§ 01 · The Hook
The bait, then the rug-pull.
Dara Denney opens with the same one-two that built Doberman Dan's career: a credentialed contrarian claim plus the receipts. "Most creative strategists are using AI completely wrong" lands in the first ten seconds, immediately backed by "I've spent over a hundred million on meta ads and worked in this industry for over ten years." Then comes the reframe that powers the rest of the video — treat AI like your junior strategist, not your replacement.
What the video promised.
stated at 01:03“Today I'm gonna walk you through my five best use cases for Claude Cowork and creative strategy. You'll see my actual prompts and how I'm thinking through this platform right now.”delivered at 13:48
Where the time goes.
00:00 – 00:47
01 · Cold open — "most strategists use AI wrong"
Contrarian claim, credentials ($100M spend, 10 years), and the reframe: AI is a junior creative strategist or marketing assistant, not a replacement.
00:47 – 01:21
02 · Promise — 5 best Cowork use cases
Sets up the value: actual prompts, actual workflows, amplify your strategic thinking rather than replace it.
01:21 – 02:25
03 · 60-second Cowork primer
Chat vs Cowork vs Code distinction, desktop app required, Pro plan minimum (she uses Max), connectors needed — Chrome and Slack at minimum.
02:25 – 06:12
04 · Use case 1 — Ad library teardown (Ridge Wallet)
Prompt structure shown. Cowork autonomously opens Chrome, scrapes Ridge's Meta Ad Library, returns an HTML report: 291 active ads, format breakdown, partnership vs brand ad split, core messaging pillars (durability, toughness, lifetime guarantee), inferred personas, top 10 ads by impressions.
06:12 – 09:45
05 · Use case 2 — Weekly creator social media report
Cowork pulls her own activity across LinkedIn / IG / YouTube / X / TikTok, asks clarifying questions mid-task, generates a brutal "do less" recommendation set. She has this scheduled to run weekly and ship to Slack.
09:45 – 11:52
06 · Use case 3 — Competitor / brand analysis
Same engine, different prompt — runs monthly on competitors and admired brands. Output is summary + spreadsheet + HTML, with insights like "celebrity collabs are 10x multipliers" and "founder-led content punches above its weight."
11:52 – 13:48
07 · Use case 4 — Persona research decks from reviews
Three-step pipeline: scrape product reviews to CSV → cluster into personas in a Doc you can edit → convert Doc to visual presentation. Capped 47k Ridge reviews at 3k for demo speed.
13:48 – 14:33
08 · Connectors aside — Canva, Gamma, PowerPoint
Claude has a native Canva connector. Gamma is a personal tool (paid partnership disclosed on IG/LinkedIn but not on YouTube). Native PowerPoint also viable.
14:33 – 15:58
09 · Close — Ogilvy was the research director
Lands the thesis: strategists spend more time researching than producing. Cowork automates the research grind. Soft CTA for a follow-up series on briefing + QA workflows.
Visual structure at a glance.
Named ideas worth stealing.
00:28concept
AI as Junior Strategist, Not Replacement
Reframe AI from "replaces you" to "junior strategist / marketing assistant you delegate to." Lowers the threat, justifies the price, gives permission to dabble.
Steal forAny AI tool pitch where the audience is professional and threatened — JoeFlow, Mod Producer, Paperclip. "Mod Producer is your junior producer."
01:50list
Three Claude Surfaces
- Chat (info retrieval)
- Cowork (agentic action — opens browser, reads files, completes tasks)
- Code (developer CLI for building apps)
Clear segmentation that does two jobs: educates the newcomer and positions Cowork as the action/value tier.
Steal forAny multi-surface product. The wedge "information vs action" is also reusable as a category positioning frame — JoeFlow captures decisions, not just words.
02:45list
Ad Library Audit — 7-bullet prompt recipe
- Format breakdown (video vs image, duration distribution)
- Partnership ads vs brand ads distribution
- Core messaging strategies / pillars
- Inferred target personas from creative POV
- In-depth Top 10 ads by impressions
- Round-up of longest-running ads + trend summary
- Creative velocity — uploads per day/week/month + last upload
Reusable prompt skeleton for any DTC brand's Meta Ad Library. The exact bullets are visible on screen long enough to screenshot.
Steal forJoe's MCN / creator-tooling content. Run this on a competitor's ad library and post the teardown as a thread or essay.
11:52list
Three-Step Persona Deck Pipeline
- Scrape product reviews to CSV (cap at 3k for speed)
- Cluster into named personas in an editable Doc
- Convert Doc to visual presentation via Canva / Gamma / PowerPoint connector
Each step is independently usable, so you can drop in at whichever stage your work needs. The intermediate Doc step is the unlock — keeps a human in the editing loop.
Steal forAny audience-research workflow for a creator product launch. Maps directly onto Joe's offer development.
09:48concept
Schedule + Ship to Slack
Use Claude's scheduled tasks + Slack connector to run the report weekly and auto-deliver. Turns an ad-hoc prompt into an asynchronous standing report.
Steal forAny recurring intelligence Joe wants — MCN content performance, JoeFlow user signups, Paperclip agent activity.
Lines you could clip.
00:12
“Most creative strategists and digital marketers are using AI completely wrong. And it's not necessarily because they're bad at prompting or even that they're using the wrong tools. It's because they're asking AI to do the wrong job.”
00:33
“AI only really started to click for me when I stopped using it as my replacement. Instead, I treat AI like it's my junior creative strategist or my marketing assistant.”
01:11
“The goal isn't to replace your strategic thinking — it's to amplify it so that you can spot opportunities faster that you would have never seen without it.”
08:55
“Personal pros without a professional hug. Okay. But I love getting this report.”
15:02
“Ogilvy, when he finally had his namesake agency, he didn't bill himself as the creative director or even the owner. He billed himself as the research director, which really just goes to show how important research is in this process.”
14:33
“As creative strategists, I think we often think about the immediate output, which in many cases is the actual ads or the briefs. But the majority of my time as a creative strategist is actually spent researching.”
How they spent the runtime.
Hook length47s
Info densityhigh
Filler8%
Things they pointed at.
02:25toolChrome connector
02:27toolSlack connector
03:38toolClaude Opus 4.6
03:40toolClaude Sonnet 4.6
11:52toolRidge product reviews (47k corpus, sampled at 3k)
14:35toolGoogle Docs (export pipeline)
15:00bookDavid Ogilvy — "research director" anecdote
How they asked for the click.
15:20next-video
“If you're curious about how we're implementing AI into the rest of our workflow, specifically in briefing and QA, we've really made some great strides there. Let me know because I'm happy to expand on this into a series.”
Soft, comments-driven CTA — asks for engagement signal rather than a hard subscribe push. Works because the audience is professional and would resent a hard pitch.
Word for word.
HOOKopening / re-engagementCTAthe pitchmetaphorstory
00:00HOOKSo this is how I'm using Claude Cowork in my creative strategy workflow today. So here's the thing. Most creative strategists and digital marketers are using AI completely wrong. And it's not necessarily because they're bad at prompting or even that they're using the wrong tools. It's because they're asking AI to do the wrong job. So I've spent over a $100,000,000 on meta ads, and I've worked in this industry as a creative strategist and media buyer for over ten years. And I will say that AI only really started to click for me when I stopped using it as my replacement.
00:30HOOKInstead, I treat AI like it's my or my marketing assistant. So today, I'm gonna walk you through my five best use cases for Claude Cowork and creative strategy. You'll see my actual prompts and how I'm thinking through this platform right now. Because the goal isn't to replace your strategic thinking, it's to so that you can spot opportunities
00:49faster that you would have never seen without it. But first, I wanna give you a sixty second rundown on co work. Please feel free to skip it if you're already familiar. So Claude has three big functions, chat, co work, and code. Chat is your basic chat interface. You probably already have a lot of experience with that. Now co work is where Claude can actually do the work for you. It can open your browser, read your files, and complete tasks on your behalf. Some people might call this an agent.
01:14Encode is for developers. It's a command line tool that helps you build apps and websites. So if this isn't you, just feel free to skip that one for now. Now this video is very specifically about Cowork because I want us to concentrate on getting real work done, not just information being generated. Now in order for you to access Cowork, you are gonna need the desktop app, not just in browser.
01:35And as far as pricing, I will say at a minimum, you are going to need the pro plan. I personally use the max plan, but that's just me. And the final thing that I'm gonna note here with co work is that in order for Claude to access different parts of your computer in different apps, you're gonna have to give it access through connectors.
01:55And for everything that we're gonna go through today, make sure that you at least have Chrome connected as well as Slack. Now let's dive into this first one, analyzing ad libraries. Now essentially what this does is it's going to analyze a Facebook ad library for you, aka the ads that are currently being run by any brand right now, and it's going to give you a report that contains incredible
02:18and actionable data for you to use. You can do format breakdowns and a deep dive on the creators they're working with via partnership ads and just an overall scope of their creative strategy. Now the reason why this is so impressive is because I used to do this exact report for my clients every single month. Now here's the prompt that we're gonna be using to generate this. Let's go ahead and put this into Claude CoWork and see what happens. So the basic prompt structure looks
02:44like this. Number one, you're gonna say, can you do an analysis on RAND? Let's go ahead and just do Ridge Wallet. Right? And then we're also gonna get a link that goes directly to their ads library. I found that when I'm doing this, if I just say go to Ridge's ad library, it often won't get the right thing. So I just wanna make sure, hey. Let's have it
03:08exact. So we have our prompt right here, and the first thing that you're gonna do is just select a project that you're gonna work in. I always just have everything be put into a clod folder. And then another thing that you can potentially select is, oh, do you want it to be Opus, which is like the highest computing power? Sometimes I use Sonnet, but since I have the max plan, I'm gonna go ahead and just use Opus for the best results.
03:34I already have the Facebook ads link directly to the ads library open, and we're just gonna go. And you can see here immediately,
03:44it starts the progress. So it's gonna show you everything that it's gonna do step by step. And you can also see that the connector is also searching. And you can if you want to really go through and see how it is starting to conduct these tasks. Now what you'll see here too is, oh, Facebook's domain is direct from direct fetching. But if if you can see right here, Claude, not me,
04:10just pulled up this entire window, and it's starting to go through Ridge Wallet's ad library, which is pretty cool. So I'm gonna go ahead and collapse that and let it do its thing. Sometimes it can take a few minutes to actually generate the entire thing, and you can actually see it moving in the background. That's not me. It's crazy. Ah,
04:32and there it is. K. It's giving me a little summary. Let's open it up, shall we? Wow. Okay. So we can see for Rigs, they have 291
04:43active ads for product lines, for creator partners, and 94 ads were launched on April 4 alone, and the longest running ad is a 158. Alright. Let's dive in. This is such cool data. K. We have a video versus image breakdown. Pretty even, fifty fifty split. We can also see the video duration distribution. I actually see this quite a bit with videos where they're forty five to sixty seconds long. Now I can see the percentage of partnership ads is pretty low, only 4.5%.
05:14Now something that I'd probably do a follow-up here is ask, hey. Where are these ranking in terms of impressions? Sometimes this report will automatically show it. Sometimes it doesn't. Core messaging strategies, durability, toughness, lifetime guarantee sales. Wow. Wow. Wow. And we can see, like, the big pillars here, specifically for wallets, it looks like. Now this is what I really love is seeing the inferred target personas. Now, something that I really like to do as part of my process, especially when it comes to persona mapping, is mapping the personas that you seem to be targeting from your creatives versus who your actual customers are, aka the analysis from the actual reviews. And then the top 10 ads by
05:56impressions. Okay? None of the partnership ads, it seems, but some of them are, like, the legends in Bloom. Of course, I think they're actually probably pushing spend to that, maybe on, like, a reach or an awareness campaign just because it is limited edition.
06:12Next up, let's talk about some other social media reports that I'm generating. Now one is gonna be for creators because in addition to owning an ad agency, I am also a content creator. And then the other is for brands that you're interested in spying on, aka your competitors. Now let's first dive into the creator one because I actually made that one first because I actually used to have my assistant generate this report for me every single week and month, and now it is completely automated by Claude. Okay. So here's the prompt. Let's run this thing. Okay.
06:41So again, you can have it compile in whatever type of format you want via Pitch Deck. I like the HTML graphics, but you could also do it in a sheet format if you wanted. So I have all of my links here, and I don't have any Chrome tabs open. Sometimes too, what I'll say is it'll hit a wall and say, hey, we can't access these. So I will tell it to make sure you open
07:08these links with the Chrome connector. Let's go. Oh, and this is cool. It's gonna ask me a user question, which I love. When you say last week, which date range do you mean? Great question.
07:21Um, let's do let's do this one. What engagement metrics matter the most to you for best performing? All of the above, baby. How detailed should the do less of recommendations be? High level detailed with examples. Call me out. Actually I did this maybe a week or two ago for something, and it . Up. And it just pulled up my tab too. We love to see it, and it's on my LinkedIn.
07:46Crazy. Okay. I'll get back to you when the report's ready.
07:58Okay. Now it's here. Very cool. Let's open it up. Now, of course, you can give it other specifications if you want it in a different color or whatever, but let's see. Okay. Last week, I did 10 posts on LinkedIn. Not bad. Six on Instagram, one on YouTube, which was a short, two on Twitter, and two on TikTok. And here was my top performer.
08:22It says it went viral. It did not. They put ad spend behind it, but good to know your brand breakdown and process content drove the highest saves. Not surprised. So the Oats Overnight post,
08:35that many saves, and the three processes real. Very good. Yeah. I did get a lot of impressions on that Caraway Home one and the AI video content, of course. See, it's just so cool to have this type of data at a glance for creators, especially too when it does it by platform in the different posts that I have. So I get a LinkedIn breakdown. That's awesome. Very cool to for me to refer back to.
09:02And then Instagram breakdown. Do less Amazon partnership post. Whoops. Fair enough.
09:11Observation had an incredible 4.8 save rate. And this is where I'm like, ugh, you're reading me to filth. YouTube and x are significantly
09:20underserved. Do less generic sponsored content. Happens. You're cross posting heavily on LinkedIn and IG and TikTok. Yeah. Fair.
09:30Personal pros without a professional hug. Okay. But I love getting this report. You can also tell co work, hey. Be sure you send this to me in Slack. And now that they have scheduled tasks, you can make sure that this runs every single week, which is what I do. Now the next one is for a brand or competitor analysis that I like to run for a few different use cases. Number one, you can do this for your competitor. So I'll actually have co work do this analysis for me monthly so I can keep tabs on what the competition is doing organically.
09:59Again, I used to do something like this, which while it did focus more on the paid aspect, I would also pull in elements from time to time organically. And another way that I'm using it is with my clients' competitors to really map out what their strengths and gaps are and if there's anything that we should capitalize on. And sometimes just because I'm hella nosy, I'll run it on some of the big brands or brands that I really admire. And what I really like about this too is when you're finished generating this report, you can also ask co work follow-up questions to that data.
10:30And here. Now, again, I'm having the final format be in a summary, a spreadsheet, and an HTML file. And you can really do this for as many brands or as few brands as you want. I've worked with a lot of beauty brands before, so I'm like, oh, I wanna keep tabs on these specifically. Let's go. And now we wait. Now the end result is going to be this. So I did have the summary right here,
10:57and then I have the report. So what's really cool is at a high level, because the follower counts, wow, Laura Gallo and Jones Road Beauty are neck and neck. So cool. Celebrity collab are 10 x multipliers. Not surprised,
11:11for beauty. And founder led content always punches above its weight. And then it gives top reel by likes because I really specifically wanted to look at the video content, and then it breaks down the top five reels each, and I can go ahead and it has a direct link. So cool. I love that effect. And it also, like, points out what they're doubling down on utility
11:33driven content. Same thing for Laura Galloway with their their but they're doubling down on celebrity partnerships and media moments. Interesting. Multiphase product launch campaigns, which is also very interesting because they're the only one that doesn't have a really front and center founder.
11:51Very, very cool. And the final one is how to build persona research decks. And we're actually gonna be breaking this down into three concrete steps because at any given time, you might just want to do one part of this process. Number one, I'm gonna show you how to scrape for reviews into actually break that data out into personas and then putting that data in a finalized deck. Now one thing I'll flag here is that for a brand like Ridge that has 47,000
12:16reviews, you're not gonna wanna do all those or it just might take a while, but for purposes of this video, I'm gonna do an even 3,000 just so that we can go a little bit more quickly. Let's go.
12:32Now I'm gonna go a little bit ahead. You can see here that it was able to generate for me the CSV of all these reviews. Amazing. We love to see it. It even broke it down into the different
12:44product variants, which we love. And now what I'm gonna do after this is I'm going to be pasting another prompt right into the same chat of that. I'm gonna be saying, hey, from a CSV of these reviews, do the following. Now what I like to do with this first is to have it put everything in a document for me so that I can review it and really understand, okay, what type of personas are they locking in on from these reviews.
13:13This can actually make a really excellent context document that you can upload to a cloud project. If you're interested in hearing more about that, definitely let me know. But I also like it because it's easier to edit. So I can export it and edit it in Google Docs. Awesome. But once it's in a place where I think it's pretty good, then what I'm gonna tell it is, hey. I want you to turn this deck into a presentation
13:36with visuals that has a graph, yada yada, and it's gonna work. And then you're gonna have something like this that you can review with your team. Now being able to have this kind of data and documentation without you actually having to do it, I can't emphasize truly what an unlock it is. And if you're seeing this and you're worried about the branding and the blah blah blah, all of that can truly be edited. And I will say too, Claude has a connector directly to Canva. So there are some specific like Canva style documentations
14:06that we prefer. We will actually have Claude go directly to Canva aspect. And if I'm just doing my own personal work, Gamma is something that I also personally use. And full transparency, I have partnered with Gamma on my Instagram and on my LinkedIn. They are not sponsoring the YouTube channel at this point, but it is genuinely a tool that I use and really, really like. So if you are creating a lot of reports,
14:31definitely try out the native PowerPoint, but Gamma and Canva, like, there's so, so much you can do in automate. And honestly, it has made just a massive,
14:44massive impact on my day to day. Because it's, you know, kinda funny. Right? Like, as creative strategists, I think we often think about the immediate output, which in many cases is the actual ads or the briefs. Right? But the majority of my time as a creative strategist is actually spent researching, and it's so funny actually. Ogilvy, when he finally had his namesake agency, he didn't bill himself as the creative director or even the owner. He billed himself as the research director, which really just goes to show how important research is in this process. And, yes, while I do have a lot of research
15:18CTAhere that is all automated, I'm still doing a lot of it manually, but it is this type of stuff and the visuals particularly that has just really done so much for me, and I love it. And I hope you guys enjoyed it too. Now I know that a lot of this video was super research heavy, but to be honest, as a creative strategist, I spend a lot of my time conducting research. So these have been a massive, massive unlock for me. However, if you're curious about how we're implementing AI into the rest of our workflow, specifically in briefing and QA, we've really made some great strides there. Let me know because I'm happy to expand on this into a series. And I'll see you guys next week. Thank Thank you so much. Bye.
§ · For Joe
Steal the credentialed-contrarian opener.
Dara Denney playbook
Open with a contrarian claim about your category, back it with credentials in the next breath, then reframe the threat as a tool the viewer controls.
- Build a cold open in this exact shape: "Most [audience] are doing [thing] wrong" → "And it's not [obvious reason], it's [contrarian reason]" → credentials ($X spent, Y years).
- Reframe AI tools you sell as "your junior [role], not your replacement." Lower the threat, justify the price, give permission to dabble.
- Segment your category by surface — "information" vs "action" is a reusable wedge. JoeFlow captures decisions, not just words. Mod Producer ships shows, not just clips.
- Show the actual prompts on screen long enough to screenshot. Audience steals them, audience evangelizes — you become the source.
- Use a branded lower-third (CLAUDE IN ACTION) at every demo transition. Tiny wayfinding device, costs nothing, dramatically lifts perceived production value.
- Close on an authority callback (Ogilvy as research director) that lands the thesis in one story instead of restating the bullets.
§ · For You
What this could mean for you.
If you're a strategist or marketer trying Cowork
Stop using Claude as a search engine — start using it as a junior teammate you brief, schedule, and review.
- Install the Claude desktop app (Pro plan minimum) and connect at least Chrome and Slack before you try anything ambitious.
- For a competitor audit, give Cowork the literal Meta Ad Library URL — not the brand name. Generic asks bounce; exact links land.
- Steal her 7-bullet ad library prompt verbatim: format breakdown, partnership vs brand split, core messaging pillars, inferred personas, top 10 by impressions, longest-running ads, creative velocity.
- Have Cowork output as HTML or a spreadsheet, not just chat — much easier to share with a team or paste into a doc.
- Use scheduled tasks + the Slack connector to turn an ad-hoc prompt into a weekly standing report that arrives without you asking.
- When researching personas, cap your review CSV at 2-3k rows for speed — diminishing returns past that point.
- Keep a human in the loop on the intermediate Doc before generating the deck — that's where editing is cheap; once it's in Canva it's not.
