Getting the transcript
Reading the captions from YouTube. A video nobody has opened here before takes 10 to 30 seconds; this page fills in on its own.
Getting the transcript
Reading the captions from YouTube. A video nobody has opened here before takes 10 to 30 seconds; this page fills in on its own.

Sharbel A. · @sharbelxyz
This video has no Most replayed graph yet: YouTube shows one only once a video has enough views. These are the moments viewers replayed most in Sharbel A.'s most watched videos.
Most replayed moment at 13:32
3.9x that video's typical replay level
built-in memory, which is great. Here is how I would think about memory providers. Memo is interesting if you want a dedicated memory layer for personalized AI agents. It focuses on extracting, storing, linking, and retrieving memories efficiently. Their
Said at 13:25
Most replayed moment at 6:37
5.8x that video's typical replay level
trading strategy. Tests it. If it's better, it keeps it. If it's worse, it discards it and tries again. So, that's exactly what I built. Okay, here's the system we have at play. I gave it two years of crypto data,
Said at 6:31
Most replayed moment at 1:53
3.7x that video's typical replay level
let's install it together. Okay, so for step one, we need to actually start by installing Bullpen's CLI so that we can actually do everything that we want to do. And for that, let's first open Claude and put in the dangerously skip
Said at 1:47
The graph counts replays. It does not show where viewers stopped watching.
Words
2,286
Runtime
13:27
Speaking pace
170wpm
Reading time
10min
170 words per minute, between the 160 25th percentile and the 181 median of 349 measured videos. That distribution comes from the 349-video hook study.
Opening (first 30 seconds)
Your AI agent can't create anything. It can write you a script, it can plan you a campaign, it can describe the perfect shot in beautiful detail, but if you're actually asking it to generate a video or a stunning picture, [music] it just hands you a homework assignment. Well, that just changed because Hicksfield just shipped a connector that gives your agent actual hands. And the part almost nobody's covering is that it does not only work inside of Claude. It works with Hermes
85 words, the words spoken in the first 30 seconds at 170 words per minute.
Free, no signup. See how the first 30 seconds hold attention, with rewrites.
Sentence shape
| Measure | This transcript |
|---|---|
| Sentences | 159 |
| Average words per sentence | 14.4 |
| Longest sentence | 56 words |
| Questions asked | 0 |
| Sentences containing a number | 19 |
Most used terms
Filler phrases
22 in total: like 13 · actually 4 · I mean 2 · literally 2 · uh 1.
A literal whole-word count of the same phrase list the Prepublish browser extension uses, so a phrase inside another word is not counted and a phrase used in its ordinary sense still is. It is a count and not a judgement.
Free, no account. See where attention is likely to drop, with a rewrite for each weak line. The free check shows the scores and the one issue costing the most. Or run it on the words above first.
Free · No login · See a sample audit first if you prefer.
What this transcript is
Every word below is the caption track YouTube publishes for this video, pulled from the video itself and reproduced unchanged. It is not Prepublish's writing, not a summary, and not a re-transcription: it is the video's own published captions. English captions, generated automatically by YouTube, in the video’s original language. Source: the video on YouTube. A channel that would rather this page did not exist can ask for its removal through the contact page, and it is removed.
Your AI agent can't create anything. It can write you a script, it can plan you a campaign, it can describe the perfect shot in beautiful detail, but if you're actually asking it to generate a video or a stunning picture, [music] it just hands you a homework assignment. Well, that just changed because Hicksfield just shipped a connector that gives your agent actual hands. And the part almost nobody's covering is that it does not only work inside of Claude.
It works with Hermes and any other agent you use. So, in this video, I'm going to show you exactly what shipped, how to connect it to Hermes or any other agent, how to run your first generation, and then I'm going to point my agent at a product and have it take my business to the next level. Let's get started. Here is what landed recently. Hicksfield released an official MCP server, one single endpoint that your agent connects to it once, and from that point onwards, it can generate videos, images, cut clips, reframe them, and write the file straight into your working directory.
This video is sponsored by Hicksfield, by the way, which is great because that gives us a good amount of credits to play around with. I'm excited for this integration because I absolutely hate having hundreds of tabs open for every individual thing that I run. And this update allows us to connect it straight to Hermes or whichever agent you use on a daily basis. The link is in the description if you want to try it out for yourself.
Connecting it takes about 30 seconds, and I will show you that in just a minute. And here we can see the same endpoint works in Claude, Grok Bot, ChatGPT, Cursor, Claude Code, OpenClaude, and even Hermes. But, before I connect it to anything, let me tell you about what this connector actually does. The MCP exposes the full Hixfeld model stack to your agent. This means it has access to Seedens 2.5, GPT image 2 for images, and there is more in here than just video generation.
You can use this for marketing, for faceless videos, for website building, for image generation. There is a clipper that takes a long video and cuts it into vertical shots. There is a reframe tool that changes aspect ratio without cropping. There's a virality predictor that scores a clip before you post it. I'm not going to cover all of those today. I am going to cover the one that excites me personally the most and that I feel could help my business out.
And the fastest way to show you is to connect it. So, let's do that. If you're on Claude, all you have to do is go click this little plus icon, click on connectors, add a connector, add custom connector, and then over here we're going to call it Hixfeld, and then we're going to copy this link, mcp.hixfeld.ai/mcp, and we're going to paste it over here and click continue. Boom, just like that. From here we're going to pick out the recommended options and click add.
It's going to open up this window, which is just us allowing access. We're going to click allow and open Claude. That's it. As simple as that. That's the entire setup. And boom, you can see, connected to Hixfeld. And we know that it's live because if we go to connectors, we can see Hixfeld over here. It's turned on. Hermes is MCP native, so it's the same endpoint added to my MCP server's config. All you have to do is click on Hermes, copy this, paste it into your Hermes chat, and it will give you a link to authenticate and that is it.
You're authenticated and boom, over here, Cleo's installed, authenticated, workspace selected, and we have it running. Now, let's get to the fun part. Let's test it out. I'm not going to start with a campaign. I am going to start with one prompt so you can see the mechanics with nothing else in the way. So, here the first prompt I've given it is using Higgs field, make me one product shot video of a matte black water bottle on a concrete surface, hard side light, and attach it in this chat so I can watch it.
And by the way, notice what it did there. I did not tell it which model to use. It picked based on my request and I don't know whether it picked Seasons 2 or Seasons 2.5. I'm assuming it picked Seasons 2.5 just like that. So, it has the ability to understand your job and through that pick the specific model that will work best for your use case. Now, let's watch the video. >> [music] >> There we go. Took literally less than a minute and we have a video of exactly what we asked for.
Now, this cost me 45 credits, which roughly comes down to 2 and 1/2 dollars. Uh I will come back to that number at the end because this is the number that will decide how worth it is this. But, that is the mechanic. One prompt, you get one file, but a single product shot is a demo. It's not a business. So, let me show you what I would actually use this for. Now, there are three major ways that you can use this. Let's start with number one.
UGC ads. I mean, UGC ads is a massive use case for this type of workflow and you can see a lot of the content being created and a lot of the use cases are UGC factories. So, with that, what I've done is I've told it take that water bottle you just made and turn it into a 15-second UGC ad, someone holding it, talking to the camera, natural lighting, like a real customer review. Let's see what it gives us. Our ad is ready, so let's go ahead and watch it together. >> using this matte black bottle every day.
It keeps water cold for hours, doesn't leak in my bag, and the finish feels premium. If you need one bottle, get this. Seriously. >> I like that it kept the same exact bottle shape. It kept the bottle true to dimension. I would change the actor's voice. He does sound AI generated, but then again, most of the UGC videos that are printing right now sound exactly the same. So, if I were to use this for, say, I had a I don't know, a bottle company, then I would try to switch up the voice just a little bit so that I could differentiate myself from the rest of the competition.
But, that is a finished ad in 4 minutes. This product literally did not exist 4 minutes ago. This generation, since it's 15 seconds, cost us 75 credits, which roughly comes down to about $4. This second use case is a little bit more selfish, but I need a new intro animation, and I do not want to wait 2 days for an animator to make one for me for every single video. So, what I did is I asked it, "I want you to make me a 10-second cartoon animation for my YouTube video intro that I want to include as a B-roll." The video's titled Hermes Agent Can Make Videos Now, and it's a Pictory's new MCP functionality inside of Hermes.
And then I described what I wanted in my video, and I asked it, "But first, help me conceptualize the prompt before we run it." So, I want to have a hand in what it ends up creating. So, it asked me a bunch of questions. I told it, "Yes, I'll send you a picture. Here's what I want. Here's what I have in mind." And it came back with a concept, which I approved, and it gave me a first still, which I was not a fan of the anime style, even though I picked it.
So, I I changed my mind. I told it, "Never mind. I want a 3D Pixar or Disney-style cartoon version." which it gave me this, which I liked a lot more. So, I told it, "Yes, this one is much better." I told it, "Great. Let's Let's get started." And it just generated the four different stills. Still number one, still number two, still number three, just like that. And I'm going to tell it, "I love it. Let's do the shots." Now, all we have to do is wait and come back to it in a few minutes to see what we got.
And it just finished. So, let's give that a watch over here. So, shot number one, shot number two, shot number three, shot number four. That was not bad at all. A couple of things I would change in my next generation would be to hand it specific logos. For example, Hicksfield, it generated the logo from scratch, which is something all AI agents tend to do. So, giving it the actual logos, the Hermes agent logo, giving it better direction.
I really like this shot right here. This one, for example, again, same thing, change the Hicksfield logo. So, I handed more brand assets. Maybe I would create a a brand asset folder where I would input all of my brand assets, other brand assets, the brands I'm talking about, so on and so forth. But, from this show, I really like shot number one, shot number three. Those are very good shots. The other two, I would ask it to regenerate, but all in all, this cost me 136 credits, which comes down to roughly $7.5.
Now, that might sound like a lot, $7.5, but think of the alternative, hiring an animator, waiting 1 plus week for an animation. I can understand why a lot of people are switching to optimizing their workflow with AI from video editing to video creation to animation itself. With that being said, let's head on to use case number three, which I'm which I'm also very excited about. Something I'm excited to try this out with is for web design.
Let me show you this. You can build out interactive websites through the Hixfeld MCP. And there's an entire category. I mean, look at these. These look wonderful. And that's exactly what I want to try it with for my own website. So, what I've done is I've asked Claude to build a responsive site in white and black, blah, blah, blah, with a red color accent and a strong CTA. But, first I want you to help me scope what you need from me, so ask me questions and run me through the prompt before you execute.
And there you go. It just built this in 5 minutes and we have this new hero website. Just like that. This animation right here that plays as you scroll. I really like this. I think I'm definitely going to be using this more often in my other websites. That being said, this cost us 41 credits. It was the cheapest out of the the four tests we've done, which narrows down to about $2. Claude is a chat window. It runs when I'm sitting in front of it, but Hermes Hermes is not.
Hermes runs on a machine on a schedule whether I'm awake or not. So, instead of typing these animation prompts every single time I have to film, I can turn these into a skill and that skill can hold your preferences, your brand colors, the style you want, the folder your editor pulls from, or whichever workflow you use video animation for, or web design, or image generation for. A prompt is a one-time instruction, whereas a skill is a repeatable process.
Then, you can point a cron with this skill that you've built out in Hermes. And every time, say I film a new YouTube video, Hermes can read the title and the film footage's transcript, and it can generate three intro options for me for that specific video, and drop them for me in the same folder. And just like that, that's how you can use Hermes' power, its ability to have skills, its ability to schedule cron jobs with this new update.
Your agent did not get an you too. Your agent got the ability to finish. Before this, every creative workflow ended the same way. The agent did the thinking, and then handed the doing back to you. You still have to open other apps. You still exported the file. You still did the last mile yourself. And that last mile is now gone. The only thing that stood between you and 50 different creatives was production time, but that constraint is now gone, as well.
My verdict is that this is the first version of agentic creative that I would actually put into a real business. If you want to try this out yourself, the connector is one URL, and it takes about 30 seconds to set up. I'll leave the link in the first line of the description. And comment "Hixfield" if you want more videos like this, or just let me know what you'd love to see more of. If you enjoyed this video, make sure to leave a like.
And if you're new to my channel, then subscribe, because I have a ton more content coming your way. Oh, and the algorithm gods just told me you're very likely to enjoy this video, as well. So, click it, and I'll see you there.
The words are the caption track's own and nothing is reworded or re-transcribed. Paragraph breaks are placed between sentences so the text reads as prose.
Free tools for your own script: paste a draft and see where it stands before you record it.
Paste your draft and see where viewers are likely to drop off, with a rewrite for each weak line.
Paste the first 30 seconds of your own draft for a hook score and rewrites.
Check your draft against YouTube's advertiser-friendly guidelines before you record it.
Read this channel's public videos and transcripts, and download a writing brief for it.