Almost Timely News: 🗞️ Generative AI Use Cases in PR (2025-07-06)

If your PR firm isn't doing this, you need a new one

Christopher S. Penn

Jul 06, 2025

Almost Timely News: 🗞️ Generative AI Use Cases in PR (2025-07-06) :: View in Browser

The Big Plug

👉 My new book, Almost Timeless: 48 Foundation Principles of Generative AI is now available!

Content Authenticity Statement

100% of this week's newsletter was generated by me, the human. You will see bountiful AI outputs in the video. Learn why this kind of disclosure is a good idea and might be required for anyone doing business in any capacity with the EU in the near future.

Watch This Newsletter On YouTube 📺

Click here for the video 📺 version of this newsletter on YouTube »

Click here for an MP3 audio 🎧 only version »

What's On My Mind: Generative AI Use Cases in PR

This week, let’s tackle AI use cases in PR and public relations. I had the great joy and privilege of keynoting ant Ukraine’s Tech PR School this week, supporting Ukraine’s PR community. I’ve been an ardent supporter of Ukraine’s people and industry for years now, as evidenced by the fact that the last ad spot in this newsletter (valued at USD 25000 a quarter) has been dedicated to United24 since the first days of the unlawful Russian invasion of Ukraine.

That said, I wanted to share some of the use cases more broadly so you could get a sense of the ways PR and communicators could be using AI more skillfully.

Before we get into specific use cases, I will say this as a blanket statement I feel confident in. If you are a PR firm or team and you are not paying the $20 a month per seat to give your team access to Deep Research tools (Gemini Deep Research, OpenAI Deep Research, etc.) you have lost the plot entirely and you are behind. I rarely say that about AI, because it’s such a loaded phrase, but in this case it’s true. If your PR team does not have access to AI-powered Deep Research tools, you are behind very, very badly and borderline irrelevant.

Now, let’s talk turkey. As I say in my keynotes, there are seven major use case categories for generative AI overall, and you can learn them in much greater detail in my Generative AI Use Cases for Marketers course:

Extraction: pull data out of other data
Classification: organize your data
Summarization: turn big data into small data
Question answering: ask questions of your data
Rewriting: turn one form of data into another
Synthesis: merge small data into big data
Generation: make new data from your old data

Let’s walk through the PR-specific use cases I taught in the keynote. I strongly recommend you watch the video version and follow along!

Extraction

One of the things that is so complicated and difficult about today’s public systems is that data is literally everywhere and usually in formats we can’t access easily. The good news is that today’s state of the art models can also see, what are known as vision language models (VLMs).

Probably the most well known VLM is Google’s Gemini, but literally every state of the art model has SOME vision capabilities. ChatGPT and Claude can see screenshots, while Gemini can ingest videos.

Which means if you can see it, you can capture it.

Here’s a very simple example I demonstrated in my keynote. Suppose you wanted to extract data from your Google Business Reviews. Sure, you could buy software to do that, and many people have, but why? Instead, turn on any screen recorder - heck, even your phone’s camera, and just record yourself scrolling through the reviews. Then load the video into a tool like Gemini and ask it to transcribe exactly what it sees into a machine-usable format.

The prompt I usually use is straightforward and goes a bit like this:

Transcribe the Google Business Reviews in this video into JSON format. The required keys and values of the JSON format are: reviewer name, review date, review stars count, review photo (true/false), review photo subject (if true), venue response (true/false), review verbatim text, venue response verbatim text.

If you’re unfamiliar, JSON is a machine language format for data. It’s one of the most common data formats and one that AI understands especially well - far better than spreadsheets, because JSON is language and not tabular data. We’ll keep our data in this format for now because the other use cases depend on it; at the end of the process, you can always ask AI to turn it into something more human-usable like a CSV table.

For the demo, I used a famous Kyiv restaurant, Ostannya Barykada, the Last Barricade. It’s a unique restaurant and museum combination dedicated to the recent history in Ukraine, from Euromaidan and the illegal annexation of Crimea by Russia in 2014 to the current illegal invasion.

If you can see it, AI can help you transcribe and convert it. This is incredibly valuable, especially for data sources that are legendary for blocking any kind of scraping.

Classification

The next use case is classification. This is where we organize our data, prepare it for use elsewhere. In the context of Ostannya Barykada, maybe we don’t trust how people use their star ratings. After all, people have a tendency to rate things in extremes, either 5 stars or 1 star. As a business, I’d want to know if the stars matched how people actually felt when they left reviews.

So what we could do is have our AI tool add this to our dataset, right in the JSON file (hence why we kept it in JSON). I could use a prompt like this:

In the JSON dataset, use the review_text_verbatim key and value fields to perform sentiment analysis. Score each review solely based on the contents of review_text_verbatim on a floating point scale of 1-5 with 2 decimals of precision. Add this to the JSON as a new key value pair named review_sentiment_score.

What we end up with is a nice score of the sentiment itself that we can analyze externally. Never, ever attempt to have generative AI do math (with one exception). It doesn’t go well because it’s fundamentally incapable of it. You can’t ask it to do the math here - but you can ask it to write the code to do math.

Summarization

Next, let’s take our data and summarize it. Summarization is one of the most powerful, simplest use cases for generative AI. It takes big data and turns it into small data, which is super useful for communicating information in a compact fashion.

Suppose we wanted to communicate to our stakeholders, the owners of Ostannya Barykada, how their public perception is going. That’s the job of a communicator and PR professional, to show how their work is landing with the public (the public in public relations, after all). We now have sentiment scores, reviews, review stars, and time. How could we do this?

Well first, we’d want to understand what the key priorities of our client are. What does a restauranteur care about? Butts in seats. That’s it. Butts in seats, paying to eat food. Literally nothing else matters, because if your tables aren’t full at each seating, you’re not making money.

To understand why our tables might or might not be full, we can use the review data we’ve just gathered. We’d want to summarize it with generative AI and some math code (don’t worry, you don’t have to code) so we could show our results.

Here’s a simple example prompt we might use.

Using the JSON dataset and the Simple Statistics JS library at https://cdnjs.cloudflare.com/ajax/libs/simple-statistics/7.8.8/simple-statistics.min.js, let’s build a public perception report for the owners of Ostannya Barykada based on the Google reviews we’ve been analyzing. The report will be in HTML, CSS, Tailwind, and use the Simple Statistics JS library (read the docs at https://simple-statistics.github.io/docs/ ) to compute a rolling mean of both star reviews and sentiment. The report should highlight three things the restaurant does well, based on the review verbatims, three things the restaurant needs to improve on, and should show the two charts as bar charts with the rolling means as lines superimposed on the bars. The report should use the colors of Ukraine and use the Noto Sans font for its body copy and Noto Serif for its headings. Think through what would resonate with a restauranteur to help them make decisions that will fill seats at their seatings.

We’re not having AI do the math. No. We’re having AI write the code to do the math, because there are literally thousands of useful pieces of code in libraries that are freely available which can do the math for us - correctly - and put it in the output. And we're not writing the code - AI is. We're just copy pasting the results. That's as technical as this gets.

Quite a useful summary.

Question Answering

The fourth category is question answering, asking questions of our data. This is the use case that every PR practitioner should be most fluent in. Let’s say we’ve done our analysis of Ostannya Barykada and we’ve concluded we need to do a public relations campaign to get the word out about the restaurant. For that, we need to understand Kyiv’s media scene, to learn how the public finds restaurants and chooses them. For that, we can use our Deep Research tools.

Deep Research tools really are just AI agents in disguise. They’re billed as a single purpose tool, but what they really are is a combination of extractor, rewriter, and summarizer all in one. They crawl the web and public Internet based on our requests and come back with incredibly rich, detailed data.

Let’s start with a simple query to understand the landscape. We’ll use the Trust Insights CASINO Deep Research Prompt Framework to help with this. As I said at the top, if your PR firm/agency isn’t using a process like this, fire them immediately.

Here’s an example prompt.

Let’s build a prompt to understand the Kyiv food media landscape. I represent Ostannya Barykada, the restaurant in Kyiv, and we want to reach more media and more of the public to let them know about our unique dining experience as both restaurant and museum. Our research goal is to find up to 10 traditional media outlets and up to 10 new media outlets (food YouTubers, Tiktokers, Telegram influencers, etc.) in the Kyiv scene that we can screen and later reach out to. We need to know who they are, who their audiences serve. Ostannya Barykada serves two primary audiences: local Ukrainians who love 100% Ukrainian sourced food (no external food or drink from outside Ukraine) and tourists looking for authentic Ukrainian food. Using the Trust Insights CASINO prompt framework, ask me one question at a time until you have enough information to build the full CASINO prompt.

Your AI of choice will walk you through the CASINO prompt framework, asking you questions, until it has enough information to build the prompt. From there, take the prompt into the Deep Research tool of your choice and have it execute the report.

Why this extra step? Why not put it straight into a Deep Research tool? Because for many services, you get a limited number of Deep Research reports per month. OpenAI, for example, on the Plus plan limits you to 10 full reports and 15 abridged reports a month. You don’t want to burn those scarce reports on bad queries. Second, by having you do this process, it forces you to think more clearly about the task in general. We want AI to help us improve our thinking, not do our thinking for us.

Once we have our media list, we can then potentially issue follow on reports. We might want to ask about a specific journalist or influencer and how to reach them best, what pitch angles they're receptive to, etc.

For example, in our survey of the Kyiv media space, we find traditional media outlets like The Village Україна (Ukraine), a local news site in Kyiv. We also find Yevhen Klopotenko, a YouTube and Instagram influencer and foodie. He seems, based on the research, like a prime candidate, so we'd commission a Deep Research report on him. You'd follow the exact same CASINO framework and process.

Rewriting

That brings us to the next category. Suppose as PR professionals, we want our pitches to land. Instead of taking the firehose approach and blasting every publication and influencer with our pitch - and my inbox is full of these - we instead crafted pitches that were highly likely to land.

In our profile of Yevhen Klopotenko, we specifically asked for not only what channels he has, but what angles and even a personality analysis of him based on how he speaks and writes on his channels. The next step should be obvious: using a language model to take our stock pitch and automatically customize it for him.

We'd start with our generic pitch. Let's say the restaurant wants to publicize its award-winning sunny dumplings, made with local potatoes and fried onion cream. The generic news release is still useful; however, we want to tailor it to our most influential news outlets.

With a straightforward prompt, we'd take our news release and ask our AI of choice to calibrate it for that person or outlet, something along the lines of:

Let's tailor and customize the attached news release to maximally appeal to food influencer Yevhen Klopotenko. Using the research report, identify which channel would be the most appropriate for our news, what themes he cares about and how we could pitch to those themes, the kinds of language most likely to resonate with him, and three different hooks and angles. Pay attention to language that will land as well as what to avoid. Present your pitch angles in descending order by alignment and fit to Klopotenko's priorities.

After you do the analysis, follow up by having it customize the pitch. This takes a generic pitch that might or might not land and tunes it to fit as best as possible.

Synthesis

One of the things AI tools do better than anything else is find patterns. especially in disparate data sources or in data that's much larger than we typically deal with. Suppose we have a client like Ostannya Barykada. They want to understand the overall market environment they operate in.

How might we gather this information? We could certainly - and should - repeat the process of extraction to obtain reviews for competing restaurants and venues in the area. We'd take that data, run it through the same processing steps as we did for ourselves, and then compare and contrast what's working at other restaurants with what's working at ours.

We might also want to use traditional PR tools to gather coverage about the market we're operating in, or at the very least, commission a Deep Research Report about that. We might even want to compile ideal customer profiles for the different customer segments in our market, such as patriotic locals and tourists visiting the area.

Then we'd take all that data and put it into the generative AI tool of our choice and ask simple, straightforward questions like:

What should our PR strategy be?
What things should we do less of, or not do?
How well aligned is our PR strategy to the client needs?
How well is our work aligning with the public that the client serves?

Synthesis is all about connecting the dots, so the more data we can bring, the better a job AI will do. The worst possible mistake you could make here in synthesis would be to provide no data and hope the AI can either naively search the web for it or know the facts already.

In the video version of this newsletter, you'll see that I'm doing synthesis inquiries from the ideal customer profile, the media landscape, and the company's strategy, along with the company's actual review data. This is a ton of fact-based, grounded data that is far more likely to give useful answers.

Generation

We arrive at our final destination, the last use case: generation. You might think this is strange - after all, isn't the entire point of generative AI generation? Yes and no.

Generation is the weakest use case for generative AI (weird, right?) specifically because of Principle 4 from my new book, Almost Timeless. Principle 4 is: AI doesn't know when it's lying. If we use generative AI naively, just asking it to do things without providing a ton of data, we're going to get what I can only describe as hilariously poor results.

Which is how most people use it.

Which is why many people get frustrated with it.

Think about it. If two people use generative AI to make, say, social media content for Ostannya Barykada, and one just says, "hey, make an image of people eating at the restaurant", and the other provides ALL the context we've built so far, then says, "make an image of people eating at the restaurant, based on these photos from the real restaurant, designed to appeal to the ideal customer profile, focused on what the restaurant is best at", who's going to get the better result?

There should be no question in our minds which result will be better.

That's why generation, as a use case, comes last - because if you rush past the other 6 steps in the process, you're not going to have nearly enough context to create powerful, useful, and compelling content with generative AI. Instead, you're very likely to make unremarkable slop, slop that won't catch anyone's eye or move the needle on your communications campaigns.

Beyond basic generation, we should also be thinking bigger - much bigger. What should the overarching communications strategy be? What kind of big picture strategic plan could you provide from all this context, from all this background information?

Let's ask.

Based on all the background information provided, let's build a comprehensive marketing communications strategy for Ostannya Barykada for the next four quarters. Use all the information we've accrued thus far to build a strategic plan (why we're doing things), a tactical plan (what we're going to do), and then a quarter by quarter execution plan, by channel, by quarter, with highly detailed lists of what we need to accomplish. Root the plan in the goals and KPIs set in our overall strategy, and provide KPIs for each quarter in the tactical plan. Provide metrics for success with each of the executions in the execution plan so individual teams know what they need to measure against that roll up to our KPIs and goals.

In my work with Trust Insights clients, more often than not, people are astonished at how AI can fill in blind spots and identify opportunities they never would have thought of. Today, with these tools, as a PR professional you could be building bespoke, incredible, truly personalized and customized strategy and execution for your clients. The time you save on building the materials (compared to weeks and weeks of manual research and hundreds, if not thousands of billable hours) means that you can deliver exceptional service and exceptional results.

If you look at the materials I created for this newsletter and the talk that preceded it, is it apparent that I don't know anything about the restaurant industry, or the Kyiv food scene? No - the opposite is true. Using AI, I've created rich, robust materials that ANY client would LOVE to have on their desks - and not just for PR, but for the entirety of marketing communications.

That's where generative AI, properly used, is really going to boost your results. No one wants another mindless blog post. Everyone wants customized, personalizations insights and plans of action that they can put into motion or hire you to put into motion for them.

Wrapping Up

In retrospect, I probably should have chosen a restaurant that I could actually get a complimentary meal from, rather than in a city that is many, many flights away, as this amount of work would be valued in the thousands of dollars range. Perhaps one day, when the war is over, I'll visit Ostannya Barykada and hand them this data, and maybe I'll get some free varenyky.

However, the point still stands: even with relatively little knowledge of a sector, as long as AI is our partner and not our replacement, we can create astonishingly good results together.

Your challenge, should you choose to accept it, is to try these exercises that I did in this newsletter for your company or your clients. I specifically went as absolutely low tech as possible, using only one set of tools that costs $20 a month. Unlike last week's issue, there's not a single line of code in here other than plain language prompts, to illustrate the point that all this is accessible to everyone.

I'll also underline the point I made at the beginning. Use this newsletter to evaluate your current PR efforts, your current PR team, your current PR agency. If they're not doing these table minimum basics, then you're not getting the most out of your PR efforts and resources. It's time to make a change to a more modern agency or team that is.

Shameless Plug

We got to just a tiny fraction of the use cases of generative AI you could do in PR. I teach this content in both our online course, but also in custom workshops. If you'd like a version of this content that's tuned and tailored to your PR agency or team, reach out to the Trust Insights team and we can work together to find the right options for you.

How Was This Issue?

Rate this week's newsletter issue with a single click/tap. Your feedback over time helps me figure out what content to create for you.

Here's The Unsubscribe

It took me a while to find a convenient way to link it up, but here's how to get to the unsubscribe.

If you don't see anything, here's the text link to copy and paste:

https://almosttimely.substack.com/action/disable_email

Share With a Friend or Colleague

If you enjoy this newsletter and want to share it with a friend/colleague, please do. Send this URL to your friend/colleague:

https://www.christopherspenn.com/newsletter

For enrolled subscribers on Substack, there are referral rewards if you refer 100, 200, or 300 other readers. Visit the Leaderboard here.

Advertisement: Bring Me In To Speak At Your Event

Elevate your next conference or corporate retreat with a customized keynote on the practical applications of AI. I deliver fresh insights tailored to your audience's industry and challenges, equipping your attendees with actionable resources and real-world knowledge to navigate the evolving AI landscape.

👉 If this sounds good to you, click/tap here to grab 15 minutes with the team to talk over your event's specific needs.

If you'd like to see more, here are:

ICYMI: In Case You Missed It

This week, lots of new content on the blog as I've started a completely new process for you to enjoy.

Skill Up With Classes

These are just a few of the classes I have available over at the Trust Insights website that you can take.

Premium

Free

Advertisement: New AI Book!

In Almost Timeless, generative AI expert Christopher Penn provides the definitive playbook. Drawing on 18 months of in-the-trenches work and insights from thousands of real-world questions, Penn distills the noise into 48 foundational principles—durable mental models that give you a more permanent, strategic understanding of this transformative technology.

In this book, you will learn to:

Master the Machine: Finally understand why AI acts like a "brilliant but forgetful intern" and turn its quirks into your greatest strength.
Deploy the Playbook: Move from theory to practice with frameworks for driving real, measurable business value with AI.
Secure Your Human Advantage: Discover why your creativity, judgment, and ethics are more valuable than ever—and how to leverage them to win.

Stop feeling overwhelmed. Start leading with confidence. By the time you finish Almost Timeless, you won’t just know what to do; you will understand why you are doing it. And in an age of constant change, that understanding is the only real competitive advantage.

👉 Order your copy of Almost Timeless: 48 Foundation Principles of Generative AI today!

What's In The Box? Here's a 5 Minute Tour

Here's a 5 minute video tour of the course so you can see what's inside.

Get Back to Work

Folks who post jobs in the free Analytics for Marketers Slack community may have those jobs shared here, too. If you're looking for work, check out these recent open positions, and check out the Slack group for the comprehensive list.

Advertisement: Free AI Strategy Kit

Grab the Trust Insights AI-Ready Marketing Strategy Kit! It's the culmination of almost a decade of experience deploying AI (yes, classical AI pre-ChatGPT is still AI), and the lessons we've earned and learned along the way.

In the kit, you'll find:

TRIPS AI Use Case Identifier
AI Marketing Goal Alignment Worksheet
AI Readiness Self-Assessment (5P & 6Cs)
12-Month AI Marketing Roadmap Template
Basic AI ROI Projection Calculator
AI Initiative Performance Tracker

If you want to earn a black belt, the first step is mastering the basics as a white belt, and that's what this kit is. Get your house in order, master the basics of preparing for AI, and you'll be better positioned than 99% of the folks chasing buzzwords.

👉 Grab your kit for free at TrustInsights.ai/aikit today.

How to Stay in Touch

Let's make sure we're connected in the places it suits you best. Here's where you can find different content:

My blog - daily videos, blog posts, and podcast episodes
My YouTube channel - daily videos, conference talks, and all things video
My company, Trust Insights - marketing analytics help
My podcast, Marketing over Coffee - weekly episodes of what's worth noting in marketing
My second podcast, In-Ear Insights - the Trust Insights weekly podcast focused on data and analytics
On Bluesky - random personal stuff and chaos
On LinkedIn - daily videos and news
On Instagram - personal photos and travels
My free Slack discussion forum, Analytics for Marketers - open conversations about marketing and analytics

Listen to my theme song as a new single:

Advertisement: Ukraine 🇺🇦 Humanitarian Fund

The war to free Ukraine continues. If you'd like to support humanitarian efforts in Ukraine, the Ukrainian government has set up a special portal, United24, to help make contributing easy. The effort to free Ukraine from Russia's illegal invasion needs your ongoing support.

👉 Donate today to the Ukraine Humanitarian Relief Fund »

Events I'll Be At

Here are the public events where I'm speaking and attending. Say hi if you're at an event also:

AFT, Washington DC, July 2025
AMA Pennsylvania, York, August 2025
SMPS, Denver, October 2025
Marketing AI Conference, Cleveland, October 2025
MarketingProfs B2B Forum, Boston, November 2025

There are also private events that aren't open to the public.

If you're an event organizer, let me help your event shine. Visit my speaking page for more details.

Can't be at an event? Stop by my private Slack group instead, Analytics for Marketers.

Required Disclosures

Events with links have purchased sponsorships in this newsletter and as a result, I receive direct financial compensation for promoting them.

Advertisements in this newsletter have paid to be promoted, and as a result, I receive direct financial compensation for promoting them.

My company, Trust Insights, maintains business partnerships with companies including, but not limited to, IBM, Cisco Systems, Amazon, Talkwalker, MarketingProfs, MarketMuse, Agorapulse, Hubspot, Informa, Demandbase, The Marketing AI Institute, and others. While links shared from partners are not explicit endorsements, nor do they directly financially benefit Trust Insights, a commercial relationship exists for which Trust Insights may receive indirect financial benefit, and thus I may receive indirect financial benefit from them as well.

Thank You

Thanks for subscribing and reading this far. I appreciate it. As always, thank you for your support, your attention, and your kindness.

See you next week,

Christopher S. Penn