An agency owner walks through the workflows where an AI agent now drafts YouTube packaging, brand vision docs, recruiting outreach and social carousels, and where a human still has to step in.
An AI agent can absorb entire marketing job functions, packaging, brand docs, recruiting outreach, carousels, but only stays useful if a human keeps injecting taste and turning each correction into a reusable workflow.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
You run a small marketing team or agency and want a concrete map of which tasks (packaging, brand docs, outreach, carousels) can move to an AI workflow first.
You already use an AI agent for content but only for one-off prompts, not for a saved, repeatable workflow.
You're hiring for a marketing or AI-native role and want a working definition of what 'good' looks like in an interview.
SKIP IF…
You want a step-by-step technical tutorial on building the automations. This is a walkthrough of the results and reasoning, not a build guide.
You're looking for a general introduction to AI agents. It assumes you already know what an AI agent is.
TL;DR
The full version, fast.
Eric Siu argues GPT-6 Astra has taken over large parts of his marketing team's work, not by replacing people but by absorbing tasks while people shift into managing the workflows around it. He walks through five live workflows: a content engine that audits a raw podcast recording and proposes video packaging and short-form candidates; a design workflow that turns a company vision into a full brand document and framework diagrams; a recruiting workflow that drafts job posts and outreach sequences and can operate tools like an ATS or Juicebox directly; a carousel workflow that repurposes one video into Instagram content; and a ManyChat workflow that automates comment-to-DM funnels. The through-line: AI drafts get you 90-95% there, and the remaining leverage comes from injecting taste, leaving feedback, and saving the correction as a reusable skill rather than repeating it from scratch.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Cold open: the claim that GPT-6 Astra has replaced large parts of the marketing team.
00:12 – 00:55
02 · The YouTube results
Scrolls his own channel comparing recent AI-workflow videos (4K-10K views) against older videos stuck at 400-800 views.
00:55 – 01:09
03 · How the team's work changes
Reframes 'replaced' as work changing shape: the people who stay valuable manage the workflows rather than doing the manual task.
01:09 – 03:45
04 · The content engine in action
Feeds a freshly recorded podcast into the content engine, which audits it and proposes mid-form video angles, titles, and 16 short-form candidates.
03:45 – 05:46
05 · YouTube packaging and creative feedback
Reviews AI-generated thumbnail/title packaging, gives live feedback (podium layout, 'more controversial'), and asks for a more creative third option.
05:46 – 07:10
06 · Design and communicating the company vision
Uses the agent to draft the company's vision/mission page, an API-gateway diagram, and a two-by-two effort-vs-usefulness framework.
07:10 – 09:10
07 · Turning the vision into an X article
Converts the vision doc into a published article with a banner, service tiers, and a 'usefulness and effort are different axes' framework, calling the draft 95% there.
09:10 – 10:09
08 · Reusable design systems and distribution
Builds a design-system preview so the vision and visual language don't have to be rebuilt from scratch each time, and argues against being a one-channel operator.
10:09 – 12:15
09 · Recruiting and outreach workflows
Runs a recruiting workflow that drafts a 'Forward Deployed Marketer' job post and outreach email sequences, and can operate the ATS and Juicebox via browser access.
12:15 – 13:04
10 · Carousels from existing content
Turns one YouTube video into an Instagram carousel ('30 traits of an AI-native marketer') with captions, formatted for the platform.
13:04 – 13:46
11 · ManyChat and automated DMs
Shows a workflow that sets up ManyChat comment-to-DM automations, cutting a 20-30 minute manual task down when it has to run a dozen times a day.
13:46 – 14:16
12 · The people we're hiring
Closes on hiring: the people who understand how to audit and manage AI agents are the ones the company wants, with a link to the careers page.
Atomic Insights
Lines worth screenshotting.
AI output that's 95% finished is worth accepting and hand-finishing rather than re-prompting endlessly for the last 5%.
A channel's views can jump from the 400-800 range to 4,000-10,000 once output volume stops being blocked by production capacity.
Leaving inline comments on an AI-generated document and having the agent apply them is what turns a one-shot draft into a collaborative workflow.
Asking explicitly for a third, 'more creative' or 'more controversial' option after two safe drafts reliably pushes an AI agent past its default safe answer.
A two-by-two framework plotting delivery effort against customer usefulness turns every future pricing or scoping conversation into a lookup instead of a fresh negotiation.
An AI agent with browser access can operate tools that have no API, which removes 'no integration exists' as a reason not to automate a task.
Automations that individually take 20-30 minutes turn into a full lost day once you have to repeat them a dozen times in a day.
Betting on a single discovery channel is riskier now because search, social and AI agents are simultaneously deciding what gets found.
Repeating the same vision/mission/values statement across every artifact isn't redundant, it's the mechanism that keeps an AI's output aligned with what 'good' means.
Hiring for 'can direct and audit an AI agent' tests a different skill than hiring for the underlying craft, and the interview questions should reflect that.
Takeaway
Taste and workflow ownership outlast the task
WHAT TO LEARN
When an AI agent takes over a job function, the value moves to whoever keeps injecting taste, correcting the output, and turning that correction into a workflow the next person can run.
01GPT-6 Astra and my marketing team
When an AI agent absorbs a job function, the leverage shifts to whoever manages and refines the workflow behind it, not whoever used to do the task by hand.
02The YouTube results
A channel's view counts can jump from a few hundred to several thousand once content stops being blocked by production capacity rather than by audience interest.
03How the team's work changes
Treat 'my team got replaced' as 'my team's job changed to auditing and directing AI output' rather than as a literal headcount story.
04The content engine in action
Feeding a raw recording straight into an audit workflow turns one recording into a menu of ready-to-review packaging and short-form options in minutes.
Leaving inline comments on AI-generated output and letting the agent apply them is what makes a workflow collaborative instead of one-shot.
05YouTube packaging and creative feedback
Packaging decisions improve fastest when you give the agent a completed first draft to react to and revise, not a blank prompt.
Asking explicitly for a 'more creative' or 'more controversial' option after two safe drafts is a reliable way to push past an AI's default safe output.
06Design and communicating the company vision
Repeating the same vision, mission and values statement across every artifact is the actual job of a leader working with AI: someone has to keep restating what 'good' means so the model can be corrected against it.
A reusable two-by-two framework for effort versus usefulness is worth building once because it turns every future pricing or scoping conversation into a lookup instead of a negotiation.
07Turning the vision into an X article
Output that is '95% there' is worth accepting and finishing by hand rather than re-prompting for perfection, because the last 5% is exactly the part that needs human judgment.
Generic-sounding AI phrasing is a signal to add a specific, opinionated detail back in, not a reason to discard the draft.
08Reusable design systems and distribution
A design system prevents redoing brand decisions from scratch on every new asset, which is what actually lets a vision scale across channels.
Betting on a single discovery channel is a bigger risk now, because search, social and AI agents are simultaneously deciding what gets found.
09Recruiting and outreach workflows
Recruiting and partnership outreach can run through the same kind of workflow, draft, feedback, refine into a reusable sequence, as content production.
An AI agent with browser access can operate inside tools that have no API, which means workflow automation isn't limited to what a platform exposes.
10Carousels from existing content
One piece of long-form content can be mechanically repackaged into multiple owned-channel formats without new original work, as long as the source material is rich enough.
11ManyChat and automated DMs
Automations that individually take 20-30 minutes become a full lost day at even a dozen repetitions, which is the real argument for turning a manual sequence into a named workflow.
12The people we're hiring
Hiring for 'can direct and audit an AI agent' tests a different skill than hiring for the underlying craft, and interview questions should probe for that directly.
Glossary
Terms worth knowing.
GPT-6 Astra
The AI agent the video's creator uses to draft and run marketing tasks, from video packaging to recruiting outreach to design documents.
Content engine
A saved workflow that takes a raw recording (transcript and link) and outputs an audit, video packaging options, and short-form clip candidates.
Forward Deployed Marketer
An internal role the creator's agency is hiring for: an AI-native marketing generalist who is deep in one channel (SEO or paid media) but broad enough to consult across a client's whole account.
ManyChat
A tool that automates the comment-to-DM funnel on platforms like Instagram, sending an automatic direct message when someone comments a keyword on a post.
Juicebox
An agentic outreach tool the creator's team uses for recruiting, referenced as one of the apps the AI agent operates directly via browser access.
“GPT-6 Astra has replaced so many elements of my marketing team. It's not even funny anymore.”
cold open claim with no setup needed→ TikTok hook↗ Tweet quote
08:08
“If you just can get me five to 10% or 95% there, then I have a lot more leverage.”
compact, quotable standard for what 'good enough' AI output means→ IG reel cold open↗ Tweet quote
13:00
“It has replaced the people who don't know how to adapt, but it hasn't replaced the people who know how to run these workflows.”
thesis-line summary of the whole video→ newsletter pull-quote↗ Tweet quote
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
17px
metaphoranalogy
GPT -6 Astra has replaced so many elements of my marketing team. It's not even funny anymore. I'm going to show you exactly the results and also how we got there.
So you can do the same thing as well. Okay. So this is my YouTube channel over here and you can see, okay, I use a GPT -6 Astra to run my marketing for a day, 10K views, how I use Grokbot to run marketing every day, 4K views, 27 traits of AI native marketers, 4 .4K views.
gbd6 austria how i use it to run my business 10 grokbots i used to run my business right and then you can see grokbot crushes hermes and open call okay now you can see like this one's not that not that bad either but before i was getting 600 views 1200 views 853 views 1700 views over here 437 views my channel was not doing well 462 views over here 436 views 731 views 757.
it's very demotivating to see these numbers right but Now, with all these skills that we have, we're able to drive better results, right? And so when I say it has replaced my marketing team, the way I think about it is that work has fundamentally changed where the people who are good on your team, they are managing these workflows.
And I'm going to show you these workflows so you can do the same thing as well. Okay, so this is the workflow in action. And I'm actually going to show you one thing that it's working on right now because I just recorded like a 45 -minute video.
But this is a marketing school podcast that I recorded with Neil the other day. So Neil and I, we do this podcast and you can see. that it made a framer -like UI over here.
So literally what we did was I said, hey, I want you to run the leveling up content engine, and here's the Riverside link. We just recorded this right now, okay? So it audits the whole thing.
It's like, okay, here's what we think you should be attacking over here. So we're proposing some mid -form videos, and we have these skills in here, okay? So here's how we're building reusable AI workflows over here.
It's going to be 15 to 20 minutes, why agencies are struggling, how we're adapting. That's not bad packaging. What Nike's marking shift can teach operators.
Probably could be a little more controversial with that title. I think how we turn content engagement into sales conversations. I think that was a good part too.
I built an AI content engine. Neil spotted the problem. Why SEO and GE are becoming the same conversation.
So it found the right moments over here and it's showing that the assessment, what we should be doing over here. I was like, Hey, by the way, I need some short forms as well. So it comes up with, look at this 16 short form candidates over here that it considers to be strong.
So. Your next AI workflow belongs in this quadrant. Your marketing team learns, but does it act?
Stop judging B2B videos by views. Okay. These are good.
And okay. So all these things that I'm telling you right now, this is my taste, right? This is how I want to do things.
I can just give it to, I can just continue to upgrade the skill over time. And then whoever takes over this workflow, they can continue to make it into their own. But then I'm just saying, okay, give me the entire workflow set over here.
Okay. That's cool. Now.
On the right side over here, you can see that, okay, why agencies are struggling. Here's what's happening over here. And I can leave comments, right?
Like I'm using Framer, for example. I think, is it Framer? I think it's Framer or Remotion.
I don't know. But you can leave comments in here and then the AI will pick it up, right? And then you can, we're getting to a point where you can collaborate with your team.
And then I'm going to continue to update this skill over time because what it was doing is it was running it locally on my computer, but then I want to have it run in the cloud. So it's not eating up all my space, right? But these are things that - Codex and GPT -6 Austria can build for you.
But this is an entire job. Short form content, mid form content, long form content, getting the captions, getting it for newsletters, posting it, finding the right moments. These were all different jobs before.
In fact, that was an entire team where you can now do this on your own. You can put it together, do what you like, and then you can move it on, right? So once you have it working for you, once you have a marketing workflow working for you, then you nail it first and then you scale it.
And you can see there's some short forms over here. You can see it includes an overlay. So whatever preferences I like, I can just continue to upgrade these things.
So this is the 45 -minute video that I just recorded. And I said, hey, I want you to run the leveling up content engine and use this YouTube packaging skill for three options for the full video and four separate long -form videos that cover one each. So let's see what it came up with.
This is for the full video. These are the full video angles for the full 45 minutes. Okay.
My daily driver, Grokbot, Muse, Instinct, Hermes. So I might ask it to add the logos there. Beyond the chat, four agents I use for content.
Okay. So these are fine. I would probably want to make them better.
I'd probably want to go with the packaging that we had recently with the trophy case, right? And I might just ask it to do that and go from there. So I'm going to change it up right now.
Watch. Hey, so for the full video angles over here, so for the first one, My Daily Driver, let's just change it to, you know, like we can have different text, right? The ultimate agent or like who wins, right?
I want to have the logos for each of these little cards over here. So for Grokbot, for Muse, for Instinct, for Hermes, that's Metamuse, right? So that could be one packaging.
So who wins as the text up top and then the logos. There's that. Then the other one is, because you know that I've done, we've done packaging recently where we have maybe like a, maybe there's a podium for example.
Right. And usually a podium is top three. So you might have someone else that just like on the ground.
And so, you know, maybe, you know, Grock bot is, is, is number one. And then Hermes is number two. And then instinct is number three.
And then Metamuse is on the ground type of thing. And maybe they're like little characters. And so, and then based on kind of the feedback that I just gave you right now, I want you to make a third packaging that's more creative than what you came up with here.
Right. And so, you know, we want, we want these to be a little more controversial, I would say, and then see what happens from there. So what it did earlier was it pulled the transcript, but I'm going to try to refine these a little bit.
I find that whenever I just let the AI work on it, it's not that good, right? And so this is where you need to inject your taste and maybe you can make your taste into an eval. Now, in terms of getting design work done, GPT -6 Oster is amazing for this too.
So the way I'm looking at this is if you look at what I have over here, it's important for you as a leader to communicate what the vision is. What is the vision? What's the mission of the company?
And you have to continue to... Repeat yourself. Again, vision, mission, values.
Mission, vision, values. You're the chief repeater officer. And so for us, our mission is we're the marketing partner that companies will never outgrow.
And we talk about the different features that we have, and we talk about here's how everything connects, all the things that we're working on. So we have this API gateway. We have our marketing intelligence here.
Useful work becomes workflows. And then we're always asking our clients, what do they want to accomplish ultimately? And then this is how the workflow works over here.
And then also importantly, we need to have a two -by -two diagram that shows that we continually want to move towards building reusable workflows that are less delivery effort over time and are high customer usefulness versus just doing custom premium work all the time because that's harder to scale. And then we have this three -layer cake over here, managed services and then gateway access, and we charge people on outputs and outcomes.
This makes it easy for outsiders. or people on our team to understand what it is that we're trying to do. And so this is something I'm going to continue to update over time.
And it formed this for me, right? Like I made this myself. I didn't get any other help to make this.
And it also came with different design concepts. So GPT -6 Austria is really good with design. And the other thing I'll say too is I actually, I said, hey, I want you to actually make this.
our vision into an X article. So this is what it looks like. This banner, I didn't ask it to make this banner.
So it shows you services that's offered. This is what it looks like. This is in our color over here.
This is a single brain philosophy, right? So what services offer will actually look like. So custom build that can become a reasonable product.
Custom work deserves a place. Usefulness and effort are different axes, right? Here's the framework we're using.
Okay, great. It inserted this. I asked it to add some visual diagrams in there.
Here's how it works and actually tighten it a little bit as well. That's pretty cool. High effort at high usefulness is premium in customer work.
Customer has a meaningful need and meaning to take substantial effort. Okay. I might edit these a little bit, but I think this is like 95 % there.
Efficiency should create room for both sides. One capability, three ways to buy. Okay.
So, you know, it does sound a little bit like AI over here. So I think you need to add in your magic five to 10 % or so. But if you just can get me five to 10 % or 95 % there.
then I have a lot more leverage because this actually incentivizes me as a business operator to create more content. And that is one of the highest leverage things that I can do. And so what you just saw earlier was my YouTube results is that they're starting to compound more because I'm no longer constrained.
And because before when I felt like I was constrained, it actually.
It disincentivized me from making the videos. I didn't want to make the videos because I didn't trust that we can get the edit out on time. I didn't necessarily trust the packaging.
There's a whole host of things there. But I know that if I can build the workflows there and I can hand off the workflows to someone else to manage, that's a lot easier to manage. Because a lot of that trust is inside the workflow.
And if you trust the person to ultimately just manage it, then it's going to be fine. Having to trust a lot of different moving parts within the process, that leads to a lot of breakage and that leads to a lot of people just not wanting to make content ultimately, right? But as you get better and better with this, as you continue to tweak it and you hand off these workflows to other people, that actually encourages you to make more content.
It actually makes the content more fun. And so for me, I basically made a design system preview over here. So once you have something working, you should make a design system for it.
And then you publish it to your, push it over to your repo. And so for me, it's a couple of things. One, establishing what the vision is and communicating it to my team over and over and over having a design system for it.
So it continues to scale. So I don't have to keep redoing this from scratch every time. And then also is.
Can you repurpose the vision that you have? Can you repurpose the content that you create? Can you repurpose just like you saw earlier, right?
And then you're able to tackle these things. Because to me, I believe that you can't just be a one trick pony with one channel anymore. Back in the day, maybe 10, 15 years ago, you could just hit SEO as a channel.
You can't do that anymore. Today, it's very much a search everywhere optimization game. So it might be called SEO first.
but still search everywhere optimization, which means you want to be found on all these different channels and you want to be, you want to, we're also trying to hit agents as well. We want to be, we want agents to find this as well. Right?
So. That's cool. In business, how do you go about recruiting influencers?
How do you go about recruiting or thinking about partnerships, for example? And to me, these are forms of marketing, right? When you're going out there, you're trying to source the right people to partner up with.
You're trying to source, maybe you're trying to do like a content collab with somebody else that's still marketing. So you're doing a lot of outreach, but a lot of people don't know how to do outreach, right? So what I did here from a recruiting standpoint is We're looking to recruit these four deployed marketers on the SEO side and the paid media side.
We want more paid media people. We want more SEO people. And we're putting them into this four deployed marketer role where these people are basically business consultants.
And they understand that first, they're very AI native and they also understand business really well. They're deep in one area, but they're wide kind of across the board. Right.
And so that's good. But if you can give them. Give your people a recruiting workflow to manage, and you can have a recruiting workflow.
Basically, now we have these bots that are managing our applicant tracking system. They're managing our agentic reach out tools like Juicebox. And you can see here, it made the sequences.
And it can manage it because it can manage your computer, right? And so if you can manage your computer, not only do you give it connectors, but if you give it browser access, it can manage the web apps that you have. And GPT -6 Astra has gotten really good with browser use, right?
So you can see here. You know, hey, we're hiring an AI for SEO at Single Grain. We're calling it the Ford Deployed Marketer.
Think deep SEO, AU expertise, broad marketing skills, and a builder's instinct. Someone who works with clients and spots recurring problems and turns them into repeatable workflows. We're building marketing agents that live in Slack and Microsoft Teams.
This role is about shipping systems that improve results and give people time back. Okay, so initially the reach out wasn't that good, but I went a little back and forth with it, and it got a lot better, right? And so it's like.
The subject line, AI for an SEO, a builder role, I'd say great. Because people are tired of having to do the same things over and over. And they want to lead these companies.
And we know the psychology there. And so we've made it into the workflow. This is also running for other roles that we're recruiting for as well.
And now everyone can become an outreach person because you have these workflows, these reusable workflows that they can use. You're also thinking about reusable workflows for your customers as well. So that's what it is.
Now, last but not least, carousels. I'm thinking about reach constantly, right? I'm just thinking about how can it be easier for me to make stuff, okay?
So from one little video that I made on YouTube, it created these carousels over here. And we know that this type performs really well on my Instagram. And we said, okay, these are 30 traits of an AI native marketer.
It has the captions as well. Okay, so you can see here's the foundation over here, right? That's pretty good.
You can see, okay, build the foundation. Okay, six square slides over here. So we can have...
So basically from one piece of content, we've created four new pieces of content that can then be published to multiple channels. And so when I say kind of tongue -in -cheek that this has replaced my marketing team, the way I think about this is in many ways it has. It has replaced the people who don't know how to adapt, but it hasn't replaced the people who know how to run these workflows.
And so from an SEO standpoint, from creating content, or even this ManyChat format system that I have over here. Nobody has to log into ManyChat, which will then, that's the mechanism when you're like on Instagram where you're commenting on something and you're getting access and you get an automatic DM. ManyChat is the system that puts that together, right?
But it used to take maybe an hour. Even if you make a copy of automation that you have, maybe it takes you 30 minutes. But imagine you have to do 14 in a day.
Okay, well, your entire day is gone. And so you just have a workflow for this. This is a computer use.
And whenever I come up with a new video or a new resource, I can just say, Run the mini chat format skill. And then, you know, here's exactly what I'm trying to do.
You can just pull the transcript from this video and then boom, we're good to go. Right. Also, one of the reasons why I'm making this video right now is for future people that are joining a company.
They can just watch this and understand this is the way to think about this. Right. Because I will tell you the AI native people that are working at the company right now, they understand how this works and how.
the future is going to look. And by the way, when I interview people right now, it's like, Hey, how do you think the future is going to work? Cause like, I just think I'm going to be an auditor and a manager of these agents.
And those are the people that we want to hire. And by the way, if that's you, you can go to our single grain .com slash careers page because we want to work with you. All right.
So that's it for the video. We'll catch you in the next one.
The Hook
The bait, then the rug-pull.
Eric Siu opens by claiming an AI agent has taken over so much of his marketing team's work that it stopped being funny, then spends the rest of the video walking through the actual workflows, not just the results, behind that claim.
Frameworks
Named ideas worth stealing.
03:32concept
Nail It, Then Scale It
Get a marketing workflow working reliably for yourself first, then hand it off and scale it to the team.
Steal forany process you're about to hand off to a new hire or contractor
06:41model
Effort vs. Usefulness Matrix
Stop or rethink
Custom, Premium
Deprioritize
Reuse, Scale
A two-by-two plotting delivery effort against customer usefulness, used to decide which services should become reusable products versus stay bespoke.
Steal forpricing and scoping conversations with clients
06:52list
Three-Layer Service Cake
Managed services
Gateway + workflow access
Outputs and outcomes pricing
How the agency structures what it sells, from full-service delivery down to charging purely on results.
Steal forpackaging a service business into tiers
10:33concept
Forward Deployed Marketer
A hiring role for someone deep in one marketing channel but broad enough to consult across a client's whole account, described as a business consultant who is AI-native.
Steal forwriting a job description for a hybrid specialist/generalist hire
CTA Breakdown
How they asked for the click.
VERBAL ASK
14:10link
“if that's you, you can go to our single grain.com slash careers page because we want to work with you”
Soft close tied directly to the video's argument (people who can run these workflows are who they're hiring), not a generic subscribe ask.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
Eric Siu runs the same four marketing prompts — a website build, a thumbnail redesign, a growth-strategy memo, and a batch of social clips — on GPT Sol 5.6 and Claude Fable 5, and picks a winner based on which one actually finishes the job.
Claude Desktop and the Codex app both quietly shipped native, multi-tab in-app browsers within days of each other — used right, one prompt now opens, drafts, and stages a dozen tabs at once.
AI Edge's host installs seven free, high-star GitHub repos live on screen, turning stock Claude Code into a social-research engine, a browser operator, an ads department, and a 279-agent company.
Greg Isenberg and Remy put the invite-only Instinct agent through real errands, a haircut, a restaurant table, and a Bali visa, and find out it keeps a copy of your email even after you disconnect it.
A software architect sketches phantom-fs, the abstraction layer that lets dozens of Claude Code agents share one synced, conflict-free codebase from anywhere.
A 16-minute walkthrough of replacing a $99+/month ManyChat plan with a free, self-hosted Instagram DM automation, installed end to end by a Claude Code agent skill.
00:27
00:35
00:42
01:20
01:30
01:41
01:52
02:03
02:13
02:24
02:35
02:45
02:56
03:07
03:17
03:28
03:39
04:00
04:11
04:22
04:32
04:43
04:54
05:04
05:15
05:26
05:37
05:58
06:09
06:19
06:30
06:41
06:51
07:02
07:24
07:34
07:45
07:56
08:06
08:17
08:28
08:38
08:49
09:00
09:21
09:32
09:43
09:53
10:15
10:25
10:36
10:47
10:58
11:08
11:19
11:30
11:40
11:51
12:02
12:23
12:34
12:45
12:55
13:17
13:27
13:38
13:59
14:10
13:20
Video of the Day23:10
24:49
27:30
19:28
15:58