The AI Skill Path for Founders
Skills:
AI Systems Design90%Prompt Craft80%AI Tools for PMs80%Advanced Prompting70%Agent Foundations70%
Key Takeaways
The AI skill path for founders is a map to help them improve their AI skills and see the benefits for their business, with six levels of progression from land lover to admiral, utilizing tools such as chat GPT, cla Gemini, and perplexity.
Full Transcript
We are all trying to figure out this AI thing, but for us founders, it's especially important. We now live in a world where where we feel that if we don't keep up, we're going to be left behind and someone else is going to take our cake. And the biggest problem of all, there is no clear map, no path that we could look at and say this is worth investing my time in to get better at this AI thing. And this is going to be the outcome that I get. I couldn't find anything, so I built my own here. I'm going to share it with you today. It's going to help you see where you currently are and what it already unlocks for your business. And then of course also showing the path of what to do next and if it's worth investing the time. As always, you can read this here on pirateskills.com. You can watch this on YouTube or you can listen to it as a podcast on Apple or Spotify. Have it your way. But some things are going to be very interactive here on the website. So it's definitely worth shopping by. All the links are in the bio. As always, as I said in the beginning, there is no map and that's a real problem. Everything keeps shifting under our feet. Next week, there's a new best model. The skills we might have developed or tools we sought after are irrelevant now because the base foundation model because the base foundation model can now suddenly do it out of hand. Underneath every level in this AI skill path, there is one thing that remains steady. That is the relationship to initiation. Who is going to act and when? In the lower tiers of AI usage, it's it's very reactive. You go to the AI, you say here, let's talk about this topic. Let's re research this. But you are the absolute bottleneck here. Then if you progress, you start building with AI and you build little software tools that make you more productive. It's it's really active, but you your user base can now use this now at a at a higher level. But where you want to get to is proactivity where the eye comes to you and where it suggests a menu of things and you become the curator of these ideas. And this is the fundamental thing that happens from level to level that you give the AI more agency and you become less and less of a bottleneck. So here we are the actual AI skill path. As we are pirate skills, I was allowed to have a bit fun with the title. So this is the path from land lover to admiral. Level one is the land lover. Imagine the typical chat GPT user. You go to the AI, you ask a question, you get an answer. It's all in one session. Next time you open a chat, there is no memory. But even there is a lot of value we can unlock for our businesses. We always look at the builder and the growth side. So developing a product and growing that product into the marketplace. On the builder side, we can now do smart searches uh and replace Google. A lot of times if we don't understand something, we can get an explainer. We can just feed it presentations or ideas of what we want to build and it can process that. It could break down complex builder tasks into its components. And on the growth side, And on the growth side, we are learning to draft captions for example, do strong or do stronger idea brainstorming, cold email drafts, content outlines. All of this is now possible even though not at the level where we maybe we are happy with the tone of voice. The typical tools people on level one are using and this is not a judgment of the tool. This is just the level at which they use that those tools are chat GPT cla Gemini perplexity and you could see that each level unlocks certain benefits for you and you unlock this level by having gotten these benefits from it. You can be pretty sure you're not on level one yet. We have gone through this. We have progressed to what did we progress? We might have progressed to the deck hand. Still not a great position on a captain's ship to be on, but this is where you are now more sophisticated users of these tools. On the biller side, you can now persist context by, for example, using chat GPT projects than just normal standalone chats. You might be more sophisticated in how you prompt those tools so you get better more reliable results. You might be even using it as like a copy paste programmer where you say like okay write me this code and then you take the code put it into uh another tool and then there is your website. Very tedious and annoying. Or you might be using on the cloud side of things um tools like artifacts where little mini web apps are already within the cloud world. You might be using clot code a co-work for this but it's it still feels like you don't own anything. There is nothing special that you can later share uh with other customers. Maybe a custom GPT is like the first edge that you can get here at level two that has value beyond your own workflow. On the growth side, you can now, for example, in this chat GBT project, define a clear brand voice system and make sure that those captions, articles you can now write in collaboration are much more to the level that you want the quality and the tone of voice to be. You might have discovered an amazing feature called deep research. Like many of those um tools now have this where they really spend hours sometimes to build this perfect paper for you uh 50 to 80 pages long doing a deep like competitive research or market analysis. You could even give a task like lead research especially like with claude co-work now you could have it collect people from LinkedIn research them get their email addresses. It's it's still it's not manual anymore. that's that's not it. Um, but it doesn't feel like like a system. It's more like I do it one off and then I do it again. Can even create great AI images and video, especially like with Gemini's nano banana model. It's insane. Or the VO3.1 model. or you're using um a tool like notion in AI where you have a lot of context um around your project available to the AI and you managed this context very consciously. You're still reactive but you now unlocked a lot more values out of pretty much the same tools. Let's progress to level three, the navigator. You're not just building for yourself anymore and improving your own productivity, but you are often now doing the quintessential VIP coding thing by building simple web apps or internal tools that help you um be more productive, but that you can now also share with your teammates and with your customers. And it's the moment where you have this experience of I can talk to my website and voila there is a change and if I want something interactive or dynamic or a little database that is now totally possible. Yeah, you can prompt something and there's going to be maybe not a full product but more like a prototype that comes comes out of it. And on the growth side you can now build yourself a little tool that does a whole batch of creatives. Imagine you have um a type of quote that you want to do. You research 30 quotes. They can all be styled and batched in one go. You can build workflow automations with tools like Nitn. You can have a prompt and boom, a new landing page for a specific audience shows up. You can build yourself highly personalized email sequencers with tools like Clay. This is this is a very interesting level to get to. And if you do not want to become too technical, this is a good point to arrive and stay at for for quite a while. This is still happy land. This is where you mostly see a box, things happen magically on the other side, but those platforms like Lovable, Vzero, Base 44, they take a lot of the hard decision off your shoulders and let you just do the thing. You just have to be able to live with the compromises. But this level three is for lots of business types that don't require a lot of software, a perfect place to get to. You will see the advantages as I um explain the progression, but navigator is the first serious level uh that gives you an immense amount of benefits beyond chat GPT. On level four, you have to make a big jump. Getting away from tiny chat boxes, facing the terminal from time to time, and owning real decisions you make that normally engineers would do. You have to decide about a lot more tools that you use and how you want to set up your tool stack for the stuff you want to build and how you want to grow what you have built. This is this Corsair level is where you take a lot more ownership which is a good and a bad thing. Good thing is you truly own everything around your codebase. If you are if you're feeling locked into a service, you can just get out, switch it, no problem. You get access to amazing features that will trickle up the levels, I'm pretty sure, like agent skills, which is a way to clearly communicate resources and clear processes to an AI that make them much much more reliable when they would otherwise maybe feel a bit stupid. And you can now connect with MCP servers to external tools. That gives your agents the ability to do a lot more of the work for you. And what also lots of developers love is that you can have everything documented automatically without worrying about if someone else comes into the team that you have to explain everything from the ground up. Everything can be documented by those tools. The tools that most builders use here is as I often say like one central AI hub, an AI IDE like cloud code, codeex, cursor, anti-gravity and they keep on improving and keep coming better. You're probably going to host your website or your web app on Versel. You have a database somewhere like Superbase, but you can do a lot more things here. This is the first level where you can seriously consider building building a full-fledged uh software as a service either B2B or B2C and also being able to fix something if something goes really wrong. This is the level where people need to progress to when they kind of maxed out on the chat builders and they're trying to fix things or improve things, make them more performant or more scalable. This is on the builder side. But on the growth side, you also now have way more opportunity. You can choose your analytics tool as you like. For example, as you know, I love Post Hog and you can just use your code agent, have it implement everything for you, build the dashboards. You can now track the behavior of your users much more sophisticatedly. If you want to set up an AB test for your landing pages, your onboarding journey, for your signup experience, that is now a prompt away and you own the entire funnel. You don't have a gazillion tools running around. This is the time where you build your own stack and and most of the customer experience happens in places you have vibe coded yourself. So, you are in full control and no one else can take it away from you anymore. By the way, this is the level that I focus on getting you to in the builder forge and the growth forge. Full control, no babysitting tools. This is the real thing that you could just stay on and and grow your experience, grow the software that you're building, grow your audience an infinite amount. So, what should there be beyond this awesome Corsair level? I want you to think about this. This is still us being very interactive with AI, telling it what to build, telling it uh what to do, what to analyze. This is still not the AI coming towards us and proactively helping us to build the software and to grow it. That is what level five is for. At level five, you're now the captain. You command the mission. you are in control but you let the AI do things for you on the builder side. That means instead of before you are handling everything in cloud code, you give an agent like open claw or agent zero access to cloud code and it codes for you. This is amazing allowing you to do things like remote coding where you are just on your Slack or your website or Telegram chat with OpenClaw and tell it hey I got this idea for an feature or could you complete this I had to get in a taxi to go to a meeting and see some friends and it it can do so much for you. You can let it, for example, review your own PRs or those of your colleagues and have it check maybe for malicious code or exposed API keys and and warn you if something happens. You can have it read your logs on like an 2hour interval. And I do this for example, I let it read the forcell logs, all of my outgoing emails and look for anomalies. And especially in the beginning when I did this, it found so much And the good thing is it could then propose the fix already for me for me to review. It already maybe had uh done a little preview scan, but I could just say like, "Oh, this solution is great. Let's publish it." And I was done with it and I could do this from Slack. Yeah. On my phone on the go. That was amazing. On the growth side, you can now build system where you autonomously build complete content loops. Imagine having it research the news, then turning this news into a newsletter and sending it out uh to your audience by just to the ones that are the right fit for it. You can let it loose on your CRM and be the true intelligence in your system. I don't touch my CRM anymore. I just talk to my agent that then manages this CRM for me. I'm on this level five captain level. Um I'm not beyond that and there is so much still to be explored on level four and five uh that I guess that I will be quite busy on this level for a while. I love getting these proactive daily briefings. Uh you can let it uh search your calendar, your emails, um your to-do software, and be proactive about suggesting what to do, even acting on it already if it feels uh like it can do significant work for you up front. And then of course if you think about things like lead enrichment can have it pull in dynamic data on a daily basis. Research for strong signals that a company or a person in a company is currently looking for a solution like yours. Research them. Enrich the data in your CRM. present to you with a solution uh on how to approach this customer or doing that automatically for you. That is level five. This is what people who talk about open claw are so excited about. And I could just pick four use cases that I found particularly salient. Yeah. For the builder and the growth track. But there's more more and much much much more to be discovered. like on level six. Level six is where you really become the admiral. Instead of just communicating with one agent, you're orchestrating multiple agent like a complete fleet that is working harmoniously together. And there are a couple of things that can be done at this level that just was not accessible before. And I see most companies in the future raising themselves up to this level to get a competitive advantage. On the builder side, you're now managing a master agent that has lots of sub agents below it, directing the work that you said you wanted to get done. You could do things like ensuring that if anything is wrong in your codebase or your logs that maybe you have a monitoring agent that informs a fixer agent and a quality control agent and that helps you automatically fix everything that is broken or that causes an error in the customer experience. For security measures, you could have adversarial review, meaning you could have a red team trying to break into your uh tools and a blue team trying to protect that, becoming better and better with every iteration. And if something bad happens, you can have the system react to these incidents and if confident in having a fix, already put that fix in and only wake you up. if it couldn't make that fix immediately in a given period of time. On the growth side, imagine having a completely autonomous growth engine, building your content, doing the outreach, and becoming better over time because it's doing auto research on what is actually working in terms of content, ads, funnels, and it's optimizing itself over time. And imagine this intelligence on your CRM being able to pick up on realtime signals and acting on your behalf so that fewer and fewer deals fall through your hands. There are dedicated tools who can do that. Of course we are all currently talk about open claw but there have already been pretty GTM go to market oriented tools like relevance AI who are doing this or crew AI who have started with the idea of multi- aent setups quite a long time ago of course agent zero as the more developer oriented version of this admiral admiral level uh is still definitely beyond me I'm very curious uh where this is going to take because I'm excited yeah for this future and for the freedom this is going to give us as founders so we can focus on what is most human understanding the needs of our customers and thinking about what are the solutions and where do we need to move long term so do you really need to go all the way does everyone need to become an admiral in this uh skill journey I don't think so every business type has different requirements and let me just give you some examples so this becomes a bit less abstract. Let's say you're building a service level business. It's totally fine if you're on the navigator level. This is where you use tools like lovable v 0ero base 44 where you can just speak build a little tool for your clients. uh having your service delivery more modern on a better website uh with a login for your customer. Really impressing them with things that normal service businesses wouldn't be able to do or couldn't afford the full-time developer it would normally take to do things like that. Then you may you may be having an info product. Yeah. Where you're selling a course or workshops. Here you can leverage the navigator level to batch your creative, automate your content workflows already. But without them everything has to be done a lot more manual and it doesn't compound as much. Then let's think about like classic agencies. Classic agencies are like marketing agencies, sales agencies, etc. They would also benefit from that level. This is where they can build a lot of internal tooling that would automate the process increasing their margin so not everything has to be done by humans in your business. It's not required to have like fully running automated agents for your customers but empowering your internal users to get a lot of the repetitive boring grunt work out of the way with some internal tools. Then there is this new beautiful concept of like an agency as a service where you are actually like providing something that feels uh like a service like providing daily content for a certain customer but what is actually running is software and large language models in the background. You can sell this like an agency service. Uh but you do not have to do the work like an agency. There I would say you would have to get to the level where you can build your own SAS. That is the captain level which is level four. And you will get the most benefits from going through the the hard jump from this less technical navigator level to the more technical captain level. And of course, if you plan to build any type of B2C or B2B SAS product, uh, which I'm most focused on with the growth lab and the builder lab, then you need to be able to own your codebase. You need to be able to fix real problems in production. You need to have an understanding of the underlying modules that make up a SAS. Yeah, this is uh what the captain level is primarily good for. You can make good judgment calls and do changes if something bad happens and it's worth investing the time in this really makes you the founder in the AI. So you see none of these business model require level five or six where you have completely autonomous running agents around but those agents will do a lot of the lifting in any of those business models as well and you can save a lot of time deliver a higher level of quality and quantity of output for your customers for your teammates and I think once you are on this journey the progression path you you will want to continue. But as I said, you don't have to. So the answer is easy. Not everyone has to become an admiral and build multi- orchestrated agent swamps. You can just use what is working for you right now. Get good at it and like unlock each and every benefit for you over time. And now you have a clear progression path uh to go along. And I hope this helped you and helps your colleagues and your friends who are also on this path. I'm looking forward to your feedback. Thank you for joining me so far. Bye-bye.
Original Description
Everyone's using AI. Nobody knows where they actually stand.
There's no shortage of AI tools, tutorials, and hype. But there's no map. No way to know: am I a beginner who thinks they're advanced, or someone further along than I realize? And what should I actually learn next?
I couldn't find a clear progression path, so I built my own. The AI Skill Path is a 6-level framework — from Landlubber to Admiral — that shows you where you are, what you already unlock at that level, and what it takes to reach the next one. Each level covers both the Build track and the Grow track.
This isn't about collecting tools. It's about knowing which capabilities matter at your stage — and ignoring everything else until you get there.
WHO THIS IS FOR
- A founder who uses ChatGPT daily but isn't sure what the next level looks like
- A vibe coder who ships apps but hasn't touched AI marketing yet
- A marketer experimenting with AI content but not building anything technical
- Someone drowning in AI FOMO who wants a clear, honest progression path
IN THIS EPISODE
00:00:00 – The problem: there is no map
00:01:11 – The principle underneath every level: relationship to initiation
00:02:39 – The 6 levels overview
00:02:52 – Level 1: Landlubber — the ChatGPT user
00:04:36 – Level 2: Deckhand — persistent context and better prompting
00:07:19 – Level 3: Navigator — vibe coding simple tools and automations
00:09:30 – Level 4: Corsair — owning your stack and building real SaaS
00:13:43 – Level 5: Captain — proactive agents working for you
00:17:29 – Level 6: Admiral — orchestrating multi-agent systems
00:20:36 – Do you really need to go all the way?
00:24:48 – Find your level and start from there
TOOLS & RESOURCES MENTIONED
🖥️ Claude Code / Cursor / Windsurf (AI IDEs): https://anthropic.com / https://cursor.com
⚡ Vercel (hosting): https://vercel.com
🗄️ Supabase (database): https://supabase.com
📊 PostHog (analytics): https://posthog.com
🤖 OpenClaw / Agent Zero (autonomous agents)
🎨 Lov
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
More on: AI Systems Design
View skill →Related Reads
📰
📰
📰
📰
Death by Hallucination: Your Agent Promised Lifetime 50% Off to Everyone
Dev.to AI
Your AI frontend agent doesn't need more autonomy. It needs a definition of done
Dev.to AI
The Agent Can Own the Loop. It Cannot Own the Receipt That Ends It.
Dev.to AI
Agentic Browser: ~98% fewer tokens than HTML for LLM web agents (Python + MCP)
Dev.to AI
Chapters (11)
The problem: there is no map
1:11
The principle underneath every level: relationship to initiation
2:39
The 6 levels overview
2:52
Level 1: Landlubber — the ChatGPT user
4:36
Level 2: Deckhand — persistent context and better prompting
7:19
Level 3: Navigator — vibe coding simple tools and automations
9:30
Level 4: Corsair — owning your stack and building real SaaS
13:43
Level 5: Captain — proactive agents working for you
17:29
Level 6: Admiral — orchestrating multi-agent systems
20:36
Do you really need to go all the way?
24:48
Find your level and start from there
🎓
Tutor Explanation
DeepCamp AI