Co-founder at Notably • PR & Comms Strategy • Chat w/ me if you want press that drives business results: bit.ly/NotablyMeet 👋
Boston, MA
- Tweets 4,174
- Following 1,399
- Followers 5,926
- Likes 3,082
Pinned Tweet
I run a PR agency that lands our clients in TechCrunch, CNBC, & other world class publications on a consistent basis.
My superpower? I can make anything sound newsworthy to the journalist I'm pitching.
These are the 2 golden rules I follow that 10x my placement rate: 👇
1/8 🧵
22
32
415
92,234
Please, please, please can we stop workslopping each other in PR?!
You know what I mean—and no, it’s not necessarily em-dashes—but that distinctly AI-esque, perfectly-written-but-saying-nothing-of-substance… slop.
Look, I’m all-in on AI. But what I mean by that is automations like monitoring Qwoted and my clients’ beats so I can get them placements, not delegating my thinking.
When you tell AI to draft a pitch, email, or anything in writing meant for other humans, you’re abdicating your critical thinking to next-token-prediction software that converges on a mediocre mean.
If the idea is original and you’re just getting help from AI to make it intelligible, fine. But otherwise, all you’re doing is plagiarizing from AI and forcing the recipient to decipher something whose meaning cannot ultimately be deciphered.
Here’s what the latest research says about workslop:
- 40% of full-time US workers surveyed had received it in the previous month
- 53% admitted sending it to colleagues themselves
- About half of workslop receivers saw the sender as less creative, capable, and reliable
And… it gets worse:
- Each instance took nearly 2 hours to deal with
- 53% of recipients were annoyed, 38% confused, and 22% offended
- 32% were less likely to want to work with that person again
And seriously, in an age where more and more people are giving up taste, original and critical thinking, creativity to AI… why would you want to surrender your greatest advantage as a human?!
2
1
9
540
Remember when I said PR was getting a software upgrade? This. This is what I meant.
652,200 people watched my jev news scanning demo last week.
today it's a real product: the stories you can jump on every morning, the angle you can take, and the journalists already covering it.
delivered to your email or team slack, every morning 7am.
in my demo jev read 384 headlines in 24.9 seconds for 19 cents. opus 5, same feed, same clock, got through 4/384 at 390x the cost per headline.
on friday i shipped a slim open source version already in newsjack.sh v0.1.19.
so why am i also shipping it again? three things i learned building the free one:
1. the news is shared
every monitor reads the same news. the oss version fetches your feeds, on your machine, every run. hosted, a headline gets fetched once, judged once, and dealt to every company that's a good fit.
> the sources multiply
we started with just news. and X and reddit are next. in the oss profile, X monitoring asks you for WOEIDs and an X api key that can reach the trends endpoint. that's a lot of setup to maintain and most people I talk to don't know how to do that.
> the local states is messy
I wrote the cli in go. to make a local agent remember what it already saw across runs, i had to bolt a sqlite database onto it. a lot of wrestling for "remember yesterday's news" while keeping it cheap for people who run it, pooling the cost just makes sense.
so yesterday and this morning i ran the new product through 14 of @prcarly's clients.
12,104 headlines read. 238 stories worth a reaction. 55 ready to run today.
carly's take: "um, yeah. this is insane."
how it works:
- put in the website of your company (or your clients)
- it reads the site and works out your beats, competitors, what to search for
- every headline on the wire gets judged by jev first
- survivors get the expensive treatment: an angle, and one click to the journalists already primed to continue the story
- every morning at 7, on your desk. email or slack
the biggest takeaway from jev for me is that the cost and speed have improved by 2 orders of magnitudes - which pushes the algorithm around the stack to be rewritten.
400 google alerts across 20 tabs was the old way to get scale. accuracy or scale, pick one.
with jev reading everything is a rounding error now, so we priced it like one: we're just bundling radar into every paid medialyst plan, every project, starting today. the pricing page didn't change.
the oss stays open. newsjack.sh has the skills, the cli, and jev mode. if you want to run it yourself, run it yourself.
if you want the managed version, first run is free: medialyst.ai/radar
1
2
477
Does pitching ever feel a bit like fishing? Drop your line (pitch) in the water and hope (with all your might) that you get a nibble?
Well, that's why I just built a pitching game called The Hook.
You get a brief with:
- the story you’re trying to land
- a reporter and their preferences
- and pitching tips for the journalist (e.g., "Favorite pitches include: what changed, with plans and completed actions clearly distinguished.").
Then you pick the winning* subject line out of multiple choices.
*The catch (or what actually makes this GOOD; and yes, that was a pun): this is an already-published piece with an already-published headline.
If you get it right, you win points and then there’s a fun mini reel-it-in game where you pull in a fish ranging from common to mythical.
But the main idea is to train the muscle that helps you write the kind of headline that’ll win you placements when you pitch… grounded in what was already published.
Crowdsourcing time:
That sound interesting to you? Would you play?
And, what else do you think will make this not only fun, but genuinely helpful for PR pros to start landing more pieces?
Lmk. It’s been insanely fun and exciting building this.
2
11
390
PR is about to get a major software update
Jev is INSANE. 🤯
in 24.9 seconds it read 384 news from this morning and told 15 brands which stories to hop onto today, for $0.19.
Claude Opus 5, running on the same feed at the same time, got through 4/384 and cost $0.77.
per headline that is ~390x cheaper, and the answer comes back before you finish reading the headline yourself.
there's going to be so many ways for JEV to help you find trending stories, find journalists covering it, and get press coverage.
everything in the video open-sourced here: newsjack.sh (in the demo/ folder, together with 30+ skills to turn your agent into a PR team)
5
7
134
50,043
Nine weeks ago, I decided that even though I’ve got a 5-month-old keeping me up at night, a 2-year-old keeping me on my toes, and a PR agency keeping me up in the day… going to start building things with Claude.
Come what may.
I wouldn’t say I’ve got AI psychosis (no grand, delusional ambitions of winning any Nobel prizes), but I will say, without exaggeration, the following…
First, I’ll describe how I’m doing it, and next, I’ll tell you what I’ve been building.
HOW
The beauty of AI is that it’s a kind of meta software. I remember teaching myself HTML 20 years ago, poring over badly-written manuals, scouring forums, just to understand how to make silly little websites for fun.
With AI, you can literally say: “Explain in simple terms how to (whatever it is you’re trying to do.” Not only that, but I went a step further, had it ingest my emails/correspondences, and asked it to do it in MY own language 😎
Software that teaches you how to software. How meta-sick is that?!
WHAT
1. Fed Claude everything I've ever written (emails, Slack, articles, podcasts) and built a custom skill that writes in my voice automatically
2. Connected Granola to Claude so that action items from every call go straight to my Gmail tasks
3. Set up real-time Qwoted monitoring so that the second a journalist query matches one of my clients, I get an alert (result: roughly 10 extra media hits per month)
4. During the Apple event, I had Claude check a live transcript every five minutes against my client roster while doing other work. It flagged a Siri announcement as relevant to my voice-AI client and I had pitches out presto
5. Building a pitching game that pulls real news stories and challenges PR pros to match the best subject line to a pitch brief (more on this soon!)
AND?!
I honestly believe that if you’re not tinkering with AI and building things, then you're kind of screwed.
Plus… I can’t remember the last time I was so fired up and enjoying myself this much…
7
16
1,133
I’m reading an 803-page, 4.34 lb-doorstopper-of-a-book (or “concusser-of-a-book” if you’re more of a dark humor aficionado, like me) titled: “The Behavior Operations Manual: Neuro-Cognitive Intelligence”.
Set me back $112, I’m only up to page 160 in a month, and… I’m absolutely digging it.
I was a psych major at UCLA. Always been obsessed with understanding people.
And this book is a manual on… the human psyche. Why people behave the way they do, how to pick up on what folks DON’T say, and how to earn trust and credibility.
I'm reading it because my whole job is people. Clients, reporters, editors... the better I understand where they're coming from, the better I can communicate with them.
A few thoughts:
— I don’t believe people are easy to understand, but I don’t think they’re opaque, black boxes either. And I meannnnn I don’t think I would be running a boutique PR agency that averages 10+ tier-1 placements a month for our clients if I hadn’t spent the last 15 years trying to understand the human psyche better. It really does help, I think.
— I use a ton of AI but I don’t think you can lazily prompt (“yO ChAt/cLaUde: tEaCh mE tHe WaY of PeOplE” 😂) your way to knowledge… nothing replaces the slow but effective labor of reading long-form.
— Being a good PR pro certainly requires technical expertise (writing, research, etc.), but the most important skill is fathoming the mysteries of the human mind. Do as much of that as you can.
P.S.
What are you reading these days?
6
13
445
The “AI-will-replace-PR” hype is backwards. AI is actually tilting the playing field further toward experienced pros. Anyone who thought a ChatGPT subscription was a good PR strategy is finding this out the hard way.
Three dangerous assumptions people make about AI:
Assumption 1: “AI can write our pitches.”
It can write a pitch. It cannot write a pitch a journalist will actually use. Unless you're Apple or Nike, announcements aren't news, but AI doesn't know that. It produces confident-looking promotional garbage that reporters delete on sight.
Assumption 2: “Domain expertise can be uploaded.”
Feeding a chatbot every internal document you have doesn't teach it what a journalist will actually cover. The story is never your product update. It's the news hook or the trend, and your company has an expert who can speak to it. Recognizing which angle a reporter will bite on, and which one they'll delete without a reply, comes from judgment, not a database.
Assumption 3: “AI can evaluate its own output.”
This is the snake eating its own tail. The instinct to spot red flags, or fancy words devoid of meaning, can only be earned through failure. I've watched enough of my own pitches crash and burn to catch problems a PR bot never will.
The reality of AI in PR is that a seasoned pro using AI can develop strategies in hours that used to take weeks. But someone without PR experience using the same tool, ships work that lands nowhere.
My advice is to stop worrying that AI is going to replace you. Instead, spend your mental energy on your critical judgment.
…and let AI compound it.
4
2
14
744
"Hey, you found one that I didn't have. That's a win. I count that as a big win."
That's what I told @medialyst's builder @elvissun on Monday. Four weeks ago I told him his AI-generated media list was kinda garbage and gave it an F.
This week it scored a 66.
Here's how that happened.
The test hasn't changed since July: brief the AI the way I'd brief someone on my own team, which is to say not much, and see whether it independently arrives at my exact list.
What changed Monday was the brief and the process. New story -- a real client announcement of mine going out this week, a niche NFL angle, twelve journalists on my hand-built list, which is about as precise as my lists get. And instead of grading a list after the fact, I typed in the prompt myself and watched it build live.
Then I cross-checked rows as they landed.
Partway down I stopped at a name I didn't recognize. Went back to my own sheet. She wasn't on it. So I read her coverage, and told Elvis to keep her, because she belonged on MY list.
Four weeks of this score have asked one question: how many of my journalists can the tool find? Monday the question ran the other way for the first time.
The tool found 7 of my 12 journalists. Then I checked every email address it served against my own and it went 7 for 7. Of the 28 names it handed me in total, I'd keep 9.
That's a 66 out of 100, up 15 points from 51 last week, and the first time since this started that the Carly Score™ isn't an F.
A twelve-name list is a harder test than it sounds, by the way. On a 43-name brief you can find a handful and still look respectable. On twelve there's nowhere to hide, and twelve is exactly how precise a real niche campaign has to be.
What's still broken is the five it missed. That's the gap between a good week and a list I'd hand to a client without checking every row myself, and Elvis knows where the bar is.
Still grading a fresh list every Thursday. Still publishing the number whether it's flattering or not. It was a 27.9 in July.
Live scorecard in the comments 📈
🤯 the biggest secret to building my AI+PR agents has been hiding in plain sight for 11 months.
"Same core story. Four different coverage opportunities."
that's from a trending linkedin post I bookmarked last september. a publicist treating one announcement as four audience angles:
→ security trades get the stopping-fraud angle
→ HR publications get the employee-onboarding angle
→ pop culture and tech media get the fighting-deepfakes angle
→ financial press get the protecting-banking-customers angle
I read all of that last september. nodded, noted, forgot it.
which is what everyone does with their bookmarks...
then 3 weeks ago @prcarly became our advisor and started grading our agent against the media lists she builds by hand. first grade: 43 journalists on hers, our agent only found 7 of them. F.
so I spent this week trying to encode her method into subagents. halfway through, I googled "multi-channel storytelling approaches" just to see what was already out there.
first result. that post from last september.
hers.
written months before we ever spoke.
so I had effectively brought on an advisor to teach me something I had already read, for free, and not understood.
and now I know why I didn't understand it at the time:
someone who works in PR reads that post and sees story angles. an AI engineer like me reads it and sees a fan-out of subagents. (which is exactly how everyone in the X bubble thinks but also the wrong half to notice first)
so two things shipped out of this for @medialyst today:
1. angles. your news gets broken into several distinct angles instead of one, and you chat with the plan before it builds anything.
2. deep dive agents. one research subagent per angle, each in its own container, 5 running at once. each reads what it finds, picks its next query from what it just read, and keeps pulling until the thread dies.
I took her biometric auth example for a real run:
↳ started with the fraud prevention angle
↳ added 5 journalists
↳ searched "injection attack"
↳ added 8 journalists
↳ searched "voice cloning fraud scam"
↳ found FBI's $262M account takeover alert
↳ added 9 journalists who covered that report
↳ kept searching "credential stuffing attack"
↳ ... (kept going for 8 minutes)
↳ 32 from the security angle, 84 more from the other three
fable 5 and gpt 5.6 sol ultra both couldn't solve this.
but the same model taught the methods cracked exactly this:
our eval this week went 42.2 → 51.3
still an F but we're moving in the right direction
3
2,974
For over two years I've been tracking how AI search is reshaping PR. Here's a summary of the most recent data PR Pros need to know:
1. 84% of AI citations come from earned media.
@MuckRack has now analyzed 25 million+ cited links across ChatGPT, Claude, and Gemini. That number has held between 82–89% across every edition of their report since July 2025. Earned media is the new baseline for AI visibility.
2. Paid and advertorial content accounts for 0.3% of AI citations.
Brands have spent a decade buying their way into earned-looking content (native ads, sponsored posts, advertorials dressed up as editorial). AI came along and decided it’s not citing any of that.
3. There's only a 2% overlap between the journalists PR teams pitch most and the journalists AI engines cite most.
Most media lists were built for traditional SEO and human readers. The new reality requires two different media lists, one for human credibility, the second for AI citation, and clarity on where they overlap.
4. Industry trend queries drive journalism citations at 2x the rate of how-to queries.
When someone asks "what's happening in [your industry]," AI reaches for news coverage. When they ask "how do I do X," it leans on owned content. If your buyers are doing research before making decisions, earned media is shaping their answers.
5. 57% of journalism citations are from articles published in the last 12 months.
Recency bias is accelerating. If your goal is citation share, one big placement per year won’t cut it. You need a consistent cadence of coverage to stay relevant.
6. AI search visitors convert at 4.4x the rate of traditional organic search visitors.
Per @Semrush. AI referral traffic grew 527% last year, and the channel is quickly catching up with traditional search in volume, but already surpassing it in conversion value.
The data in this list has been consistent for a year.
The only variable is whether your strategy has caught up.
6
2
9
1,438
📌 Sources:
May 2026 Muck Rack Generative Pulse report: generativepulse.ai/report/
Muck Rack December 2025 edition: globenewswire.com/news-relea…
Semrush AI search study: semrush.com/blog/ai-search-s…
98
this is a no brainer to submit for since it's free! Easy win for your clients, PR friends.
Waterstone Human Capital announced the inaugural 2026 America’s Most Admired™ Culture & CEO program this morning, extending more than two decades of evaluating how culture contributes to business success.
Read more on Meredith & The Media: meredithandthemedia.substack…
Nominate an eligible client, company or partner: americasmostadmired.com.
2
482
To actually get a response, every follow-up email needs one concrete job.
"Just checking in" doesn’t cut it.
Steal this 3-email sequence that I use for evergreen pitches. It landed my clients in Fast Company, Business Insider, and Newsweek last month:
EMAIL 1 (48 hours after initial pitch)
JOB: Add a new data point
SUBJECT: Quick follow-up: [Topic] story
Hi [Name],
Wanted to circle back on the [topic] story I pitched [when]. I know your plate is always full, so no worries if you missed it.
I found this [specific data point/study/trend] that adds another layer to the story—thought it might be interesting for your [specific section/beat].
No pressure if timing isn't right. Or, if you're working on something else where [Expert name, why they matter] can be of use, just let me know.
[Your name]
EMAIL 2 (a week after first follow-up)
JOB: Offer a new angle
SUBJECT: Different angle: [New angle on same topic]
Hi [Name],
I know I reached out about [original angle], but wanted to float this by you instead.
What if instead of focusing on [original angle], [expert] looked at [completely different angle on same topic]? Especially with [recent news hook/trend], this could be timely for [specific upcoming issue/section].
[Expert name] could speak to both the [specific aspect] and [specific aspect].
Worth exploring?
[Your name]
EMAIL 3 (a week after second follow-up)
JOB: Preserve the relationship
SUBJECT: Last check-in + here if you need me
Hi [Name],
I know the [topic] story probably isn't a fit right now -- no worries at all.
That being said, please keep [expert name] in mind for future [relevant beat] stories.
They're always great for quick thoughts on [specific expertise areas].
Talk soon,
[Your name]
___
TAKEAWAY
Be persistent and add value at every touchpoint. If a follow-up doesn't have a job (new data, new angle, or a clean handshake), don’t bother. It’s not worth sending.
AMA in the comments 👇
11
1,002
“Why should anyone care?” It’s what I ask myself before writing any headline, pitch, email, or post. I learned to do this back in college when I ran a blog called Featured Creature. It was a passion project dedicated to weird, endangered species that nobody had ever heard of.
There was a fuzzy white caterpillar I wanted to write about. The first headline I wrote: "White caterpillar lives in the rainforest and helps other animals survive." But then I looked at that headline and saw it… no one but me is going to care.
I had to make my little crawly guy relatable. So I tried a different door. I reframed the story around the caterpillar's wild white hair. The new headline read: "The Albert Einstein of Caterpillars."
The article took off. People started sharing the story about a wild haired caterpillar with their friends. My cater-pillar cata-pulted the blog to over 1M monthly visitors. What didn’t matter to people before was suddenly relevant.
Which is exactly what we’re trying to do with PR.
At @NotablyPR, most of our clients aren't household names. They're mid-market companies competing against brands people already recognize. Some of them work in industries most folks have never heard of.
So here’s what the Albert Einstein caterpillar taught me about making my clients relevant:
1. The subject rarely sells itself.
Reporters aren't sitting around waiting to cover your product roadmap. You have to find the angle that fits what their readers already care about.
2. A great frame is a door.
Calling that caterpillar "Albert Einstein" had nothing to do with the subject matter in the article. But it was the most relatable way into the story, and made everything that followed more interesting.
3. Constraints aren’t an excuse.
Working with obscure animals forced me to develop storytelling instincts I still use daily. Every caterpillar has an Einstein in it somewhere. Your job is to keep looking until you find it.
I’d still be writing about obscure animals if I had more time… thinking maybe I’ll start a Substack one of these days.
And the most fun part… will be making you want to read about them.
3
5
372
🐛 Here’s the link to the post for anyone curious: featuredcreature.com/so-this…
79
This is why 9 out of 10 pieces I place come from reporters I've never spoken to before ⤵
A fintech client announced a healthcare partnership last quarter and I immediately started pitching healthcare reporters cold. Next quarter, if the same client drops research on operational efficiency, I'll pivot to business media, and reach out cold again.
Now, don’t come at me: I'm not saying relationships don't matter. They do. Just not in the way most people assume. What existing relationships DO get you is name recognition, credit for having delivered before, and a higher chance your email gets opened. What they DON'T get you is priority, more willingness from reporters to consider your story, or guaranteed coverage.
Whether the reporter recognizes you or not, your pitch has to earn its way in.
Three more reasons the majority of my outreach is cold: Now more than ever, the reporters I had great relationships with move to a new outlet (or a new beat). Or the story itself pulls me into a beat I've never covered before. Lastly, I’m often pitching a completely new publication (usually a Substack or podcast that didn't exist last year).
Bottom line:
Relationships are nice. We're all human. But to get coverage you need storytelling that's relevant to the reporter's audience. Nothing else matters more than that.
1
6
395
Most pitches fail before the journalist finishes reading the subject line. At @NotablyPR we’ve refined our process for brainstorming pitches that land.
It starts with getting clear on the desired impact, then finding the specific angle readers want to read and journalists want to write.
Here are the steps we use behind-the-scenes to go from 0 to placed:
1. Start with the business goal.
Your client's announcement might be genuinely great news. But the announcement isn’t the reason for the pitch. Figure out the business outcome you're driving toward. That will tell you what kind of readers you ultimately need to reach.
2. Take inventory of what only your client has.
Proprietary data, internal metrics, or an expert who watches something nobody else is watching. The strongest angles rest on something a reporter cannot get anywhere else.
2. Do the journalists’ thinking for them.
Map the arc of recent coverage and pitch the next chapter before it exists. Or take a story already dominating the news cycle and find the layer nobody has reported on yet. When DeepSeek's breakthrough was everywhere, our client spotted that a trademark squatter had beaten them to the USPTO by 31 hours. That story landed them in TechCrunch.
4. Ask what the reader needs to hear right now.
Find the intersection between the problem your client solves and something already moving in the news. "We're launching project management software for remote teams" is a product description. "Windowed work lets employees work in focused bursts around caregiving" is a story. The second version landed Business Insider, TODAY, and Fortune.
5. Write the headline before anything else.
Write from the perspective of a reader who has no idea your client exists. The headline should work whether your brand is the one featured in the story or not. Then reverse-engineer your pitch from that dream headline.
6. Pressure-test for conflict, tension, and novelty.
If the angle doesn't challenge conventional wisdom or surprise the reader, it goes back in the oven.
The thread connecting all six steps: every angle starts with a specific reader and works backwards to the client.
Never the other way around.
Any steps I missed? Let me know in the comments ⤵
1
4
343
I dropped a stat mid-conversation with the EIC of The Deep View, and he took the exclusive on the spot.
I was on a call with @JasonHiner talking PR and AI when I mentioned it: for the first time ever, almost half of all executive roles now require AI literacy.
The stat came from our client @LaddersHQ's analysis of over 2 million job listings paying above $100K.
Jason said that was exactly the kind of evidence-based AI news they're looking to write about, so I offered him the exclusive.
Moments like this are why I love PR. Congrats to Ladders, their data did the pitching for me. (Check out the full article in the comments.)
2
2
4
603