Transcription
So, if you guys do want to optimize your output and make your autonomous agents work all year round and not ever run out of usage, well, this is how. Very simple. We're running Opus, Deep Seek, and Quinn. Quinn's going to be a local free model, which you could run tedious task, repetitive burning token task. Deepseek is running like Opus 4.6. It's as strong as that. So, you can do strong reasoning, codes, etc. But Opus 4.7 does precision tasks. codes, strong, very strong reasoning, decision-m, orchestrating, and a lot more. This genuinely lets my agents work every two minutes for the last 3 months without ever running out of usage. So, I downgraded my usage of Cloud Code membership to the $100 one cuz I could easily do that now. And then I'll also give you guys a 95% token efficiency prompt that is strictly optimized for this triple stack model. And you guys are all set. Let's get right into it.
All right. What's up, guys? So, this is Cloud Code. If you guys don't know how to get here, just download VS Code, which is this right here. Really quick, how to start up, you search up Cloud Code on the extension tab, click on it, you download it, it will tell you to log in, and then the background should look like this. Double tap it. Then you'll see a little icon on top right. That's Cloud Code. Perfect.
Now, if you guys go down to the description, you guys also have odd code repos, which literally this is everything you need to get started up with permanent memory, autonomous agents, etc. I'm giving it all for free. But for this specific video, I want to run down a triple language model that literally 10xed my output cuz I never run out of, you know, tokens, usage, nothing. In fact, the reason why I'm able to work 24/7 autonomously is because of this triple language model. And don't mind the little pop-ups that show up because I constantly have autonomous agents working in the background and automations and such.
So, we're going to click on triple language model just to run it down really quick. This is going to be a Quen uh local LLM which is free. Needs somewhat of a good uh hardware to run this, but there's different models that might run on your specific m machine. But regardless, chatting, greeting, th replies, mechanical text, classification, free, we know that that's going to go over, you know, files, memory, simple, cheap mechanical tasks that are very token, non-friendly token tasks that absolutely fry you. And then we essentially want to treat DeepSeek V4 Pro like Opus 4.6. So not really cheap reasoning. In fact, high reasoning, context analysis, summarization, code, code codebased reasoning, research, synthesis. These two do a lot of the lifting like 70%. And the last 30% is going to be cloud code opus 4.7. This is going to be very critical type work, architecture decisions, anything with codebase and high intense work. So trust me when I tell you how efficient this is and gives you even better outputs than usual, you'll be all set because for example, if you're running clawed opus 4.7, by the third day you're running out of usage and then your output just shoots down because you can't use it for another three days. So, you need to be optimized with your coins, especially with like usage, and I mean credits. But really quick, prereqs, we're going to need NodeJS, Olama, Deepseek, Anthropic API key. Perfect.
So, what are we going to do? We're going to download this really quick. Okay, perfect. We're going to Okay, perfect. What we're going to do is extract all of them in the background. Okay, once you guys extracted it, you guys should see a folder right here. All you have to do is triple language model and then you'll see MD this. You already have a clawed code.
Okay, really quick, you guys are going to see the pulse system, which I'm going to give you guys it for essentially free. Okay, perfect. So, you guys are going to get the pulse system for free, which I'm selling this currently, but I just upgraded to V2. I want to give you guys V1 for free. All you guys got to do is import it. Put I trust author, etc. And then double tap the background. Get into claw code. And then we're going to put initialize. That's it. We'll be back. And really quick, as that's going on, we're going to go to settings. We're going to go to bypass. And then we see allow dangerously skip permission. We click this. Should be fine. Trust me when I tell you how helpful this is. It essentially is one step closer to autonomy just because you'll have different options here. and bypass literally just doesn't ask you for approval anymore. It just does what it needs to execute.
Okay, seems like everything is now downloaded. Now, you see a triple language model, which you guys just downloaded previously. We're going to click on it. You're going to drag in the H. You're going to drag in the triple So, you're going to drag in the triple model router. Actually, just put the whole folder in. To be honest, you could put that in. It's fine. Perfect. We have put in the trip language modeling justest it. And this is also what we're going to do. What we're going to do as that's going on, we are also going to put the mega prompt. We're going to copy and paste this. Copy and then paste. and then be we'll be right back because this will essentially set up everything for cloud code have us download it do everything and yeah as you see it's starting to download stuff check on llama availability and things look really well looks like it's optimized everything okay perfect says implement v2 structure the better engineered one substituting deepsek chat for model specifically 7 V2.
So now that you put in everything, you should literally see a harness optimizer, a loop operator, refractor cleaner. You should have a lot of these, man. Even a chief of staff, which is like the coordinator that delegates the percentage of work into Quen Deepseek. By the way, in order to go to Deepseek platform.deseek.com, and then you're able to buy the amount that you want monthly. Literally, it doesn't autorenew or anything. You just buy it and then you go to API key. Copy and paste the API key into clawed code. And now now I have the autonomy loop which is essentially we're going to copy and paste this drop it down here and then this should automatically work autonomously throughout the day. So after you copy and paste that every two minutes this is going to auto loop. So work on things keep doing it autonomously. So, this chat literally this chat alone will be running like this 24/7. And then, by the way, if you guys do want to learn which ones to optimize and buy, and once you guys input the autonomy loop in, it will literally work 24/7. Like, non-stop. No excuses. Self modifies, constantly works, keep pushing, does not stop at all. I made this pretty vague for you guys so you guys are able to implement it into your current workflow and then it will work actually amazing.
Really quick, we are going to look at where we are now essentially. So, so you guys could download Quinn 314B, which I think this one is eventually the majority of you guys, but everyone with a high tier PC should download 32B. And right away, you could just download this. You copy and paste it. You just put the Quinn, you should be all set if you guys want to go into the SOP route. So, where are we now? Once you guys put in the mega prompt for the triple language model. So, this right here and then the mega prompt setup, you should automatically have these downloaded. Like it would be instantaneous. These are all set up. It just tells you put the Deepseek API. Perfect.
So, really quick, where are we now? After we're copy and pasting the autonomy, all we got to do is map Pulse's brain, pass prompt, map Pulse's brain, create block list, and that's what I recommend. This should start working autonomously throughout the whole process. And now, what do you guys have now? The triple model router. So, you guys already have this set, your project, everything's already set. No hallucinations, all this good stuff. So, you guys should be handy and dandy.
So, what is tier one? This is Quinn. Quinn's going to do flags, template slots, ddupes, hash, labels, things that burn your tokens now is for free. Deepseek is going to be Opus 4.6 reasoning, even great reasoning, long context, summarize, enrich, and a lot more. And then precision tasks, which is essentially Opus 4.7, architectural decisions, self modification, author, voice, codes, etc. So, how does this turn out monthly? To be honest, you're spending $2 a month. Like I when I mean so far I haven't spent $2 yet. It's actually crazy because right now they have like a 75% sale. So I literally don't touch anything. Your opus now you could drop it down to the $100 membership or even $20, but $100 is the lowest one for claude code. So now you make your agents run autonomously. Like, bro, we have infinite amount of work that happens every two minutes and you know, it goes over your pipelines, optimizes stuff, regenerates certain things, fixes your current pipelines. Like, this wakes I wake up with a thousand tasks done, which is absolutely insane.
If you guys do want this, all you guys got to do is go to the description, subscribe if you haven't, and it means the world. Now that you have the triple language set up, you now have 100% output quality, but literally 99% of token usage reduction. So, you're perfectly fine. Run autonomous agents. You should be all set every single week before Cloud resets.
Additionally, what are we going to do? If you guys look down in the description, there's an email template, CRM template, everything. What this is going to do, this whole folder is outreach to local businesses, create them a mockup, and then embed it into the email automatically every single day. It will outreach 500 people if you want it to. And this is how you scale your agency. You have autonomous outreaching. So, let's get right into it. Let's see this.
Okay, really quick. Now, after you guys put the triple language model in, you guys are now going to look at the description. You guys have a cold email outreach that scrapes up local businesses and outreaches them. reaches to them at 9:00 am sharp autonomously. So, make sure you unzip it, put it in the repo, as well as during the outreach. It's going to look like this, right? These are my essentially local businesses in the first outreach that it's going to show. Uh, and this is how it's going to look on their end. So, it's going to say, "Hi, blah, blah, blah." And it shows the services I provide. It shows the free preview of websites. They click on it. It takes them here. This is literally not me at all lifting a finger like at all. If you guys want to learn how to like do this once again, go to my previous video. It has the top 10 best website design repos. And you guys just drag the whole folder, put it in here as well. So you'll have three total. You'll have the triple language model and then you'll have the website repo which I have right here. Website repos. And how did we end up getting such beautiful, elegant, professional designs? Nothing crazy. You could customize them. You could do what you can. If they have anything personalized for them, you could adjust it and then sell it for 500, maybe even 250 bucks, maybe even a,000 bucks. So, this is how you do it.
Now, we're going to ask Claude to create us a mega prompt for having such beautiful, elegant designs for potential clients that it's going to scrape for. Hey, what's up, Claude? I want you to create a mega prompt for you to scrape local local businesses at 9:00 a.m. around 200 local businesses. Make sure they're enriched contacts and MX verified for email outreach. And once you scrape them, I want you to website generate them a professional UI designed website and then implement it into the HTML outreached email for the first one. And then you're also going to have three different follow-ups days after. So third day you're going to do a follow-up. Seventh day, 10th day, etc. But for the first email, we want it as an HTML, beautiful, elegant design, and it's a essential picture that shows our services, a button for them to preview their website. So we're going to send that. It's going to create us a mega prompt, and we're going to copy and paste it. And as you can see, it's creating us everything for it. 200 enriched leads, personalized mockups deployed, etc. And then it will create everything for you. This is the mega prompt.
As it's creating the mega prompt, we're going to go to a website called Netlifi. Netlefi. And this is how you're going to host the websites. So for them to click on the link that's actually linked to the email. You're going to create the website deployment. So create an account. It's literally $9 a month. So once you create an account, I want you guys to go to settings right here. You go to a authentication. You create new access. put no expiration. You do the description for it, generate key, and then you give it to your actual clawed code. What this is going to do is help it auto deploy into Netlefi as a subcategory, as a subdomain. And by this time, I'm assuming you already have a domain. So, you're literally just going to put add project link domain. And that's it.
So, this is finally done. all the scheduling, everything, the follow-up, day 1, day three, day 10, everything's already set. You're going to create a window schedule test. So, what this is going to do, you know, these that are pop up constantly popping up. These are autonomous agents essentially. Got to think about it like one will pop up at 9:00 a.m. to start scraping leads. Another one will pop up at 10:00 a.m. to start outreaching to leads. So, now you have a whole funnel and you will do it through a schedule task action. That's it. Like a window scheduled task. So, you're going to say, "Window schedule task this at 9:00 a.m. Set it up." Perfect.
What you're also going to do is go to Gmail. Gmail password app. It's an app password. Okay. Perfect. Once you guys are in the app password, you guys have to do a two-step verification. Once you guys set it up, you set it up. Okay, perfect. You'll get to this point. I'll say app password. You create your app. Boom. Boom. Boom. Create. And then it'll give you a password. Right. You're going to copy and paste that password. Put it here. What is this going to do? This now gives Cloud Code without you paying for API keys access to your email. So now it can outreach for you. And now tell it, I want you to turn on the task scheduling for email outreach and scraping so we can start rolling this at 9:00 a.m. sharp tomorrow. And then you're going to drip email send instead of mass emailing it all at the same time. So now it digested what this is all going to do. We created a taskuler and now it's going to audit original setup. So now that we set that up, taskuler 9:00 a.m. is going to set up automatically send and scrape email leads etc. So now the pipeline is already set in stoned.
So, as it's doing that now, this is the Flux domain uh mockups. So, you're going to have a subdomain. It will create it. You're going to say create any subdomain for the mockups and link it to those emails, which I already did. So, now as it's doing that, if you guys have any issues with Nellifi, so when you have any issues with Nellifi, you're all you're going to do is ask Claude, hey, make sure it's a subdomain so you could deploy these mockup generations into the email, right? And then it will create a subdomain which any URL. This is my URL domain. And so when the clients click on it, they'll see this in the URL, but they'll actually see their domain that's linked to the email. So it got done. It says at 9:00 a.m. it's going to scrape 200 leads. Enrich them MX verified, which is making making sure email works essentially. CRM sync email 234 followup and then follow up slowly but surely.
If you guys want to know how this is going to help, you're essentially getting not just cold emails, but you're getting outreach core, hard and fetch, mockups, soft failure, CRM, you're getting case studies, even you're getting uh emails, replies, uh a reply like ping that tells you, hey, someone replied. if you want to build like a nice beautiful CRM like this. So essentially it's everything beautiful. Has outreach. It has outlier. It has everything sync possibly you could possibly think of. Like bro, this is actually really cool. You're essentially going to tell your AI agent, hey, can you optimize the current CRM I have? Add in features that I never even thought of and things that you haven't thought of that you know is going to benefit me into the current Flux Social CRM and add what is necessary. maybe add things that each potential lead could have and benefit with my service that is prioritized for them personally and that I could maybe pitch to them right next to their name. So, you could literally tell it this or if you want like a crazy CRM, you just tell it, "Hey, create me a mega for a CRM for my agency that provides outreaching, AI receptionist, AI systems, websites, infrastructure building, etc. and everything I possibly need to keep up with my leads. call them, outreach, tell me if they replied or not, recent leads that were scraped, potential leads and messages that they maybe sent through email to me and replied. And that's literally and it will create you a mega prompt as it did what with the email outreach and then you should be fine from there. That's literally what I told it. But for the social media realm, I was like, "Hey, make sure the social like a social media CRM is set and situated and you're able to optimize it and updates every two hours and it tells me which hashtags are best and relevant and my outreaching CRM and that's it. It did everything. It fixed everything. It set everything." And that's how you got to do it. You just got to put in the reps, guys. It tells me which people bought. It tell it tells me competitors and outliers of the current marketplace of you know AI and stuff like that. So if you guys do want something like this, make sure you guys follow me on the IG because I end up giving you guys a bunch of free game. Like I'm talking about look at these repos I have. I have all these repos and it's crazy. It truly is guys and every single day I give out around like 10 of these repos for free. So make sure you guys follow my IG and let's see how this turns out.
Okay, seems like it's set. It said that I built an Intel page, etc. Like I already have a CRM, but now you guys have the infrastructure, by the way, to literally just tell your cloud, "Hey, build build me a CRM for my current pipeline of outreaching and replying and having people message me and potential leads and boom, you'll have a CRM exactly like this. You guys have the infrastructure now. You guys have the UI designs. You guys have everything. So you should be fine. But let's see what else it built because it told me it did a lead intel. Look at that. So essentially each potential client could make me this much. And by the way, if you want this exact CRM, literally screenshot the YouTube video, send it send it to your cloud code here and tell it, "Hey, build me the sim a similar CRM." And it'll build you a similar CRM. So literally, you have the same exact infrastructure for copying repos to cloning UIs to everything. So let's see this. Look, $12,000 I can make at a dental clinic a year. That would be crazy, guys. And it tells me what I should do, what I should sell. Look at that. Beautiful. Well, if you guys want something like this, once again, follow my gram. I literally keep you guys in the loop every single day, all day.
So, this is like really cool. So, if you guys did enjoy that, drop a like. Please subscribe. I gave you guys my pulse system, which I use currently, but I made a V2 version, which you guys can get on the site, but this one's totally free. I was just selling it a week ago. And then additionally the triple stack model and 95% token efficiency model which you guys are all set. This is strictly optimized to the triple stack language model setup. So enjoy it guys. Let me know down below what else do you guys want because I'm always giving out free game and so far I've given out a lot. So make sure you guys follow my IG. I'm always giving out free softs there as well as YouTube right here. So everything you need is down in the description below. And please, please, please tell your friends because everybody in my niche of Claude code and that follow me on Instagram, I genuinely made them from yesterday not know what the hell Cloud Code is to becoming experts and monetizing themselves. How did I do it? It's because every day I'm posting three to four things and it's genuinely very useful. In fact, everything I post, I give it to you guys for free every single video. So, with that being said, follow me on IG and let me know what you guys want tomorrow or after tomorrow cuz I genuinely want to pump out a bunch load of content on Cloud Code. I cannot wait to help you guys as I do on Instagram.