want a larger volume of usage. The team is happy to walk through the options on a one-on-one call.
Is Propel LM an add-on to existing plans? Effectively, yes, the standard plan comes with a limited number of credits built in, with the ability to scale up usage through additional pricing as your needs grow.
An exciting moment for our industry
Zach closed the session not as a founder making a pitch, but as someone who has spent his entire adult life in PR. His view is that this is one of the most exciting moments the industry has had in a long time, and that the future belongs to practitioners who get onto the AI curve rather than away from it.
The optimism is grounded. He doesn't believe AI will replace communications professionals; he believes it will multiply their output. To stay relevant, productive, and successful, comms teams will need to adopt these tools, and the good news is that with something built for the work, that adoption isn't scary or technical. It's intuitive. The AI is there as a partner, handling the busywork so you can focus on strategy, relationships, and results.
Watch the recording
If you want to see Propel LM run a campaign from a single prompt all the way to a folder of personalized, ready-to-review drafts, and to watch the journalist intelligence, list refinement, and analytics in action, the full demo is worth your time.
👉 Watch the recording for free
To talk through pricing, see more of the agent, or book a one-on-one walkthrough, reach out to us at support@propelmypr.com. We're already lining up the next sessions, and we'd love to have you join us.