0:00 If you want to build a bootstrapped SAS startup, you need to do exactly three things. Build stuff, sell stuff, and create systems and processes so you can build and sell stuff faster. That's all you need to do. No more, no less. But there has been quite a shift, especially when it comes to building. I'm a software developer, which means I knew how to code before we had AI. And when I started my journey back in 2021, being a technical founder was one of the biggest competitive advantages. After all, we could do ourselves what other startups had to pay an engineer 100,000 a year to do. But with AI, things are really changing very fast now. For the past 200 years, many roles have been replaced with technology.
0:46 And it seems like software engineers, UX designers, and other IT professionals are the next in line. The good news here is that the barrier of launching and running a profitable bootstrapped sales business is now lower than it has ever been. And the bad news, well, being a technical founder just isn't the mode it used to be. So, if you want to build a business for yourself in 2025 and moving forward, you absolutely crucially need to adapt to this new era of AI, whether you're a technical founder or not. And in this video, I'll show you the six AI tools you need to master in order to build stuff, sell stuff, and create systems and processes to build and sell stuff faster. Let's get started. If you're new to this channel, welcome here. I'm Simon Harberg
1:33 and I run a SAS portfolio of four tools to handle social media, link tracking, customer support, and graphics design. And we're on track to hit 2 million AR this year. We are just four people on my team, myself included. And the reason we're able to stay this lean is because we're also using AI extensively for both building, marketing, and operations, which is exactly what we're going to talk about in this video. Let's start with the first category. building. For a long time, building products roughly fell into two groups. Either you'd be a professional coder building everything from the ground up, or you'd use no code tools like Bubble or Web Flow, which is fully UI based, drag and drop, and requires no coding skills at
2:18 all. Obviously, building a product from the ground up comes with the massive disadvantage that you'd need to be a professional software engineer with several years of professional experience in order to realistically build something great. And using no code tools like Bubble comes with the massive disadvantage that you'll get very limited in what you can build very fast. But now we have something that goes somewhere in between. Andrew Carpathy, one of the co-founders of OpenAI and former head of AI at Tesla, calls this new concept vibe coding. It's a way to use AI and LLMs to build products by only expressing what you want the application to do and let the AI take care of the rest. One of my absolute favorite tools to do this is called Lovable. With Lovable, I can turn
3:06 my ideas into websites and functional apps using just natural language. I simply describe what I want and Lovables AI will start building it right in front of me. [Music] This works in iteration. So I typically start off with one prompt and Lovable will start building. Then I give it another prompt with feedback and things I want changed. Then it follows up. We keep going back and forth like this until I'm happy. Now, the awesome thing about Lovable is that if you're non tech and you don't know all the fancy programming terminology, Lovable does an insanely good job at simply filling in the blanks for you. It will make very reasonable assumption based on the information it has. Now, I'm a programmer, so I understand code. So, I
3:52 can click this dev mode button right here, and I'll see the exact code base it produced. So here I can confirm that Lovable did a very good engineering job following many best practices. And if I do know some programming lingo, Lovable obviously understands that too. So for instance, I can tell it to use Chakra UI as the component library here and gets it and it will start replacing all the components to use Chakra instead. Just amazing. Now, if I were non tech and just wanted to publish this app, I'd click the publish button and Lovable helps me take it live. But since I'm a tech founder, maybe I want to handle hosting on my own. And at some point, I may want to continue this
4:38 work using other tools. Lovable totally allows this. I just click here to sync this with my GitHub repository. Now, the code is live on my own GitHub account and I can take it from there. And if I want to continue working on this project with Lovable, it syncs any changes to the code back into Lovable from the same GitHub repository. Incredible tool. On my team, whenever we got a new idea, we stopped doing all the wireframing and prototyping in Figma. Total waste of time. We just spin up Lovable and start jamming. As Andrew Copathy said, the hottest new programming language is English. This doesn't mean that we shouldn't write code at all. And it's obvious that the more techoriented founders will benefit from this concept the most. But it does mean that you can
5:25 now get very far without being an actual professional schooled software engineer. And even if you are, you can increase your output by a whole lot just by using tools like lovable. The next tool in the build category is cursor. So, Cursor is an AI first code editor based on VS Code that packs a whole range of features that are going to boost the productivity of you and your engineering team. It's a much more advanced tool and it is much more tailored to professional engineers. Though, it still works on the same principle using natural language to prompt AI to do most of the coding. Cursor is aware of your entire repository and it's able to understand it, analyze it, ask questions about it, and execute tasks that are consistent
6:10 with your entire codebase. When needed, cursor will perform web searches when working on your project directly from inside the editor. It's able to search and combine information from your codebase and external documentation across the internet to perform complex operations across your project. If you're using a language like TypeScript which will highlight semantic and logical errors in your code, cursor will use these very intuitively to correct itself and make your code better. If you're developing for the web, you can give it access to the browser so it can see the errors in the console, which again gives it the ability to correct itself and iterate on its own solution without you having to intervene. Recently, I transformed an entire code base just using cursor. I gave it a very clear task description and it would just
6:55 go at it for hours all on its own while I was doing something else. On my team, we typically combine Lovable with Cursor. So, we start out with Lovable, sync it to a GitHub repository, pull it down, and open it with Cursor to do the more advanced stuff. Sometimes we then go back to Lovable to do more lightweight stuff, and we'll go back and forth like that. Now, there are alternatives to these tools. Instead of Lovable, you could use Replet or Bolt. And instead of cursor, you could use Klein or Windsurf. But the main point here is that if you want to truly adapt to the age of AI, you need to master a vibe coding tool and a more advanced AI coding tool. I think Lovable and Cursor are the best on the market right now. Now we got building covered. Next, we need to sell. This falls into the
7:42 category of marketing. This category is a little more tricky because marketing really can be a whole lot of things. And there are already a lot of AI tools doing copywriting, creating ads, doing lead generation and outbound, qualifying prospects, and much much more. And which tools you should master here really highly depends on which type of marketing you're doing. On my team, 95% of our marketing efforts are focused on one thing. And as a bootstrap SAS business, I recommend you focus on that, too. and it is creating content. As you've probably noticed, good social media content today goes beyond good copywriting. Today, it also highly depends on strong visual creatives, both graphics, images, and videos.
8:29 So, the first tool we're going to look at is Cling AI. Cling AI is a tool that can do image and video generation using either a prompt or reference images. It's a really powerful tool because you have it all in one platform. For images and graphics, I use two modes. Text to image works really well if I just want something quick. Though often enough, I want Cling to create images similar to a specific style. Either to make it consistent with other images I post or sometimes I find something really cool on Pinterest and I use that as a reference. That's for images and graphics. But Cling AI really becomes truly powerful when it comes to video. It is best-in-class when it comes to generating highquality video. I use
9:15 these clips for social media, but I also use Cling AI to create small set extensions like these laptop clips which I can reuse for anything that needs to be shown on a screen or even more advanced 3D graphics like the one you saw earlier in this video. For this, I use Cling AI to generate some of the video material. Though this does require a little extra work in a 3D tool like After Effects or Blender, I use Cling AI heavily in my content. So whenever you watch one of my YouTube videos or posts on LinkedIn, there's a high chance you're looking at something Cling AI had a part in creating. And if you want to get on top of highquality content in 2025 and moving forward, you have to master a tool like Cling AI. The next tool we need for marketing is FeedHive,
10:01 which is an AI powered social media tool. It's one of the best tools on the market to manage and create content and have it go out on all major social media platforms. The old way of doing social media marketing typically included a large team, manually crafting every piece of content, and perhaps using a dusty old tool like Sprout Social to schedule the posts. With FeedHive, you can reduce all of this to a single person managing the content and you can still show up everywhere with high quality content. Feedive uses AI in three crucial areas. First of all, to help you write, it has a built-in AI assistant you can use to vibrate your content. The AI that's used for this is a fine-tuned model trained on millions of high-erforming social
10:48 media posts. So, it intuitively knows what works really well on social. Secondly, to help you grade your content. After you're done writing, FeedHive will use AI to produce a performance prediction score. This is AI estimating how likely this post is to do well. Also, here it's using a fine-tuned model, but it's also taking into account relevancy and which topics are currently trending. Finally, to help you recycle. I cannot overstate this. Simply recycling old content is a massive productivity hack. Not every single post you do needs to be brand new. Feed Hype will look through your posting history and it will use AI to suggest pieces of content that it believes will perform well if recycled. It will obviously use
11:34 previous performance as an indicator, but it will also take into account how recent it was posted. And sometimes it will suggest a post that flopped originally, even though it has all the characteristics of a post that should do well and probably would do well if you try posting it again. You can also do basic image editing with Feed Eye, but I highly suggest using a specialized tool like Cling AI and simply bring those creatives into FeedHive and use it for your social media content. Now, full transparency, FeedHive is one of the SAS products my team and I run. So, I have a bias here obviously, but I do firmly believe that FeedHive is one of the best, if not the best AI social media tool on the market. And I don't know of any other tools that cover both writing, performance prediction, and recycling
12:19 using AI. And not only that, you can also get instant lifetime access to FeedHive and all its cool AI features for a single one-time purchase. That's right. We're actually selling access to this tool as a lifetime deal. If you go to foundersstack.pro, you can buy access to my entire SAS portfolio consisting of four powerful AI tools, all included for a single one-time purchase. I'll leave a link in the description below. Of course, there are alternatives here, too. Instead of clinging AAI, you could use Runway or Luma. And instead of FeedHive, you could use AI tools like Predis or Ad Creative, which are also good for social media. I personally think Clingai and FeedHive are the best tools on the market for
13:05 this job right now. Great. So now we know how to build and we know how to sell. Now we need to create systems and processes so we can build and sell faster. The last thing we need to master is ops and automation. The most important tool you should master here is NAN. NAD is the most powerful automation tool on the market right now. You can use it to build traditional automation workflows similar to what you can achieve with Sappure or Make and you can use it to build fully autonomous AI agent workflows using a combination of a broad range of different AI models. And of course, you can build workflows that are a mix of both. And this is exactly where NAN becomes a total game changer. Combining traditional
13:50 deterministic workflows with agendic decisionmaking in certain steps makes NAND an extremely useful tool for internal operations and automation. If you self-host NAN, you get unlimited workflow runs. So the tool itself becomes effectively free. Then the only cost you need to consider is the AI you use. We're using NAN for so many things in my startup already and I have covered a bunch of use cases here on this channel. In particular, I have a video showing how to combine NAND with FeedHive to create social media content completely on autopilot. And I have a video showing how to use Naden with Abbase, another tool we also run to entirely manage your blog and SEO strategy using AI. I'll include links to these two videos in the description
14:36 below. And also, Abase is included in the founder stack offer I mentioned before. For a single onetime purchase, you can watch these two videos, follow the tutorials, and once you're done, you'll have unlimited social media and blog content for your business automatically published every week using AI. Now, let's talk about AI models. Many people think that chat GPT or OpenAI selection of models are the only one that exists. Obviously, that's no longer true. We have Claude, Gemini, Groc, Deep Seek, and a handful of other popular models available. But that's still only the absolute top of the iceberg. There are many, and I mean many, many, more models available that
15:22 you haven't even heard of. Most of them are super niche. They're not general models like GPT4 or Gemini. They're built to solve something very specific. And most of them are open- source and free to use. You only need to pay for the compute. And if you still haven't realized that the current landscape of AI is so much bigger than the tiny handful of the most popular models that keeps popping up on social, you are absolutely missing out. And that's why I want to bring this to your attention cuz it's such an unlock. Let's start by talking about the tool. It's called Replicate. Replicate is a platform that hosts the infrastructure you need to run AI models. They make the whole thing as simple as calling a simple API endpoint
16:09 and it will then run your model and return the results. You pay as you go, so there's no monthly or yearly subscription. You're only charged for the compute that you use. Replicate already has a ton of AI models ready to use and they keep adding more. Chances are if you need a specific problem solved by AI, for instance, upscaling an image, add audio to a video, do classification of a specific kind, Replicate already has a model that can do exactly that. But if not, there is a much bigger collection of open source AI models available, too. It's called Hugging Face. On Hoging Face, you can find almost 2 million different super specialized and entirely open- source AI models that you can download with a
16:55 single click of a button. If something can be solved with AI, there's a model for it on HuggingFace. You can download the model you need, you can fine-tune it, then package it and upload it to replicate where you can run the model through an API. Now, it's easy to integrate this specialized AI into your custom workflows in NAD. These two tools in combination unlock almost unlimited AI power. And most people out there, even in tech, have no idea you can actually do this. I would suggest some alternatives here, but I can't honestly come up with an alternative to NAN. Yeah, make maybe, but honestly, no, they don't really even compare. Naden is just so far ahead. There are a few alternatives to replicate, especially one called Fell AI is getting traction,
17:42 but personally I believe replicate is leading here. And there are no alternatives to Hucking Face that I know of. And I just want to make it clear that I'm not paid anything to talk about these tools. This video is not sponsored by any of the tools I mentioned here. Although FeedHive is one of the tools my team and I run, so there's obviously an affiliation there. Check out foundersstack.pro Pro where you can get access to my entire SAS portfolio for a single one-time purchase. And go check out this video next. Here I will show you how to use FeedHive with NAN to create content on social media. I will see you over there.
