OpenAI Releases ChatGPT AI Agent Skills

0:00What's up everyone? Welcome to the Startup Ideas podcast. Today is December 22nd. I'm going to talk about a huge news item. OpenAI quietly launched something not a lot of people are talking about. I'm going to talk about one trend you could take advantage of, build a business around. I'm going to talk about one app that I'm using that, [music] you know, I I think you should too. I'm going to talk about one startup idea and just give it away to you. And at the end, I'm going to give you a mobile app framework so you can actually go and build a mobile app in 2026. And that's

0:30today's episode. The Startup Ideas podcast. [music] Sipping time, baby. So, the news item I want to talk about, um, that I don't think a lot of people are talking about is quietly OpenAI has launched skills. So, we've seen Anthropic and Claude skills, a lot of people talking about that, including myself, how amazing it is. But quietly, we have Codex now officially support skills. Skills are reusable bundles of instructions, scripts, and resources that help Codex complete specific tasks. You can call skill directly with dollar sign {dot} skill name, or let Codex choose the right one based on your prompts. So, what's really cool is they're just following the agent

1:18skills.io standard, the one uh by Anthropic. A skill is just a folder, basically an MD uh skill.md file for instructions and metadata. Um, and all of the sudden, your ChatGPT and Codex has just become a lot more valuable. So, there's a ton tons of ways uh that you're going to be able to use agent

1:42skills within Codex or ChatGPT. Um, you know, it gives some examples here in the official in the official announcement. Some technical stuff cuz it's talking about Codex uh specifically, but like a notion spec to implementation uh skill. A uh have Codex read and update your linear tickets uh skill. Have Codex automatically fix your GitHub CI failures skill. So, it does a lot of things and um you know, I think a lot of people don't really understand the difference between what is a skill, what is a sub agent, and what is an MCP. But so, just to you know, give everyone a little bit of a foundation, a skill is like a written

2:25guide that tells, you know, Claude or, you know, Codex or ChatGPT exactly how to do a specific task. For example, analyze spreadsheets this way, or write emails in this voice, or design, uh you know, design the interface like this. So, you can basically call and reuse the same skill again. Um, and it just it allows for more consistent output. So, what is a sub agent? Well, a sub agent is making a few extra copies of, you know, the LLM that you're using, you know, Claude or or Codex or ChatGPT, each with its own job, such as you review the code, you write tests, you update docs. The main the

3:06main LLM can hand out pieces of a big task to help these workers so work happens in parallel and stays organized. And an MCP is just a universal power plug that lets the LLM access other tools. Like we saw uh that, you know, it's going to update your linear tickets. Well, it needs access to that, and that's through MCPs. So, highly encourage everyone kind of,

3:29you know, plays around with this. Um, you can you can see over here uh on agentskills.io. I'll include it in the show notes. It kind of explains a little more about what uh what agent skills are. Says here, it's an open format, a simple open format for giving agents new capabilities and expertise. Agent skills are folders of instructions, scripts, and resources that can discover and use to do more more accurately and efficiently, and who doesn't want that? Says how, you know, what an agent skill can enable. Uh domain expertise, so package specialized knowledge into reusable instructions from legal review processes to data pipelines. New capabilities, where you

4:12can give agents, you know, new capabilities like creating presentations or building MCP servers. Uh repeatable workflows, interoperability. So, uh yeah, there's a a lot there. There's some example skills that they give as well on GitHub. And I just think that not not a lot of people are talking about it. So, I had to I had to let you know about it. Um, so that's the one news item that I think

4:39you should be paying attention to. That that could be super valuable for you in 2026. If you have a business that's doing at least $50,000 a month in revenue, I've got something interesting for you. It's called offline mode. It's a two-day event that me and my team are putting on at a 20,000 plus foot square foot mansion. Yes, this is what it looks like. On January 23rd and January 24th in Fort Lauderdale, Florida. I'll include a link in the description if you're interested in coming. But it's basically for people who have a business that's kind of cranking, but they really want to put it in rocket ship mode. They

5:17want to create a set of businesses that generate tons of money, tons of cash flow, tons of product market fit, tons of impact. Um, but they're not just quite there yet. It's also about, you know, making your business AI first, how you can actually, you know, build not just one product, but multiple products. And you're going to leave with, you know, tactical uh answers to your questions. So, um if that's you and this sounds interesting,

5:43uh I'll see you there. So, that's the news item. The the trend that uh I've been hearing about, um is is something called face yoga. Don't know if you've heard about it, but uh if you go I'm going to go onto Idea Browser and just check the trend face yoga. You know, I'm sure you've heard of yoga,

6:07obviously. Um, but people are cert you know, doing these like face yoga techniques to to get slimmer faces. Um, and there's a lot of opportunity here if you want to create different apps. Um, cuz there are, you know, a ton of opportunities um you know, in yoga, right? There's people making millions of dollars a year

6:29with these yoga apps. Um, but face yoga is kind of a sub niche that I think is interesting. Um, and is getting a ton of volume, you know, you're seeing you know, 110,000 searches in August. Um, the average [clears throat] CPC is about a dollar. The competition is low. Um, and I just think that you're going to see Yeah, I think there's an opportunity to build products, apps, you know, around face workouts, facial exercises, face yoga. Um, face lifting yoga exercises, jawline trainer, face slimming exercises. So, there's a lot of opportunity here, a trend I never heard of heard of, but

7:09face yoga. Face yoga involves facial exercises, stretches, massage to tone facial muscles, potentially reducing signs of aging like wrinkles and sagging by firming the skin. Much like body yoga strengthens muscles, offering a natural and holistic way to improve facial definition. Um, you can even see here people are creating uh YouTube videos on it. 10-minute cheek lift face yoga tone, right? Like there's an opportunity to build an app that's, you know, just some of these exercises, and and you can charge for it. Um, or make it free and do uh advertising or sponsorship. So, um face yoga, a trend I have recently heard of that uh hey, might get your creative

7:55juices flowing. Uh one app that uh I wanted to talk about, um that I've been using literally for 14 years. Um, and you know, some of you've heard of this one. It's the Things to Do list app. So, it's by a company called Culture Code. Not affiliated with them at all. And it's the to-do list app. People often ask me like, how do I how do I stay focused? How do I, you know, I'm cuz I'm working on multiple companies and and a lot of people work for you

8:28know, work with me and stuff like that. I keep everything in to-do list, and it's just Sorry, into Things, not to-do list. As my to-do list. And what I like about it is has like what I'm focusing on today, what is my upcoming, and what is something I want to do someday. Um, and it isn't a monthly fee, it's just like a one-time fee. Um there is some stuff to be desired here, I will say. Like I could use some AI features and stuff like that. But overall, if you're looking for like a simple to-do list app, I just find this Things to be the simplest the simplest

9:07one there. Um, and I've tried other things. I've tried more intense ones. Um, ones that are on like the GTD framework, and and sort of more Yeah, like kind of more the kitchen sink type app. But I don't know, I always come back to Things. Does what I need to do. And I recommend it. I recommend it for people who are looking for a simple to-do list app. Let's give away a startup idea. So, uh the startup idea of today is is one um that uh was on the like, you know, here you can check here. Every day for the last 24 days we've been giving away a

9:48free startup idea. Here's day 22. I'll include the link uh in the show notes if you want to You get these ideas for free, you just got to claim them. So, today's idea I really like. It says, "Call an expert service for non-developers stuck at 80% done." So, how many of us have built a Vibe Coded

10:07app and are like 80% of the way there. We use things like, you know, Cursor or Replit or V0, um and then we kind of just give up. So, the idea is something called Last 20. Last 20 is a phone-a-friend for builders who are almost there. Upload your code, describe the block, and matched with an expert who's hit that wall before and knows the way around it. 15 minutes on a

10:31screen share, that's it. Problem solved. Basically, uh hire a Vibe Coder, ship it. Maybe you charge $15 to $50 per session, or maybe you do a yearly thing for agencies, maybe it's like 100 to $500 a year who can't afford stalled client work. Experts set their rates, so you basically create this marketplace and then you charge uh percentage, maybe

10:53it's 5, 10, 15%. So many of us are becoming Vibe Coders, right? There's literally millions of us, but we're, you know, at the end of it we're kind of just like, "Ah, [sighs] how do I do this one thing?" Um and there's so many, you know, I think this is so true. There's so many YouTube comments of people being like, "How do I do this?" Agency Slack channels asking, "Has anyone shipped this integration for?" Okay, so here is the the main, you know, hire a Vibe Coder phone-a-friend uh opportunity. And if you uh look at the search data, it's

11:26absolutely beautiful. Um you know, AI helpers, uh AI tutors, all this sort of stuff is just going up into the right, and I I think that's why it's such a huge opportunity. Uh it scored a nine on 10. Uh the problem is a severe pain. I don't think it's, you know, the feasibility is a five on 10, like it's not that hard to do. Um and the why now, nine on 10, perfect timing. So, um you know, I think uh I always check the the value letter in Idea Browser, it's just some get your creative juices flowing around uh how you can actually go and build

12:04this uh this product. So, I would start with like a lead magnet around an AI project roadblock guide with with the goal just to get people's email addresses, get their phone numbers. Then you do a front-end offer, so premium access to an AI help tool, then the core offer, expert help sessions. You do like a pay-per-session access to expert advice, and then eventually you layer on the annual subscription offering

12:30priority access and additional features. Um and then eventually you do the agency thing, right? You do the enterprise package. So, I think this is sort of the ladder um and uh you know, the key stages here. Just like why I like the value letter, I've talked about this before, but it's basically you give free value to build trust and capture leads,

12:50that's the um the guide. You do the front-end offer, which is like a low-ticket product to convert leads to buyers, that's that premium access to an AI help tool, so maybe it's like a Chrome extension that allows users to capture AI project issues and submit them to experts. And I know someone's like, "Oh, Chrome extension, it's going to be so hard to create." Um well, with Cursor and things like that, you know, it it it's not out of the realm of possibility to create it, and you can use, you know, you could you probably can create something pretty

13:25lightweight that that does this. Um and yeah, so I I overall I think this is an interesting idea. I think the I think it would it would do quite well, and um I hope someone takes it. I hope someone does it. And uh please let me know how it goes and as you build it. Um

13:45cuz I would I I would actually use this. You can even, by the way, just do this idea for like one product. So, Replit has Replit Bounties, which is sort of this idea where you can hire someone to help you with Replit, but this is basically Replit Bounties for all the different Vibe Coding tools, Cursor, V0, you you name it. So, uh you can even just do like V0Bounties.com um and or just focus on one platform, or Cursor CursorHelp.com or TextACursor um TextACursorCoder or something like that.com. I do think that Last 20 is like a very catchy name, so so I I do think someone should take

14:30that idea, but um th- thinking about doing this on one particular channel first, one particular platform, and then eventually expanding this to all the different Vibe Coding tools, I think that's that's the move. We're going to end this episode with a framework. Uh this is a framework I saw on Instagram by a account called Builder Cult. Um so, it it's basically here's the six-step playbook to viral validation. So, let's just talk about, you know, if you're building an app, how can you get viral validation? So, the first step is warm up your account. So, you sign up to an Instagram account, you sign up to a TikTok account, warm warm

15:13it up. Watch videos in your niche every single day of the week, comment, like, share, follow like-minded creators, and this prevents shadow bans. Um so many people just create a brand account and then uh start posting stuff and it's not, you know, not really working. Well, it's not really working cuz the platforms are trying to prevent

15:34shadow bans. Step two is to design the idea. So, you you know, your app must have one visually heavily uh visually heavy element, example, pimple pop popping. Um must be explainable in three words, track your acne. This is just an example. And must solve a fundam- fundamental human insecurity. Uh lose weight, uh productivity, height, class, this sort of thing, right? So, you want to the the apps that do the best are just sort of the the simplest apps that you can explain so simply that, you know, just in a few words that you can very important uh have one visually heavy element because that's going to be

16:16helpful when you're creating organic content that you hope to get go viral. Step three is to build the simplest MVP you can. So, do something like literally three screens minimum, embarrassing embarrassingly simple. Um give yourself two to three days to build it. You can use, you know, rork.app, that's a mobile app builder. You can use anything.com, that's a mobile app builder. You can use Vibe

16:39Code app, that's a mobile app builder. Um but yeah, give yourself a couple days, iterate on it, show it to friends, show it to people, DM it to people, and and get their get their feedback. Post Step four is post daily until one goes viral, that's the goal. So, you're going to want to create one video per day for 30 days minimum, uh or four videos per week. You're going to want to test different angles of the same app idea, and hopefully one will catch. They

17:07say in this one will catch. That's not true, not necessarily it will catch in 30 days. It might take you 60, 90, 120, but, you know, it's shots on net. Um but you'll know when one hits because the comments will explode. Step five is to build the community um you know, before the product. So, you have this like waitlist, right? So, you create a Discord. Um they say Telegram, I'm not a huge fan of Telegram. Discord, WhatsApp for the waitlist, or an email newsletter for the waitlist. Post updates like building the app, here's what I'm adding. This keeps early believers engaged while you uh while you're building it and you're iterating,

17:44and converts into day one users. And then finally, launch the app once it's secure and stuff like that. Finalize features, convert the the rork [snorts] or anything or Vibe Code app into a full-fledged product using um cloud code. Launch with a hard paywall, something like Superwall.com. Um Post content daily, you know, different angles, same value, and then hopefully scale from zero to 10K MRR purely organic. So, that is today's episode. Uh I hope you enjoyed it. I'm trying this new format where I'm giving uh a news item, a trend, an app, a startup idea, a framework, all the alpha to give you the highest probability of

18:28winning because I want to see you win in 2026. If you've gotten to this far, please please like and comment and subscribe, share this with a friend. I read every single comment that you write, so I will see you in there. And uh yeah, have a have a creative day, my friends, and I'll see you next time.