The Easiest Software Business to Start in 2026

If you've ever wanted to start a software business and if you thought vibe coding made it easy, it just got a whole lot easier. Because for the last 20 years, if you wanted to build real wealth in tech, there was one thing that everyone chased, and that was SAS or software as a service. You build something once and people pay you every single month to use it. It's recurring revenue and it scales infinitely if you do it right. And these companies sell for 10, 20x, sometimes 50x their annual revenue. We're talking hundreds of millions of dollars in most cases. But as you may know, starting one used to be brutally hard. You had to have developers, designers, months of work, tens of thousands of dollars, and it was just completely out of reach for most people. But that's just changed. And honestly, it might be one of the most exciting opportunities I've ever seen for beginners wanting to get into technology or AI and software, [music] which by the way is by far the safest career to be betting on at this point. Now, if you are new to the channel and don't know who I am, my name is Lee Molley and I've been building AI systems for businesses through my agency, Morning Signing Eye, for over 3 years now. We work with Fortune 500 companies, NBA teams, public trade companies. So, I know a thing or two about this technology and more importantly how you can actually make money with it. And what I'm seeing right now is this open claw and claudebot explosion is [music] genuinely significant. So, let me break down what this opportunity actually is, the five types of businesses you can build around it, and how you can realistically start one of those today, even if you're not technical. But this isn't all sunshine and rainbows. There are a few gotchas that you need to be aware of. So, stick around to the end because I'm going to be breaking down who this opportunity is actually for and who I think should really avoid it. So, a quick refresher on why SAS itself has been such a big deal for so long and why this shift that we're seeing with Claudebot and the opportunity it presents matters so much. Most businesses trade their time for money, you do some work, you get paid, you stop working, and you stop getting paid. But SAS is different in that you build something once, a tool or an app or service, and then people pay you every single month to keep using it. It's recurring revenue, money that shows up whether you're working or not. That's why every tech person for the last two decades has been trying to build a SAS. It's like the ultimate prize that everyone wants to do and sell. But here's the problem. Starting a SAS used to require building a lot of stuff out. Think of it like a pie chart. You had to build the actual functionality, the thing that does something useful in the app. That's maybe 20 or 30% of the work. [music] Then you have the interface, which is all the screens and the buttons and the dashboards and the settings, everything that a user actually sees and clicks on. That's another 30 to 40%. Then you've got the infrastructure like the databases, hosting, login system, security, billing, all the invisible stuff that actually makes the app work at scale. And then after all of that, you had to go and market it. You had to get people to find it and pay for it. So if you draw a circle with all of that work, you'd see why most people never even tried it or I've definitely done my fair share of discouraging people from doing this, especially as a beginner, cuz it's just like it's seriously not feasible for the vast majority of people. But because it's such a sexy and lucrative model, people are still drawn to it despite these enormous challenges that it presents. To create your own SAS, you needed 50 to $200,000 worth of work in 6 to 12 months just to get something in front of your first customers. But the actual valuable part of the software, the piece of functionality that does the thing was only a small slice of that and everything else around it was really just a wrapper to get it into people's hands and allow them to use it. So the reason I'm making this video now is because a few weeks ago something called OpenClaw or Clawbot basically exploded onto the scene. But if you haven't heard about it, OpenClaw is like a personal AI assistant that lives on your computer, but unlike chatbt where you go to a website and ask questions, you can connect it to something like WhatsApp or Telegram. So you've got it on the go and this thing will actually do stuff for you. It can [music] be set up to clear your inbox, book flights for you, manage your calendar, research companies, write code, and build apps for you, control your smart home, and the key thing is that it kind of learns about who you are over time. So, it's like having your own personal Jarvis or assistant that never forgets anything and can work 24/7 for you. It's been absolutely blowing up with 70,000 stars on GitHub in about a month. And for reference, that like never happens. Even the former AI director at Tesla is calling it the most incredible sci-fi thing I've ever seen. Now, I've been in the space for over 3 years now, and I've seen a lot of these like, oh, this is going to be huge moments come and go. Every like 3 to 6 months, there's this new surge, this new thing that gets everyone's anxiety spiking. But this one does feel a little bit different. I made a video recently talking about the iPhone parallel and why I think this is the iPhone moment for AI, and I'll I'll link it in the description if you guys want to check it out. But the short version of this is that OpenClaw isn't really the product. Open Claw is the platform. It's the iPhone. And remember how iPhone wasn't just a phone. It was actually a platform that let everyone build apps on top and sell them through the app store. And that app store platform created a $ 1.1 trillion economy that made more millionaires in the California Dollar Rush. And right now, Open Claw is doing the same thing. It's created a marketplace called Clawhub where anyone could build and share skills. You can think of skills like a little instruction file that teaches the AI assistant how to do something new. Out of the box, a Claudebot knows how to have conversations and reply to you, but it doesn't know how to analyze legal contracts or pull transcripts from YouTube properly or research companies using a bunch of different data sources. But a skill can teach it how to do that very, very quickly. If you want to teach it to review contracts for you, you can create a skill for that or find a skill for that. If you want highquality YouTube transcriptions on demand, you can create or find a skill for that. Same for advanced company research, you can do a skill for that. And here's where it gets interesting for building businesses because some skills are just plain text instructions. It's literally just a text file explaining what the AI should do. But other kinds of skills can connect your AI assistant, your Claudebot, to external services like your own software or your data or your tools. And that's where the money is here. So here's where it gets crazy because remember that full circle of everything you needed to do to build a traditional SAS. You had the interface, infrastructure, marketing, all of it. In the new model, the skillsbased model, that circle shrinks to a tiny dot. You just need to build the functionality, like the actual valuable part of the app. You put up a simple onepage landing page. You connect Stripe for payments and you write a skill file and you publish it to clawhub. That is it. In this case, the AI assistant is the interface or Telegram or WhatsApp that they're using it through is the interface to interact with that functionality because right now with the vast majority of SAS products, you go onto their website, you log in and you click buttons on their interface. What we're seeing here is that we're able to scrap all of that building out of the interface and all that complex front-end work and using Claudebot as a conversational way to interact with all your favorite apps. And not only do you not need to build a fancy front end, the marketplace and claude hub can actually handle your distribution for you by publishing it on the marketplace, just like on the app store, they can do all of your marketing. All of the eyewalls are already there looking for valuable things to input into their Claudebot. You can just get picked up and benefit from all of the traffic that's already going to those marketplaces. And finally, you don't need to build [clears throat] complex login systems because it's just an API key. All they need to do is sign up for your app and then take the API key and plug it into the Claudebot. That's all the account creation you need. And the best part about this is that AI can literally help you come up with all of this. You can talk to a claude bottle chat GPT to come up with the idea, plan the architecture of the app, write the code, build the landing page, and it can even help you from a marketing standpoint to help you better describe and position your SAS and skill as something that's useful to a certain audience. We're at the point now where one person, even a nontechnical person in a weekend [music] with AI as a co-pilot can build something that would have taken a team of people five or 6 months in terms of actually creating something and getting it to market. And that's what I mean by the barrier to entry to SAS now has collapsed. So let me show you the five types of businesses that you can actually build based around the skills as a SAS model. Category number one are pure prompt skills which is essentially expertise in a file. This is the simplest possible version you could do. This is literally just a text file with instructions. You're packaging your expertise or your knowledge about how to do something well into a format that the AI can read and use as it needs it. For example, if you know how to review contracts and spot risky clauses. You write out detailed instructions what to look for, what questions to ask, what red flags matter, and that becomes a skill. Now, when you release that, anyone's AI consistent can use your skill to review contracts using your methodology. Other examples include a writing framework, a research process and analysis methodology. The technical setup for this is dead simple. It's just creating a text file with instructions in plain English explaining how the AI should approach a task, publish it to Clawhub, and then you're done. As of right now, Clawhub isn't really set up around having both free and [music] paid skills on the platform, but I assume that's going to change in future. But it wouldn't be difficult to put a link to a landing page where they could pay for your skill and check out via Stripe and then be sent the skill file to download and plug into their claw bot. Now, to be honest here, the moat here is basically zero because anyone who downloads your skill can read your instructions like word for word and then just copy them. So, you can think of it as selling like a prompt template. It's kind of like an evolution of of people selling prompts, which people have done for the past 3 years. These are for sure useful but not really defensible at this point. And you might be able to sell them for 10 to 50 bucks. But I honestly think these skills could be a very powerful way for certain people or business owners to be able to distribute free information or kind of lead magnets through the form of a a skill that can be plugged into an assistant. Uh which is something I'm looking at exploring soon. So this kind of textbased skill is where everyone can start. It's like the training wheels version to get into this. Category 2 is all about utility skills. Here we're going one level up to a skill that can wrap a script which is just a little bit of code that actually does something. For example, you might make a YouTube transcript attractor that actually pulls down a downloaded version of the video, an MP3, and then runs it through Open AI whisper to get a really high quality transcript, not just some kind of crappy YouTube caption one. And this is really an essential part of a lot of creative workflow right now is around content creating because you need to grab videos and get the content and modify it and tweak it with AI to turn it into something of yours. The point of this is to provide an essential utility to the millions of people who need access to these transcripts in an easy way and they can plug this into their Claudebot so it's able to get those high-quality transcripts very quickly. Other examples of these kinds of utilities would be web scrapers or PDF passes or data converters. But here's what makes this a business is that these things break constantly. YouTube's always changing something. Websites update. You're responsible as a creator to keep that working all the time. That's what they are paying for and that is the product. So you're not selling just the code or script. you're selling it works and when it breaks someone is going to fix this. So you can charge $5 to $15 a month and people will pay for that reliably to keep using that service rather than having to build and maintain their own version. Category number three is API integration skills or bridges to services. So these are skills that teach the AI how to talk to tools that people already use. For example, HubSpot, one of the most popular CRM. You can create a skill that integrates into that and teaches the AI or Cordbot how to create new leads or pull reports from someone's CRM. Now they can just talk to their assistant and say add this person as a lead and it knows exactly how to talk to HubSpot and can put that lead in there in a second. The value here is in the integration logic, not let's just call this tool. But here's how to use this tool intelligently and correctly. So you can charge $20 to $100 one time for this or maybe try to get it on a recurring model instead. Category four is backend service skills or skills as a SAS as I'm calling it. This is where the real recurring revenue sits and this is what I mean by the easiest software business in the title of this video. So here's [music] the concept. You run a service on a server. Your skill teaches the AI how to connect to your service that does something and users pay for monthly access to it. I can make this a bit more concrete with an example. So imagine you had a lead intelligence service and this is a very common thing that businesses want. We've done them at Morningside before, but now anyone using this can build a much simpler version. The user tells the AI research XYZ company for me. The AI then reads your skill, sends a request to your server with the company name, and then your server or the app that you've written might cost $10 to $20 a month to run. It goes out and scrapes LinkedIn, Crunchb, [music] the website, recent news, and then pulls it all together, synthesizes a clean briefing, and then sends it back to Claudebot. So, users see this beautiful company research report. What they don't see is your server doing all the heavy lifting for them. Here's a technical setup for that broken [snorts] down simply. So, your server [music] is a small program that receives requests and sends back response, maybe 100 to 200 lines of code, and AI can help you do all the writing for it. You can deploy that on a cheap hosting, maybe $10 to $20 a month. You have one external website page or landing page that people go to. On that page, you have the pricing of it, some information about what it does. They can sign up and pay through Strike. And once you give them an API key, they can plug it into their Clawbot so that it has permission to use their account. So, it's connected to the plan that they've just paid for. Then, all you need to do is write a skill file that's maybe 30 to 40 lines of text telling the AI how and when to use your service. What makes the Mo so good on this is that your skills instructions are public and that anyone can read those, but they're useless without your server running behind them. So unless they've signed up and paid, it's like them having a phone number, but there's no one answering it. The revenue for these could be $9 to $50 a month per user. And if you can get 100 users at 19 bucks, that's $2,000 a month. It's going to cost you 20 to 50 bucks a month to host. With this, you get real SAS economics without the traditional SAS complexity. So I mentioned hosting your OpenClaw, but where can you actually host this thing? You could install OpenClaw directly onto your computer. That is kind of risky. You'd be giving it access to your files, your passwords, and everything on your machine. And if OpenClaw ever got compromised, so would you. You might have seen guys like myself buying a Mac Mini, but that's kind of overkill. It cost $600. And there is actually a lot more like lightweight and more cost-effective way of doing it. Instead, you can get great security and performance from like seven bucks a month by having your own private VPS, which stands for virtual private server. You can think of it as like your own computer that's completely separate from your stuff and it's hosted in the cloud and it's always on and ready for you to use. Hosting as VPS is a good option because it's easy to set up, it's fast, and it's super reliable. I haven't had any issues with mine. So, for those of you who are a bit more on the technical end of that spectrum and maybe do fit into that category of yes, I want to tinker with this and I want to get into it. I'm going to walk you through a quick setup here to show you how it's done. Cuz the guys at Hostinger are really smart. They've got a whole page dedicated for setting up your open claw. I've linked it just below this video with oneclick deployment. Super easy. The KVM2 plan is really the sweet spot with the best price toerformance ratio. They've got plenty of resources to run multiple bots or automations and you can scale it up later if needed. Now that it's added to your cart, you'll choose your billing period, either 1, 12, or 24 months. I'm going to go with 12 months. We have the option to go with this ready to use AI, which means it'll deploy instantly using Nexos to sync up with LLM providers like OpenAI and Anthropic. We're going to be using our own API key. So, we're going to untoggle this. If you guys want to save 10%, you can use my code Liam Otley in all caps. Then you can hit continue. If you don't already have an account, you'll create one. You can enter your billing information and payment method and then submit. After purchase, you'll see the OpenClaw configuration page and this is an important part. I'm copying my gateway token since I'll need it to log into OpenClaw in a second. Then just add what API keys for the AI you want to use with OpenClaw. I'm going to paste in my Open AI API key. I'll also add my WhatsApp number. And then finally, we will hit deploy. I'm speeding this section up for you, but hosting it takes care of everything in under 2 minutes for you, which is like by far one of the easiest processes for getting this set up that I've seen. When the deployment finishes, you'll see your OpenClaw project running in the Docker manager. Make sure to grab your gateway token now because when I click on this port number where I'll paste in the gateway token to login. And now I'm in. [music] So, I can send OpenClaw a message and it'll reply instantly to get itself up and running with you. And now you have your own private AI assistant running in the cloud. You can start playing around with it, building some skills like I was talking about in this video. And before you leave this running 24/7, I cannot stress enough that you need to take the security aspect of it seriously. You're deploying to a public server, so make sure you harden SSH, disable [music] password-based root access, and treat your API keys like production credentials. A VPS like this isolates Open Claw from your personal machine, but security is still your responsibility at the end of the day. I've shared additional security guidance inside my school community. So, feel free to join for more info on this or study other OpenCore security guides. By the way, hosting also has an AI assistant called Kodi that can actually help you with your task and get this set up. Now that you technical people have an idea of where to host your openclaw, let's look at the next way you can potentially monetize it. Category number five is what I call the proprietary data skills or the mob builders. This is by far the most defensible category and it's skills that are backed by data that other people don't have. And this is where I think some of the biggest businesses will be built long term. So, here's the concept in simple terms. You might have some kind of valuable information, whether it's industry research or market data or just expert knowledge from years in a field. You put it into what's basically a big organized filing cabinet. These are technically called a vector database, but just think of it like a barrel where you've stored all of your knowledge in like searchable chunks. When someone asks a question, the system will find the most relevant pieces out of all of that knowledge, hand them to the AI, and the AI will write a helpful answer to them. And the key here is that the user gets the insights they're looking for and they get the answers, but they never see the raw data that you've got in that vector database. That's because the AI pulls out chunks, summarizes it, and doesn't just copy and paste stuff. So, for example, say you wanted to create a market intelligence skill. You collect pricing data for thousands of software tools over time, and the user asks, "What's your average pricing for project management tools for small teams?" Your tool will know how to turn that request into a database query and then return the answer that they're looking for not just in a bunch of raw numbers, but it will be able to turn that data into the nice, clean answer that they're looking for. So, they're paying you because you did the brutal work of collecting that info. The human consultant might charge you $300 an hour, but your skill could charge $29 a month. And even if someone reads your skill, it's useless without your data and paying for the service. So, the AI can actually interact with that vector database and get the information it needs. In this case, the data is the product and you can charge anywhere from $19 to like $200 a month for this depending on how valuable the data is. So, everything I just showed you is real. And yes, I do believe that an opportunity exists here. The economics make sense. Barrier to entry has genuinely collapsed. But I need to be honest with you guys, and this is coming from someone who's watched these kinds of hype surges happen every 3 to 6 months, 4 years now. This stuff is new, and importantly, it is unproven. And this what I I kind of call building on sand. If you're looking to get into AI, start a career, start making money, then these new frontiers are sand. They're not proven, and there's no clear paths that have been walked by people before you to make this the kind of thing that you can bet your future on and go all in on. Open Claw and Claudebot is just the first version of this. You're going to have OpenAI, Google, and Anthropic. They've seen this stuff blowing up and they are absolutely going to release their own competing platforms. So, what works on Claw Hub today might need to be completely rebuilt for three different platforms in 6 months. And the people that are winning in the space right now are the absolutely cracked AI engineers who live on X and they're shipping code at 2 a.m. and have been deep in this AI coding ecosystem for years. So if that's not you and you can't ship code at lightning speed and adapt overnight when things change, then you're probably going to get eaten for lunch by these guys. And by the time you figure out how to build one skill, they've shipped 20. And here's the biggest issue with the space right now. Businesses can't use this stuff yet. And they're the ones who have all the money to spend because these things are a security nightmare. The compliance isn't there. And any serious company looking at openclaw is thinking that's cool but like give me a call in 2 years when this stuff is actually ready and safe to be used in my company. At Morningside, we're keeping an eye on all of this stuff happening. We're not ignoring it, but we're not abandoning what's worked for us for 3 years now and for tens of thousands of other people who followed the model. We're not willing to abandon all that just to chase this shiny new object on hand. For the past 3 years, I've been building AI businesses in a very different way to the skills-based one and taught tens of thousands of people to do the same. And these are like regular people with all backgrounds and often no technical experience at all. And this new business isn't as hot and flashy as skills is [music] maybe not as exciting and honestly can be kind of boring sometimes. But it is proven and it works like clockwork and has worked for tens of thousands of people. And the reason I consider this building a [music] business on a rock solid foundation is that we don't have a technology problem in AI at this point. We have a human problem. This technology is already freaking fantastic. But the reason AI adoption has a 95% failure rate in companies according to various studies, it's not because the tech doesn't work. It's because there aren't enough skilled people to figure out what a business actually needs. [music] Identifying the lowest hanging fruits with the highest chance of a positive ROI to be able to build and deliver those solutions and critically roll it out to the team and make it stick with like actual training for the teams, adoption, change management. That's where everything starts to fall apart and that's why AI adoption is taking a lot longer [music] than people expected. organizations are slow and the vast majority of people adopt new technology slowly and that problem and those needs are not changing anytime soon. So we have a much longer horizon to actually work with those clients and create value. We've been working with our biggest client at Morningside AI for over 2 years now. So the question is do you want to bet on a problem that's highly likely to shift under your feet every few months a flashy exciting and totally unproven thing or do you want to bet on a slowmoving problem that's going to be here for the next decade because it's fundamentally about how humans adopt the new technology. In my case, I'd much rather work with the devil I know than the devil I don't. And so, the proven path of AI auditing and consulting to help businesses figure out what AI tools and systems they can use. Literally, anyone can learn enough to get started with that in a few days and become genuinely good at it in a few weeks. Then you've got no code AI automation, Vive coding, [music] building AI solutions to small businesses. And then the human part, the training, the adoption, and making that stuff stick in the company. That's what my community has been doing here for years now. And this actually works. Like really works. even my free school community are in there landing their first clients very quickly using the free course that I've put in there and the people in my accelerator program are signing bigger and bigger deals every month. So I am very happy to say that this model has completely been proven at this point as [music] an entry point for beginners into starting an AI and technology based business. And here's the thing, the skills and experience you build doing this kind of AI transformation and agency work, they're going to be valuable for the next decade regardless of which platforms come and go. how to audit a company, how to actually build and deliver solutions, how to train the teams on those. That stuff's not changing. There's tens of millions of businesses who need that and they do not have people helping them right now. And eventually, when say Claudebot for businesses actually becomes a thing, when the security and compliance systems are there, guess who's going to roll that out to them? Me, or maybe you as well because you've already got the clients, you've got the relationships, you just add a clawbot setup and training offer to your services line and keep moving. You can charge clients per head to analyze the person role in the company, figure out what things you need to set up their clawbot and then train them on how to use it. You can be charging thousands of dollars per person for those kinds of setups. So that's why I say starting an agency and getting into AI business that way is building on the rock. But if you start with nothing and are chasing the shiny new thing, [music] that's how people burn out and eventually give up on AI entirely and go back to their old life. Now, luckily for you guys, I'm such a nice guy. So I put together a full free course that walks through exactly how to start an AI business this way that I've seen work for tens of thousands of people. The link is going to be the first link in the description down there. So, long story short, this skills opportunity is real. If you're extremely technical, if you've got some proprietary data or deep expertise, if you can move at the speed this ecosystem demands, then by all means, have a crack at this. Like, totally go for it. And I do hope it goes well for you. I think this stuff is awesome. But for most people, if I'm being real with you guys, the boring, proven, and slower moving opportunity here is the much smarter bet. Building your house on the rock rather than on the sand is my whole message here. And if you are interested in this kind of stuff and want to see me actually try to build one of these skills as a SAS businesses from scratch with like a landing page, a backend, setting up Stripe, the whole thing, drop a comment down below and let me know. Just a bit of a fun video of me going through the learning process right in front of you guys and you can see how the experiment turns out. [music] And generally, I'd love to hear your guys thoughts on this skills as a SAS concept and these different types of businesses I pointed out. I'd love to get your input and how [clears throat] you're feeling about this because I know when these big things pop off, there's these like anxiety spikes we all get, but we are all in this together. And if you want my honest take on specifically what this open core stuff means for people already running an AI business or agency, I made a separate video linking it down up here, [music] which I'll link right there for you guys. At the end of the day, this market is big enough for everyone and it's only growing every single day. That's all for the video, guys. Thank you so much for watching. up mly and I'll see you in the next