# Free Docusign, Bloomberg & 10 more

Channel: The Next New Thing
Video: https://www.youtube.com/watch?v=zw_KDA-NynE
Duration: 26 min
Language: English
Words: 5192
Transcript page: https://viewrankai.com/tools/youtube-transcript/zw_KDA-NynE

---

[0:00] I'm going to show you a tool that lets you copy DocuSign, a $9.36 billion company. You can copy it for free. I've got a tool that will let you analyze stock like not like a research analyst, but like a team of analysts. And I've got a tool that will help you create tasks and then use OpenAI or other agents to complete those tasks and so much more coming up with the top 10,

[0:23] actually 12 GitHub repos of the week. Let's get into it. Presented by Zapier, the AI automation company. All right, Adam. This week we're going to go just straight down the hot GitHub repos of the week based on popularity for this week. We're going to start with the very first one, which is trading agents. I've told you before how much I

[0:40] hate all these trading agents. Andrew. I hate them so much because I'm not trading using AI, but I get it. This is uh this is a tool that helps people trade. I've got a really good video on it on YouTube that will of course link everyone through. This guy used it. I love that he walked everyone through step-by-step of using

[1:00] it. And he he showed how it analyzed Nvidia. It told him that from lots of different directions, lots of different uh basically user profiles. And it ended up with a target of 268. And then he looked up how much Nvidia was at the time that he did it. Um and it was 199. And then he also walked through and not just how to install it, but also its reasoning and and how it

[1:22] was thinking. But should we pause this and go buy some Nvidia stock real quick? No. Okay. [laughter] All right. This is basically a trend that we're seeing, right? AI is coming for finance. People are thinking that they could use it to make better investment decisions. Smart, useful. What? Yeah, I think I think it's interesting. You know, AI is very good at creating models. We do that at our studio all the time for our businesses. You know, a question I'd have about this is like is it back tested? Like do I trust their model? You know, in in the it's not the AI that's trading, it's the AI that's

[1:55] creating a trading model. And is the model any good? So, is it backtested? I don't know. What's the Do they have performance numbers of if you would run this from the year 2000 to 2010, what would we have made? The word shows up. Backtesting date fidelity, so maybe Uh so I I think what I see there's probably going to be a lot of like hucksters out there. This one might be legit, it might not be, but if I'm going to use any of these models, I'm going to be testing them long before I'm putting

[2:22] real money behind them. The big value I see in this is you've got research that you can use whether or not you decide to to invest in this. Okay, let's go on to the next one. Uh Rufflo, your age your AI agents finally remember your projects across sessions. We've talked about this uh before. This is an ongoing issue.

[2:44] What do you what Oh. What do you think of Rufflo? Rufflo is a a kit of agents similar to G stack or anything else. There's agents for design, for coding, for review, and then the the top of it that's the most important is this word that keeps showing up here, orchestration. So, Rufflo is an orchestrator that says, "Hey, I know how to use all of these 90 agents." You know, think if if you're just talking to Claude and you say, "Claude, build me this." You're talking

[3:12] to one person with one set of expertise. And now you take G stack, for example, say, "Oh, now I've got all these skills in there." So, Claude can orchestrate all these, you know, 80 skills that Gary Tan is giving me. Now, Rufflo is like, "Well, I'm going to take that all and just build it into one package for you." So, instead of relying on Claude to orchestrate, now for myself, Andrew, I rely on Claude to orchestrate, and it does a an eight out of 10 job on most days. So, I'm not like really running out to go and install something else. And by the way, a year

[3:44] ago, it did a two out of 10 job. And if it's an eight out of 10 today, like, man, in 6 months, I I it's going to be a a nine out of 10. And in 6 more months, it's going to be very good at this. Context windows are getting longer. It's getting better at thinking and planning. Okay, that's the second most popular repo of the week. The third most popular is This is a developer well known because he's a big teacher. A lot of people talk about him on X. And he basically said, "You know what? I'm going to give you everything that I have just about in my

[4:13] setup." And I really like this because yes, his engineering skills here are not for me. He's way more advanced than I am. But he also includes some personal productivity skills here like how he edits an article, how he uh the skill that he uses to write an article, his productivity skills. Uh oh, here. This one

[4:33] We talked about Caveman last week. He's got his own version of Caveman. And what Caveman for him does is essentially the same thing. It simplifies what he writes before he sends it over to uh to Claude. And then before Claude writes back to him, it simplifies it again so he's not paying as much for tokens. My favorite ones are Where is it? In miscellaneous maybe? It's somewhere in

[4:53] here where he's got his writing skills. And as as a writer, I like it. What do you think of this? Andrew, I love it. You you hit on the reason why, which is, you know, we see like Gary the the G stack. And it seems very curated. It doesn't look realistic. If you look at someone like my Claude setup, this is exactly what my Claude skill setup looks like. It's a hodgepodge of skills. I've taken other ideas and customized them. You know, every time you share one of these new skill kits with me, I'm like, "Ooh, I want to steal that skill." And instead of downloading the whole package, I just

[5:25] grab that one skill. I customize it a little for my use case. And I drop it into to my platform. So, I think the way that that I'd recommend anyone look at all these skill packages is, "Hey, take the ones you like. Build your own. Build up your own little kit and customize it

[5:39] over time." Okay, next. I like that. Very practical. And I like seeing how other people work. This one is so scary. Dude, this is going to look anyone up across 3,000 sites. Um Oh man, I I don't know what to think of this one. And look at the sites that it goes after. Where is it? It's like looking at Facebook and other sites and really just helping you hunt people down. It's It's the kind of thing that if you're a stalker, you're going to love. Facebook, where is it? I don't see it here. Uh I think I gave a link to There, it's under main Oh, here, site

[6:14] list. Yes. I right there. [snorts] Oh my god. 3,000 plus sites. [laughter] Like Facebook, Instagram, GitHub. Just basically, if you want to stalk somebody, that's what it's for. What's a What's a good use of this? a that's a negative spin on it. You know, the a lot of people in your audience and and in our like studio businesses, look, we're trying to find

[6:35] leads. We're trying to find sales leads. We're trying to find talent. Uh we're trying to understand people. And I would much rather deeply understand a person before I reach out to them uh uh know, to sell them something or to hire them. So, we do this already, but honestly, we do compared to this what looks like a terrible tiny version of it. We're scraping, you know, LinkedIn, Twitter, Reddit. Uh so, I'm very curious to try this out to basically enrich our leads at a at a higher level. I'm curious how expensive it will be. Cuz it looks like a lot of AI's going to have to run to scrape all these things. But I

[7:09] absolutely I will try this this week. I get it. So, you're saying, look, we're thinking of investing in someone or hiring someone, I'm already looking them up on LinkedIn. That's the baby version of this. I want the the bigger thing. What are they doing on GitHub? What do they post on Reddit? Let's really understand the person, not for nefarious reasons. Okay, fair. Great great tool here, available of course on GitHub, and we'll have a link to everything for

[7:32] everybody below. All right. Next, Warp. Get a swarm of AI agents reviewing and fixing pull requests on your repo for free. Again, swarm of agents seems to be the big one. This one is sponsored by Anthropic on uh here you can see it on GitHub. I like that they've got that support. It is incredibly popular. Open I'm sorry,

[7:53] not Anthropic, OpenAI. And um I like that they've got this open this video on X that really took off with 3 million views. I don't know like how much of this is fake and how much is real. 2.7 million views. But clearly there's a lot of interest in this. What do you think of this one? What's the purpose of it and Yeah, I mean I I there's more and more of these that are coming out. Code Rabbit, I think is kind of a competitor to this that's that's been out a little longer. Uh I think the play here from OpenAI and others is they want access to

[8:23] this data. When the AI is running, even if it's Anthropic, you're using the Anthropic models, like OpenAI wants to sponsor groups like this so that they can learn like, "Ooh, how's the Anthropic model fixing bugs today?" But from a user perspective, I I think these are pretty cool. Uh it's it's nice to have someone watching and reviewing and especially if you're pushing out a bunch of vibe coded slop or you know, slop adjacent, like it's nice to have somebody watching your back. And so I absolutely, if you're a a one-man shop, I would be running a tool like this as

[8:54] you're as you're deploying code. By the way, if you are building anything, I want you all to know about Zapier SDK. Yes, they are our sponsor, but I've been a customer of theirs long before, I mean, over a decade before they were a sponsor. They now have over 9,000 apps in their catalog, closing in on 10,000. And what it does is it allows you to connect anything you build, even if it's just your own internal agent, to all these different apps in a way that's safe. So especially if you're listening to us here, the kind of person who likes to mess around with different tools, you

[9:23] want to give it access to Gmail because you don't just want it to, you know, be a nice little chatbot. You want it to actually help you uh clear your inbox or help you write messages. But you don't want it to send messages and you don't want it to delete messages. The thing about the SDK is it lets you make the connection to Gmail, to HubSpot, to so many other tools. Like I said, over 9,000 tools, but you decide what kind of power you want it to have and what kind of restrictions you want to impose on it. And if you have a team, this is the

[9:50] kind of tool that Zapier is making available to teams so that the the person at the top of the company can make sure that no one's just randomly connecting something and it stays there and lingers and then before long you end up with a problem. It lets you organize your team, organize your your way that you're working with um with all these tools that we're talking about here today. If you build, go to

[10:10] zapier.com/sdk. All right. Next. Uh J Code run 10 10x more coding agents on the same laptop without our upgrading hardware. This is a theme we're seeing over and over this week, right? More coding agents. You know, before we even get into this, let's just talk about why people are using so many coding agents right now. Yeah, you know, I I've been frustrated myself the last maybe 30 to 60 days with with rate limits. Like I'm using Claude, I'm using Codex, but the the rate limits have been awful and they keep lowering them. And so I've been using Open Code a lot more in the last 2 weeks and I think

[10:49] J Code here is like a it's it's a new a new age version of Open Code. This is why I can run a lot of these agents, I can use local models, I can switch between different models. So as you're hitting rate limits or if things are getting expensive, it makes it very easy to switch back and forth. And then I personally love competition in the open source space between these open coding harnesses because everyone's going to have something that they're going to win on. And Open Code wants to be expansive, they want to be this big platform. J Code, very clearly, they want to be

[11:21] fast. They want to be performant. I want this to run on smaller computers, like anyone's laptop. Look at this, 27 megabytes of RAM versus the third of a gigabyte that Claude Code uses. Like awesome. If if they want to win on performance, that's going to have a a niche for sure. And that's on one active session. We slide over to the right here and we look at 10 active sessions and the numbers

[11:43] are even more outrageously different. Okay, let's go on to the next one. This one is the seventh most popular of the week. It is Dexter ship paid AI finance product to non-technical customers over WhatsApp without writing the agent loop yourself. Um Let's look at the problem before. Here's Here's the repo on uh GitHub. Um but I actually find that with this one, it's This is a great article and of course we'll link to it explaining what the

[12:11] problem is. We need to aggregate data from multiple sources, perform uh multi-step reasoning, cross-reference information, validate accuracy before making decisions that impact real money. Most AI tools just execute commands without checking their own work. We spend hours cross-checking output, fixing hallucinations, rerunning queries because agents don't self-validate. They confidently return incorrect data, mix up time periods. You see all this problem existing tools like AutoGPT and LangChain agents are general purpose, which means they lack the domain-specific validation layers that

[12:43] financial research requires. And so that's what this is. That's the problem that this is solving and it consists of four specialized agents, the planning agent, the action agent which execu- which executes tasks, the validation agent which rigorously checks output for accuracy, and the answer agent synthesizes validated findings into financial research output. I I do see, Adam, we do have a lot of people in our audience who are big and heavy in finance and I can see why this is so This type of thing is so popular

[13:11] right now. Yeah. You know, and I I'd point out here that that you know, this word validation and there's governance, there's guardrails. You know, the These are things that that often are missed by vibe coders when they're first starting to build up something. They You focus on the happy path. that's what we call it in engineering, which is, you know, what's the thing if if the user does exactly what I expect and the computer works and the AI works and everything's perfect, here's the flow of the app. That's what you describe to the AI. And especially when you start to get into real customer data, you get into money, you get into

[13:44] security considerations, you really need to be focusing on this like validation and guardrails side, too. And it and it's honestly kind of scary when we see agents like this that don't hype up the validation and that don't hype up that no, we're actually checking, we're double-checking, and we think before we act. Like that's the place where I where I want them thinking. And if you're building something like this, I want you thinking about that, too. Like go and look at this. I love open source because you can read this code yourself, you can read the documentation, you can point Claude code at this repository and say,

[14:18] "Hey, I want to add validation like they have it. Help me do that." Uh but I I highly recommend you like really focus on the validation of these uh coding or the the trading models specifically. Next, the eighth most popular of the week is Anthropic financial services. Um let's take a look at the official launch Actually, the official launch on

[14:43] Twitter. We're expanding Claude for financial services. So, it's an expansion with Excel add-on, new connectors for real-time data, market analytics, pre-built agent skills including cash flow models, initiating coverage reports. I'll look at some of these uh in a moment, but I also want to show some of what's on GitHub where we're basically looking at a collection of different uh agents that are here to support your investment decisions. Very similar to what we're

[15:12] seeing throughout right now. Yeah, and this is really easy to install. Also, this is a a Claude co-work and and Claude code plug-in. So, if you just open the Claude desktop app, it's like three clicks to install all of this and you get the skills, you get the sub agents, you get the slash commands. It they've expanded here the data connectors. Plugins are a pretty cool kind of combination of all these other things we've been talking about. And I think part of the reason you're seeing so many finance AI apps now is that I think about 3 months ago Claude released the base financial services plugin. And

[15:49] other people looked at it and they said, "Oh, I could build an app around that." You know, Claude made it nice to converse with, but well, I could build a user experience. I could plug that into WhatsApp and I could actually bring this to customers. Take this from just some AI capability and and turn it into a package. Uh so, I my guess is if you look at the core of some of these other finance apps, you're going to see a lot of echoes of what Anthropic has released here. I mean, it's a it's an amazing

[16:15] pattern, like absolutely open source. Like, go and look at it. Go and read this. Um and and build something around it. But, [snorts] overall, I I I think these the Anthropic I've tried a lot of them. I haven't tried the financial services one, but I've used a bunch of their other plugins. They are like top-notch. I love these. Uh Claude make a trading agent that generates a $100 million

[16:37] profit per month, make no mistakes. Uh-huh. Me using VS lookup AI, tornado coming at me. [snorts] And of course, uh Claude pay me 1 million dollars, make no mistakes. All right, great. Let's move on to the next one. Number nine, most popular. This one is very practical and what we're seeing is again from the Alibaba team. They are getting better at turning your prompts

[17:01] into videos. Uh I've got a few links here that anyone's going to be able to download in the uh comments, but I want you to just take a look at [snorts] their official video. Uh everything's here is in Chinese. There should to a language selector at the top of the readme there. Really? I didn't know that was there.

[17:20] top of the there. There you go. Oh, look at this. Okay. This video is taking a little bit of time. Essentially, this is this is a video creator from the Anthropic team. We're seeing AI get better and better. I don't think this is the best that we have out there, but it's interesting. Here.

[17:46] This is my favorite part of it. We present Pixel video instead of we present. So, they're they're they're very AI-ish in this. It's pretty interesting. Um like this she does not look like a real person. I guess maybe maybe one that has a lot of the Instagram filter. is this using local models? Good question. Um I think it is. I mean, that's

[18:08] What is it that I saw it? That's very impressive then if it is. I could be Oh, yeah. Using Comfy So, I use ComfyUI a lot as well for local uh video and image generation. Uh that's What is ComfyUI? It's a it's also open source. You can download it and install a bunch of image models locally. Like so, you know, disconnect from the internet and you can still use them. Uh and whatever fits on your computer like ComfyUI will help you set it up. I I find the interface very confusing and hard to use. It's a lot of like dragging nodes and wires around, but I could just ask Claude to say, "Hey

[18:44] Claude, make me a ComfyUI you know, setup that can generate this sort of image." And and I paste it in and and the dang thing just works. It's pretty awesome. All right. We're seeing a lot of this more and more. Uh just type it, you'll get your audio. Just type it, you'll get your video. This is not really the solution for like lip syncing from what I've seen, but we're still looking at um a really useful tool that's just

[19:08] continuing to get better. This This is for scraping. Um scraping is a huge one. I had to scrape in order to put this these slides together. All this research was scraped. I had my browser open up uh X because X is very um protective of their stuff. Um, I even had my browser open up YouTube so that I can pull in data. I

[19:30] found that would be the easiest. So, scraping tools are super helpful right now. Yeah. I mean, I've used Scraping a lot over the last uh I don't even know. At least year, but but probably multiple years. Uh it's a very good tool. You know, the problem is as the structure of a page changes, you know, let's say I build a scraper today to scrape LinkedIn. Well, what happens

[19:52] when LinkedIn updates their design? Well, now my scraper doesn't know how to find the data I was looking for anymore. And so, in a tool like Scraping, instead of telling it, "Here's exactly the spot to look for." You kind of give it contextual clues of the data you're looking for, where it's at, how it's structured. So, as the website changes, Scraping will adjust and build a new scraping plan. And if it fails to find the data, then it'll let you know, and and you can go in and edit it. But, for the most part, I find it to to self-heal

[20:22] very accurately. That's actually really the most interesting part for me of it, the self-healing. What are you scraping? Oh man, everything. I mean, everything you just said. I scrape YouTube stats. Like, I scrape LinkedIn because LinkedIn won't share out impression data over the API. Uh yeah, any any of that stuff. That's That's why I'm so excited also to test the the uh person scraper you talked about before cuz it's very difficult to build and organize scrapers. And if someone else is going to do all that work for me, then I am I am all about it. Plus, like, by the way, Andrew, these these things like Scraping

[20:57] has like anti-bot detection built in. And this is a battle. If you're trying to scrape LinkedIn, like, LinkedIn doesn't want that. You mentioned this on on X. Like, they don't want that. And so, Scraping has to keep up and say "Ooh, here's how we can get around the CAPTCHAs. Here's how we can jiggle the mouse in just the right way so it looks like a human is using it." You as the developer or Vibecode, you don't have to worry about any of that. That's all

[21:19] taken over by by Scraping Bee. I've got a link here for anyone who wants to have a walk-through of it. Um this is a good video that shows you how it works and it's of course linked right down here. Um let's move on. So that's the 10th most and I usually would have stopped here because I wanted to do a top 10, but I had to show these two. Next one, DocuSeal. I love the way that this guy had written it up. He's basically saying, "Look, we all are paying a ton

[21:45] for DocuSign." I actually don't pay for DocuSign. I use a cheap alternative, but I see so many businesses using it. I don't even like paying for the cheap alternative. It's such an easy thing. Um this is now something that will take Maybe this is not a team of 50 paying 24,000, but I I wonder actually these days I do see more and more people being willing to just throw this at Cloud Code and uh and use this open source tool themselves. But we're looking at a really inexpensive uh solution to getting documents signed that you can also build into your own tools. So you don't have to just use this right out of

[22:21] the box as like a Docu- a DocuSign alternative. You can build with Yeah, that's my favorite feature. I I think you know, our business is actually use DocuSign, but I but I hate that it's an experience we don't own. We can't brand it very effectively. It's a separate link to a separate site with other emails. I want it to be in We have

[22:38] these like beautiful digital contracts. I want it to be in there instead. So I'm interested to check this out and and see uh see how secure it is, see how reliable it is, but honestly it looks pretty high quality just from the little demo we've done so far. All right. And finally, this is one that

[22:57] we did a full video on. This is Symphony. How would you describe Symphony? And I've got our video walk-through right here. Well, I've described it as someone's side project. Uh but uh OpenAI made this the the idea is that hey, we're all using tools like Jira, Trello, Linear to manage our tasks for our teams. And now we have AI doing the work, but it still takes a person in

[23:20] the middle to tell the AI what to do. Well, Symphony is an agent that you can run on your computer, and it will connect to your Linear and connect locally to CodeX. And as soon as a task is created inside Linear, your agent locally automatically, you don't have to do anything, the agent will grab that task and ooh, I can do that work. It'll move it to the in progress state, it will start working, and then it'll ask questions, and then once it's done working, it'll commit all that code, drop it in the review column, and say, "Great, human, can you review this for

[23:52] me, please?" I'm looking for that in this video here. That was the coolest part that you can actually just see the cards get moving through the Kanban board as the agent is working on it. Here are the downsides that you pointed out. Number one, it's really slow. Just to create a basic markdown file, it created all these different tasks. And then yeah, you can see it actually very quickly uh crossed it out as it was doing it. Um so here like one frame, next one, you see that it's doing it. But also, it's a lot of steps for something as simple as a simple markdown remember right, it was 300,000

[24:29] tokens that it burned to create one demo markdown file. I asked regular CodeX to do it just on my computer, and it and it used about 14,000 tokens. Still seems like a lot, but far fewer than 300,000. I mean, the benefits here though is like why I'm already in Linear every day, and so I can stay in Linear. If I'm a project manager, I can like I can reply instead of sitting down with a coding agent and working locally, I can just reply in a comment and the agent will see, oh, Adam just commented this. Well, let me go and work on that. Let me check that

[25:05] for him. So, you can have a conversation with the agent all inside linear. Multiple people on your team can share it. So, it's much easier to to share around. That said, I think Symphony as as a a tool all on its own is feels like a hobby project. They've started building some of these features directly into the Codex desktop app, but that's a little bit higher quality, although you know, I think the best is most

[25:29] likely yet to come on task management. By the way, are you Do you call your computer Big Boy? I do. Every one of my computers, you know, they have a they have some name. It's a It's a beefy, heavy, big laptop. So, this one got called Big Boy. All right, there it is, the top 12 of the week. Keep letting us know what you think of this program. We're trying to evaluate it and show you what's going on on GitHub and use as many of these as we can. Let us know in the comments and if you want this whole report here with all the links that I personally curated and

[26:03] spent a lot of time going over, you will get it in the in the description, not in the comments. Bye. Bye, everyone.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt