How I Automated 90% of My Content Workflow (ChatGPT + Riverside) transcript

Today, I'm going to show you how I use AI to automate an entire content workflow to save me at least five hours a week. By the end of this video, you'll learn how to copy my exact three-step system to automate your own manual workflows. I'm confident that you'll be able to use AI to take back at least a few hours of your time after watching this. All right, so I publish a new podcast episode every Sunday. And for each episode, I used to have to manually research the guest, prepare an interview guide, record the conversation, write instructions for my video editor, and package the whole thing into a newsletter post, a YouTube video, social posts, and clips.

And this would take me multiple days to do, even with AI's help. But now, I use a system of AI skills that work together to automate 90% of this process while still letting me apply my taste and judgment along the way. This system has three phases. First, map every manual step of a workflow. Then, work with AI to create skills for each step. And finally, connect your skills together to automate the entire workflow. Now, real quick, this tutorial is brought to you by Riverside, the all-in-one AI studio that I use to record, edit, and produce my podcast.

I've been using it for years, and I love how it records at studio level quality and lets me edit and schedule clips to all platforms. You can try yourself by going to creators.ai.com/peteryang and use code peteryang to get one month's free. All right. Now, let me show you exactly how I work with AI. So, as I mentioned, the first step is to map every step of a manual workflow. Here is everything that I used to do manually to produce my podcast.

You can see that there are 15 steps which you can imagine is pretty painful to do and even working with chat GPT and AI to tackle all these steps in a single conversation just takes forever. So just from making this list you can see that there's three distinct phases to my workflow. First I have to prepare the podcast by researching the guest and then putting together an interview guide. Then after recording the episode, I have to send edit instructions to my video editors.

And last but not least, I have to do postp production, which is getting the transcript for the podcast, getting the episode uploaded to YouTube, drafting title and thumbnail combinations, writing a newsletter post, making clips, writing social posts, and more. So, you should also put together a very meticulous list like this. Don't miss any manual steps. The more instructions and context you can give AI, the more likely it is to help you automate everything.

So in fact, why don't we actually just copy this whole list into chatbt and ask it what work from here can you take off my plate and let's see what it comes up with. Okay, here I'm using a cognto mode because I've already built skills for a lot of stuff and I don't want those to impact the output. All right, so here is chatb suggestions. It's saying that it can own researching the guest, drafting the interview guide.

It can own finding the intro real. And then down here, it can say it can own clean the transcript, making the title and thumbnails, newsletter post, making the clips, and so on. Says it can own most of it with a little bit of guidance from me, which is exactly the flow that we want. Okay, so that's step one, which is to map every step of your workflow and then brainstor AI which parts it can automate. Now, let's move on to step two, which is creating AI skills based on these steps.

So, here are all the AI skills that I created to streamline my entire podcast production workflow. You can see that there's one skill for each phase. There's podcast prep for making the podcast interview guide. There's podcast edit for sending instructions to my video editor. And then there's podcast production, which is the most complicated skill, which is about packaging. And what this skill does is it walks me through these other skills to make the thumbnail title, show notes, the newsletter post, the social post, and even video clips one by one by calling these individual skills.

So, let me actually show you how everything here works starting with podcast prep. And we're going to use a real example throughout, which is how I produced this podcast about how to save money now with ChachiP finances with Ethan, the product lead. All right. So, here I am in chat GPT. And by the way, for all these workflows, I usually stick to GPT. So, medium. You don't have to go to Astra or something crazy to save on tokens.

So, medium is plenty good enough. But I start this whole process by saying, "Hey, I'm interviewing Ethan from OpenAI about personal finance. Please do research and help me prep an interview guide." And it's automatically going to call the podcast prep skill. And this is what it's going to come out with. So, I did a bunch of research and I found out that the recommended end goal for my interview with Ethan is to focus on practical personal finance use cases.

Here's some examples. Here's a bunch of links to the research that I did about chatb finances. And then it also does a YouTube scan for wellperforming videos in this niche. So, personal financing chatbt and so on. And it even suggests packaging options right off the bat for thumbnail and title combinations that I should think about. And the reason it does all of this is because, for example, if there isn't any interesting videos about how to use chat to manage your finances, then maybe this whole interview is actually not worth doing in the first place.

So, it's worth doing this research up front to figure out if this interview is even worth doing. Now, it then puts together the actual interview guide for Ethan. Here's the introduction that I to do. Here is the questions and some notes from example answers based on his research. So the main categories that it has is what chat GBD can actually do with your money. Something about open eyes finance benchmarks which is not that interesting and then also building a personal CFO that people can trust based on Ethan's background.

So what I did next was I gave it some feedback. I said look up Chat GB Finance docs or Ethan's tweet history and focus on use cases and here's the structure of the interview that I actually want. I want to explain what chatb Finance is. I want to talk about how to connect your accounts. I want to talk about use cases and so on so forth. And then it revised interview guide based on my feedback. All right. So here's the final interview guide that I iterated on with the podcast prep scale.

And there's three clear sections. There's questions. There's some notes. There's some use cases I want to demonstrate. And I got Chachi to write all this stuff in Google Doc and send it over to Ethan and the team for review. Okay. So that was step one, which is podcast prep, researching and preparing the podcast. Next, I recorded the interview with Ethan. And once again, I use Riverside to record the interview. Here is the finished recording.

It's 15 minutes long. And as you can see, it records everything locally, so everything remains high resolution, and you don't have to worry about getting disconnected. It also has a bunch of AI tools to help you cut dead air, cut repeat phases, and so on so forth automatically. Okay. But what I like to do is instead of using the UI here, I like to use the Riverside MCP and install that in Chat GPT and then give AI instructions to prepare the podcast for my video editors.

So, let me show you how that works. Okay, so we're back in ChattP and I told it to run the podcast edit scale for the Riverside Ethan recording. And the first thing it did is it made a linear ticket to include instructions for my video editors. And then it decided to go ahead and make the editor package. And the editor package includes the thumbnail title that we did before. But it also includes the intro reel. So every episode I make has a short 30 to 40 second intro reel that highlights the most engaging stuff in the podcast.

My skill basically puts together two different intro reels based on the written transcript of the podcast. just option A and option B. And then I gave it some feedback to combine the two. And this is the intro that I want. Then it went off and it put it into the linear ticket. And next, it also decided to make some clip candidates too. So again, it's using the automatically generated written transcript from the podcast and it's pulling out interesting clip candidates.

So the first one is about how you can audit nine days of spend with catchupt and it writes both the social copy and also the full transcript. [snorts] So I gave it some feedback again to improve the clips to tell it where to start and where to stop and then it's doing some more iterations. And then I said okay now update the linear ticket with all this stuff including the clips. So let's take a look at the linear ticket now.

All right. So, this is the linear ticket for my video editing team. And there's a few different sections that we work through. There's the thumbnail and title. There's the intro reel and where to pull the quotes. There is moments to cut. So, automatically it went through the entire episode and it found stuff to cut logistic issues or technical difficulties and it gave the timestamps for this. And then it also has the copy and the stuff for all the clips that I want to make.

So again, for me to do this stuff manually, listen to the whole podcast again and pick out interesting moments would take hours. But instead, I'm using AI to look at the entire transcript, find interesting moments. I'm still applying my taste along the way. I'm just not just letting it spit out his output and pasting it here. I'm giving a feedback to make it good, but it just saves me a lot of time. All right, so that is the podcast edit part of the process.

And now let's talk about podcast production, which is the majority of the work and also where AI saves me the most time. For every episode that I make, I also create this newsletter post for my Substack. And it has the video, it has the title, it has the timestamps from the video, the sponsor, and also takeaways and stuff that's super insightful from the podcast that you can see here. Okay. So to create the newsletter post, I basically use podcast production to call the podcast post skill as the first draft.

So here's the example. I say, "Hey, let's work on a newsletter post." For this one, I wanted to be super practical, focus on the actual prompts that Ethan shared. And here it drafted the first version of the newsletter post. Now, the thing is, the first version was not that good. There was a lot of issues. The prompts weren't quite right. It didn't quite use what I want. And I told it, "Hey, these are the exact six prompts that Ethan shared with us.

Can you please incorporate these prompts into the newsletter post? Okay. And based on the feedback, you can see here in the newsletter post that I went off and pasted the exact prompts that Ethan shared for how to use chat GPT to save money. So with the newsletter post in relatively good shape, I next told it to run the thumbnail title scale to figure out what the packaging is. And what this skill does is it looks through the transcript again.

It finds out what are the main topics discussed during the podcast and then it does a bunch of research on outperforming videos both from my channel and other similar channels. And based on this research, it suggests different combinations for the thumbnail and title. Now, if you are a YouTuber, you know how important the packaging is. Find the right thumbnail and title. It's just really important for people to come and actually watch the video.

So, here's five different packages that is suggested for me. And I thought about it and I said, "Hey, why don't we do these combinations instead?" And then we went back and forth and here is the thumbnails I came up with. And by the way, once you have a repeatable thumbnail template like I have for my podcast, you can also just tell AI to use Figma MCP or whatever tool you use to edit a thumbnail directly. Usually, even with AI's help, it takes me at least like 30 to 40 minutes to come up with the right thumbnail title just because I'm like super anal about it and I want to make it good.

Okay, so here's the combinations that we come up with and that we are going to AB test for this episode. All right, so going back to this map here, uh I also have a skill called podcast show notes that helps me draft YouTube and Spotify descriptions. So basically I told you to hey draft the show notes for this for me and here is the proposed show notes that came up with the right timestamps based on the uploaded YouTube video that my editors did for me.

It also has links to sponsors, links to my newsletter post, and where to find Ethan that it found itself online. A best practice is if you find yourself repeating the same template or the same process each time, just build a skill to streamline as much of it as possible. And just like with the newsletter post, this stuff isn't perfect, unfortunately, in one shot. So, I give it feedback to make it better. For example, name the actual use cases in the time stamps and iterations there.

And then I told it to add the show notes to the YouTube description itself. So it uses browser use to copy what we agreed on and paste it here. And it can also click all these buttons to turn on monetization to make sure that it's part of the right playlist and so on and so forth, which again is just like a lot of manual work that you can streamline with AI. All right. Now, going back to the newsletter post, I wasn't fully happy with what AI came up with.

So I decided to make a bunch of manual edits to the post directly. And here's something that I think all of you can learn from. So before I made the manual edits to the post, I asked it to take a snapshot of the current post that had AI's edits and then I'm going to make manual edits next and then I want you to compare the two. So it took a snapshot. Then I made my edits and then it found that here's what changed. And and the reason I told it to explain all this is because I wanted to update the scale so that it can oneshot and avoid these mistakes the next time.

So you can see here the next prompt I gave it is summarize all the before and after changes to the podcast post scale to be able to oneshot this next time and list these changes in a number list so that I can accept or decline each change. So it went off and listed everything in a number list here and I actually read through and reviewed everything and gave a bunch of feedback and then it given me a new list and I said hey make all the updates to the scale.

So this is like a pretty common process that I use to improve my skills over time so that it doesn't repeat the same mistakes over and over. All right, so going back to this map, we have the thumbnails, we have the titles, we have the show notes, and we have the newsletter post. So, next step is to make a social post and also prepare and publish clips. So, to make the social post, I have my podcast social skill. I tell it to give me two teasers and two podcast posts.

So, here's a bunch of teasers that I wrote. And I like how you can click around to figure out which teaser I like. And also, here is the podcast post. And by the way, I came up with this stuff because I give each skill like really good examples of past teaser and podcast post that I wrote and tell her to follow the same style and the same guidelines. But you can see again it doesn't make it perfect in one shot. So I have to make manual edits.

I can tell you to pick a better intro quote and so on so forth. And then I told her, okay, I'm pretty happy with this. Why don't you go find a good screen cap from the YouTube video where we're both smiling. This one, Ethan is not smiling, so I want to find one where we're both smiling and pretending we're having fun and then attached it to the teaser post. Okay, so it went off and it found this screen cap. Here we go.

It looks like we're both having a good time. Then it attached it to the schedule post. So, let's go ahead click on this. You can see here that this teaser post is now scheduled for next Saturday with the copy we agreed on and also with the image that AI found. All right, so last but not least, AI can also help you find interesting clips from the podcast and schedule them to social media. And to do this, I just say video post.

Now, let's make a clip. Give me your top suggestions. Make chat GPT execute your tax plan number one, find hidden credit card rewards number two, and so on and so forth. And down here, I told it, hey, I really like option three because of the banter that we have. and go ahead and make the clip with Riverside. Let's take a look. So, if you use the Riverside MCP again and then made the clip here for me to review. And you notice how I cut out a bunch of pauses and other unnecessary words to make the clip that I actually want.

So, I reviewed the clip and I also reviewed the social post that goes along with the clip and I said, "Looks good. Go ahead and put both on Typefully, which is my social scheduling app to schedule the post." So, let's go ahead and review the post in Tyfully. And here we have the social post along with the full clip in Typhoia as well. All right. So, there you go. That's a super quick walkthrough of how I built a series of AI skills to streamline my entire podcast production process.

Now, as I'm showing you, I'm still really involved in each step to review AI's output, to make manual edits if necessary, because I don't want it to just pump out slop. I want it to be high taste and high craft. But these skills and chain them together really helped me save a ton of time. So just to recap, if you want to do this for yourself, because I know not everyone has a podcast, is just pick one or two manual workflows that are taking up a lot of your time every week.

And then the first step is to just list all the manual steps in each workflow. So here's 15 manual steps from my podcast process. Then just ask AI, ask or ask claude, hey, here's all the manual steps I take. which parts of this can you help me streamline? And the more mcps and plugins you connect chatpt and cloud to, the more likely it is to streamline more of these steps. Then once you list the manual steps, there's clear patterns that emerge.

There's three different phases to my podcast process. So just build the skills with AI for each phase. And by the way, I want to actually show you what these skills look like. So for example, this is the podcast prep skill that I use to prepare the interview guide. And this looks like pretty complicated and pretty intimidating, but to be honest with you, I didn't write most of this. I gave AI instructions and I came up with a workflow.

So here's the name of the scale. Here's when to use it. And here's how to use it. Exactly. Similar thing for podcast edit scale, there's a list of steps that AI should take. And also for podcast production scale there's a list of steps and there's a whole orchestration and routing layer of different other skills it should call based on the sequence that I'm in doing post-production for the podcast. So the important thing is to not get intimidated by all this and just remember my three steps map every manual step of a workflow.

Create AI skills based on these steps and connect these skills together end to end to automate as much of the workflow as possible. And real talk, I feel like I probably need to be better at actually hiring a human to help drive AI to do a lot of stuff instead of just doing everything myself because even with this fancy system setup, I still take a lot of time to produce my podcast. But yeah, I just want to show you guys how I actually do work with chat GPT.

Instead of just doing a bunch of demos or trying a bunch of new tools, this is how I actually try to save time and get work done. All right, so that's it for this episode. I hope you enjoyed it. For more of my skills and workflows, you can go to behindthecraft.com where I've included a bunch of copy and paste prompts that you can use. And as always, thanks for watching and please like and subscribe for more practical nobullshit tutorials and interviews.