How to Use OpenClaw Better Than 99% of People

Key Takeaways

Open Claw, also known as Clawbot, is one of the most powerful AI tools ever released. But 99% of people are using it without structure, without control, and without visibility into what it's actually doing. So, in this video, I'll show you exactly how to use Open Claw better than 99% of people. I will also cover six different key things that you can implement right now to turn it into a real AI project manager with task tracking, clear authority boundaries, long-term memory, and the right models for the right jobs. So, this is going to be without complex theory, without guesswork, just practical systems that you can set up once to make open claw run faster, cheaper, and smoother for you and actually work for you in the background. So let's get started. Open Claw can do a lot of things for you, but when you have too many tasks, you cannot track each task and what is happening behind the scenes. Essentially, this is looking like this. So, you're analyzing the data, you're monitoring thumbnails, you're brainstorming ideas, but you don't have a dashboard where you can actually see all these u tasks that are in progress. So, it's really hard to track each tasks. you can't see any progress and you're pretty much living in an unorganized chaos. So essentially what you want is you need a custom dashboard that will track all of your tasks in one place. What's in queue? Uh what's in progress? What's finished? This makes working with open claw that much easier and you're much more productive in return. So essentially what we have here is I have here an entire uh clawbot Trello board or canban board and I'm tracking all of my task tasks right here. So as you can see I have done today. This is essentially done today and what I did uh is I created this in lovable. So I created a canban board in Lovable. I connected it to open claw and open claw is managing this for me. So as you can see if I go to open claw I went in and I said create a tasks desker uh dexter task with external ID set to fresh. So essentially I set this up entirely on my own. I will leave a link down below for the prompting guide for lovable and prompting guide in uh clawbot or openclaw so that you can set this up for yourself. But as you can see, it went in there and it added my own uh task. And that task is right here or in my app right here. And I can obviously go I can, you know, uh delete this task if I want or not. But this is essentially one of the really cool ways how you can use open claw. Implementing this will literally 10x your productivity. The second thing you can do to make open claw even smarter is to give it authority boundaries. So what does that mean? So without authority boundaries, open claw usually asks too many questions or makes risky decisions. Now some of these decisions can look like this. This is an example that we saw just a couple of days ago that one guy actually got charged $3,000 for a course that he didn't sign up for. So what happened here? So what happened is that uh OpenClaw didn't have the boundaries in place. That being said, open claw will operate within defined authority. So you need to define the rules for open claw because you are the boss. So in this case I formatted these rules in two different brackets. You have no approval required and you have hard stop ask human. So in the first bracket as you can see it can decide autonomously. So on some tasks it will be able to decide autonomously while on other tasks it will need to ask you. So in this case you want to make sure that you're specifying this in the prompt. So you can actually just go and create prompt based on this particular uh structure and you know you can go into chat GPT even and chat GPT can create a prompt for your cloudbot or open claw and you can paste it in there and this way you can get structure so you can define this once right you may decide autonomously this this and that and then requiring human approval this this and that. So, that's essentially how you're going to be in a much safer position and you won't really be like this guy who is charging $3,000 for a course. Obviously, you're not going to give it a your credit card at the end of the day, but this is another really smart way how you can uh adjust OpenClaw to work in your favor. Now, let's talk about memory. Open Claw often has memory issues because it limits the memory you use to save tokens and run more affordably. So what you need to do is you need to create a universal prompt. But before creating a universal prompt that I'm going to show you uh very soon, you need to understand how memory works in openclaw. So you have a stateless workflow. This is just when you join in open claw. You have current tasks and then you have temporary context. However, this context is temporary. So it forgets past chats. It forgets goals and rules and there is no persistent memory which is a big downside. So what you want to do is you want to create your own document where whenever you have a current tasks you have the memory file here and it reads that me memory file. In this case state MD is a memory file. I will show you exactly how you can set this up and then it acts accordingly to your memory and it has goals and rules that it follows. So essentially how this looks like is you have it right here. So, this is a long-term memory setup for Open Claw. I will leave all of the documentation in the description so you have it. But as you can see, what you want to do first is you want to create a state MD. State MD simply put is a long-term memory file. So, it's a type of a file. You don't have to know why it's called state MD. You simply want to create a file in your open claw workspace uh called like that. And what you can do is you can copy this and you can paste this below in uh your open claw. Now, as you can see, this um state MD uh consists of identity. So, for example, you're my long-term AI operator, and not a chat assistant, current goals, which you can add right here. I didn't add this in the prompt, just so you can add it yourself for your own uh specific use case, active projects, constraints, which you can modify. So, make sure to modify this. Once you put this in, this is kind of like a prompt and that prompt will create the uh state MD folder or file. From there, what you want to do is you want to solidify that state MD folder. So, what you want to do is you want to copy this exact prompt. You have no long-term memory unless you create it. From now on, persistent memory works like this. Before starting any tasks, you must read state MD. So what this prompt will do is it will just say from now on open claw just go and use the state MD as a reference for the memory and this is essentially fixing this whole mess right here and you will get a much better memory long-term memory whenever you're using your open claw. Now another thing that I see is that most people host open claw on Mac mini or locally and while that makes sense for shortterm if you want to run it 24/7 and access it anywhere in the world plus if you want to have your security nailed down you need a VPS. So what I'm doing is I'm hosting openclaw on hostinger and I have 247 uptime. I have hostinger secure firewall plus it's beginner friendly. Now why do I always go and why do I recommend this? I recommend this because it's easy to set up. It's secure because you don't know what files can OpenClaw actually access on your computer and again it's just more efficient because you're not going to be hosting it on your computer. It's not actually going out there and using up your resources. So if you're using open claw, you know that openclaw brain, essentially the LLM that is running the open claw is the LLM that you set up at the start of your open claw journey. You set up OpenAI or Enthropic or Google Gemini. Whatever LLM you set up or whatever AI model you set up, that's the open claw brain. However, that specific model isn't always the best for specific use cases. So in this case I have three different specific use cases. So for web search you can use Gemini. For coding you can use codeex or quen for social search you can use grock. So all of these have specific use cases. Now one thing you can also do is you can ask open claw what is the best model for a specific use case and it will give you the model and then you can use that model for that specific use case. So this is a much smarter way to use open clause. So if you want OpenClaw to use certain AI models for certain actions, you have two options. You can ask OpenClaw what is the best model for a specific action and implement it that way. Or you can simply tell OpenCloud to set up a specific AI model for specific actions where you will actually need an API key to add it to Open Claw. In return, this will give you much better results in shorter time and with less money being spent on tokens, which is a huge bonus. So, here's what most people don't realize. When you first join Open Claw, you don't have any history with it. So, what you want to do is you want to create a system prompt. And this will be set of rules, set of priorities, decisions, and executions that Open Claw will be capable of. Because if you don't set this, open claw will again not have any authority and it will just go and do a bunch of different stuff. If you can create a system prone where you have specific use cases, goals, expectations, this is huge because open call will know exactly what you need and it will serve you literally like a butler every single day. You're lucky here because I've created a system prompt right here which I pasted. You can adjust this again in chat GPT or in any other LLM by yourself and you can paste this in open claw. Now, as you can see, it's quite a lengthy one, but that's exactly what you want to do because once you set this up first time, you don't have to set this up any more in the future because Open Claw will exactly know the rules, the priorities, decisions, and what it actually can do for you. So these are six points on how to use open claw better than 99% of other people because in the end open claw will be as good as the leader which is you in this case. So you want to make sure you set the rules, you set boundaries and you set up everything how I showed you in this video. If you like this video, leave a like down below. Comment down below what is your favorite point from this video and subscribe for more videos just like