Build These Projects and I will Hire You (Part 2)

Summary

Highlights

• Jason Goodison outlines project ideas that could lead to a job opportunity. 🌟
• Console.ext helps monitor errors and notifies developers even when asleep, preventing revenue loss. ⏰
• By uploading public domain books to YouTube with added visuals, you can educate and earn through ads. 🎥
• AI brand manager can manage emails, sort important ones, and optimize sponsorship deals efficiently. 📬
• Onboarding Buddy uses AI to make understanding new codebases a breeze for developers. 💡

Key Takeaways

• Create impactful projects and you might get hired or start a successful startup! 🚀
• Console.ext can help early startups manage critical errors effectively and cheaply. 🖥️
• Monetizing classic books through YouTube can provide free content and revenue. 📚
• An AI brand manager can streamline influencer email management, saving time and maximizing opportunities. 📧
• Onboarding Buddy AI can make new developers' lives easier by efficiently navigating codebases. 👨‍💻

Overview

Chapters

• 00:00 - 00:30: Introduction to Startup Project Ideas The chapter introduces the concept of startup project ideas that not only promise potential employment but also serve as viable business ventures. The author offers to hire or at least interview anyone who successfully builds one of these projects. The focus is on projects that can grow into real startups beyond their initial purpose. One discussed idea is 'console.ext', aimed at addressing critical bugs in startup systems, especially during off-hours, to prevent revenue loss.
• 00:30 - 03:30: Project 1: Console.ext for Error Notifications In this chapter, the focus is on setting up a system for error notifications using a project called 'Console.ext.' The process begins with establishing a logging platform such as Data Dog, which is noted for being complex and enterprise-focused. One of the quirks mentioned about Data Dog is the necessity to select your account region explicitly, or else risk being unable to access your account. This platform is good for logging errors but lacks the functionality to notify you immediately in critical situations, particularly when you're asleep.
• 03:30 - 05:30: Project 2: Classic Books Monetization on YouTube The chapter discusses an innovative tool developed for early-stage startups, including YC companies, to handle error notifications efficiently. Traditionally, these companies rely on customers to notify them of errors, which is not ideal. The new solution, console.ext, allows developers to override the console object and include a text function. This enables automatic text notifications whenever a specific line of code is executed, particularly in critical try-catch blocks. The setup is straightforward, requiring only a single line of code to integrate functionality.
• 05:30 - 08:00: Project 3: AI Brand Manager for Creators The chapter discusses the creation of an AI Brand Manager product aimed at helping creators, which is designed to be easy to use with a simple sign-up and integration process. It highlights the importance of managing text notifications efficiently, especially in scenarios where errors might cause repetitive alerts. To address this, it suggests implementing API requests with duplicate detection or rate limiting to ensure the system doesn't overload users with messages. Additionally, it emphasizes the need for a comprehensive dashboard where users can track all text requests, including those hindered by rate limits, to maintain transparency and control.
• 08:00 - 10:30: Project 4: Onboarding Buddy AI for Developers This chapter focuses on developing an Onboarding Buddy AI specifically tailored for developers, incorporating various notification and logging features. Key tasks include changing the recipient of on-call notifications, integrating a call function along with text notifications, and establishing a retry system to ensure alerts are received, requiring a resend after 5 minutes if not acknowledged. For the logging and notification system, fork and customize an existing open-source project like Winston, a Node.js logging library, which manages log levels and formatting.
• 10:30 - 11:00: Conclusion and Submission Guidelines The chapter provides guidance on implementing a system for handling event notifications. Major tasks include adding a console extension override, implementing an SMS/calling layer, building a management dashboard, and integrating rate limiting and deduplication logic. The potential for monetization by offering this setup to interested parties, such as YC companies, is highlighted, with an estimated starting cost of $10 per month for the service. The text stresses the ease and affordability of setting up such a system.

Build These Projects and I will Hire You (Part 2) Transcription

• 00:00 - 00:30 Do these projects and I will hire you. Part two. If you build one of these projects, I will hire you or at least interview you. These projects are meant to double as actual startup ideas. So, I will pay you for your service and you can actually go on to build these into real startups that you can forget about me entirely. The first is console.ext. Now, I think this is a really great startup idea. So, let's say you launch your new startup and there's a critical bug in the checkout flow. So, you start getting errors when people are trying to buy. The problem is you're fast asleep and you're losing thousands of dollars per hour. So, how do you even stop
• 00:30 - 01:00 something like that? Well, step one is you set up a logging platform like Data Dog, right? It's this huge, clunky, pain in the ass system that's built for enterprises. So, in Data Dog, believe it or not, you even need to select the region that you set your account up in. And if you don't log into that exact URL every single time, they will pretend your account doesn't even exist. It's It's just so bad. This logs the errors, but that's not really good when you're asleep. You need something to actually wake you up. So that's when you plug into something like pager duty and that will call or text you based on some conditions. You could say something like
• 01:00 - 01:30 after five errors in data dog call me. Most early stage startups including YC companies have never done this. They rely on their customers to text them when there's an error which is a really bad look. So here's what console.ext does. It overrides the console object to include a text function. That's it. It will text you whenever a certain line of code is run. So in my really important try catch blocks, I'm going to say console.ext instead of console. error. Now, this should be one line of code to set up. Something like console set key or something like that. It's so easy
• 01:30 - 02:00 that anyone can do it with two clicks to sign up and one line of code. A few points on something like this. You cannot send a text on every single time that line is hit in case there's a loop causing the error and you're going to send thousands of texts. So, instead have it be some kind of API request and then dduplicate it or rate limit on your side. I've recommend something like one text every 5 minutes. So, there also needs to be a dashboard where they can see every single text request and even if that text request was not delivered because of rate limiting, they should be able to see every single one. And on that dashboard, they should also be able
• 02:00 - 02:30 to change the recipient of their on call person in case they have a few employees. You should also plug in a call function in case they want to use more than just text. You're also going to have to build a retry system. So, after about 5 minutes of no reply or no answer to the call, you're going to have to call or text again. You really have to wake people up. for implementing this type of logging and notification system. You could fork and build upon an established open-source project like Winston, for example. Winston's an open- source logging library for Node.js handles like log levels, formatting,
• 02:30 - 03:00 buffering, and flushing of the events already done. You don't even have to do it. The main work would be adding a console.ext override, implementing the SMS/calling layer, building a simple dashboard for managing the notifications, and then adding the rate limiting and dduplication logic. I bet any YC company would set this up in less than 5 minutes and would be willing to pay at least $10 a month to start. It's super easy to get up and going and it's so cheap, you might as well. To take it a step further, if you automatically log all console logs for them in the
• 03:00 - 03:30 dashboard, too, this would essentially make data dog mostly irrelevant for really early stage startups. The problem is overriding console.log is super dangerous. So, if you try it, be extremely careful. Project number two, classic books monetization. So, here's a really cool opportunity. It's perfectly legal, and most people don't even know about it. Many people don't realize that classic books like Sherlock Holmes, Dracula, The Great Gatsby, really anything before 1928, it's actually completely free. It's all public domain. Yet, publishers still print these books
• 03:30 - 04:00 and charge 20 to $30 per piece. So, here's the opportunity. Take these books that are already public domain and create highquality audio versions using 11 Labs or any AI software and also create picture art for it. The audio from 11 Labs sounds like a real person and it will only take you 10 to 20 minutes instead of tens of hours. To Sherlock Holmes, she is always the woman. I have seldom heard him mention her under any other name. If you add a visual aspect too with AI art, that would really help keep people engaged. Now, instead of selling these books, all
• 04:00 - 04:30 you're going to do is upload them to YouTube with titles like full Sherlock Holmes audio book, complete novel, or something like that. Then, you're going to monetize through YouTube ads while providing free content to the world. you're really going to be helping people avoid paying for what should be free. So to implement this, you're going to need a few things. First off, you're going to need to download the source material. So there's a project called Project Gutenberg, and they have perfectly formatted versions of all the classic books. You can even sort by popular to figure out which books to start with.
• 04:30 - 05:00 Then you're going to create audio versions of each of the books, obviously. And to do that, you'd probably use a voice AI service like 11 Labs or maybe Play HD. And then you're just going to post them on YouTube and optimize for YouTube. So create some chapters, create timestamps, SEO uh friendly titles that will appear whenever someone searches like buy book name for example. But so YouTube is a high domain service. So this should get SEOed pretty easily. You're going to get really high view durations too because it's an entire book people are listening to. So YouTube ads will pay you a lot. Now if you do this project for me, the
• 05:00 - 05:30 first one would be crime and punishment. I would subscribe to you immediately. The best part, this is completely legal and it actually helps people by showing them they don't need to pay for these books. You could really start this one with one book and little to no budget at all. Now, the next project is an AI brand manager. So, here's the thing. Managing your inbox as a creator or an influencer is a pain in the ass. By the way, I hate calling myself that. You're getting hundreds or thousands of emails per day. Viewers wanting to connect, brands trying to work with you, random spam, important business stuff. It's all
• 05:30 - 06:00 mixed in. It's chaos. So, what usually happens is you either one, get overwhelmed and miss important opportunities, or you spend hours every day sorting through emails yourself instead of creating content, or three, most commonly, you hire a brand manager, which takes 15% of all ad deals. And really, they mostly just reply to the emails that are already there in your inbox. Even if you do manage to stay on top of it, you're probably not maximizing your sponsorship deals. It's just way too much manual work. What creators really need is an AI assistant
• 06:00 - 06:30 that can help you intelligently handle all of this. Sort the important stuff from the noise. Negotiate with deals on your behalf and bubble up viewer messages so they don't get buried. And it'll only bug you when it's absolutely necessary. So firstly, we need to create a smart email classification system. This will automatically sort incoming emails into viewer messages and sponsorship opportunities. And then everything that doesn't fit into one of those two buckets, just archive it. Most YouTubers actually want to answer messages from scrappy kids that are just
• 06:30 - 07:00 looking to learn. The problem is they get buried in the inbox and they just never see them in the first place. For your sponsorship opportunities, you can plug in exactly what your fees are and what types of products you're willing to promote in the first place. When sponsors reach out and say something like, "Hey, we want to work with you." This tool will automatically reply and ask for those details. You can also have it integrate with your YouTube analytics and it can automatically grab your media kit for you in case asked. And then last but not least, it will just politely decline all of the sponsorship opportunities that don't match your
• 07:00 - 07:30 criteria. If you're going to do this, you should also make a dashboard. The dashboard should show you all the emails and responses, kind of like superhuman does. And it should also give you a deals page, which kind of acts like a CRM. Shows you all the deals in the pipeline. That really is a fancy term. It's just going to look like a Trello board, right? Where each deal is a card on the Trello board, and you have maybe five different categories. So, it's moving along from new leads to vetted to ready to review and finally to closed. By automating email management, creators can focus on content while ensuring no
• 07:30 - 08:00 valuable opportunities slip through the cracks. Okay, next one. Onboarding buddy AI. So, have you ever joined a new company and you spend weeks trying to understand their codebase? Um, onboarding buddy is your AI powered tour guide through any codebase. So, here's what it's going to do. It's going to take your repo. It's going to analyze it end to end and it's going to create an interactive learning path customized for new developers. The problem is obvious, right? New developers spend weeks or months trying to understand how everything fits together. Senior devs waste countless hours explaining the
• 08:00 - 08:30 same architectural concepts over and over again. And obviously documentation gets outdated fast because nobody wants to maintain it. Onboarding Buddy is going to solve it automatically. So you're going to drop in your repo and it's going to generate a system architectural map showing how all the pieces connect. It's going to identify the critical code paths, right? The 20% of code that you actually have to understand to really be effective. It's going to create dependency graphs and show you exactly how classes rely on each other. And it's going to build interactive tutorials like step-by-step
• 08:30 - 09:00 walkthroughs of key workflows. The magic is how it's going to determine what's even important. So it's going to analyze Git history to find the files which change the most frequently. It's going to map data flow to identify the critical code paths. It's going to use AI to understand the architectural patterns. And it's going to look at test coverage to find the core functionality. It could even send out funny code report newsletters like fire ship does. Um, but specifically for your company and obviously emails, not videos. To implement this, I have a few tips for you. At the core of Onboarding Buddies Engine, is something called an abstract
• 09:00 - 09:30 syntax tree as you've probably heard it before. When a human reads code, we don't just scan lines. We parse functions, loops, variables, classes. We have understand how they all connect. And a lets a computer do that too. It acts like a tree representation structure of the source code. It doesn't care about the comments or the whites space. It just breaks the code into meaningful pieces, functions, variables, imports, conditionals, all that kind of stuff. Okay, here's what you might do. So, you're going to parse the code into a using language specific parsers. And then you're going to convert the codebase into all as for analysis.
• 09:30 - 10:00 You're going to walk the tree to find all of the relationships. and we traverse the tree map out with function calls, variable references, all that kind of stuff. Then you're going to overlay the test coverage and the documentation. The a is going to give exact locations of test coverage so we can prioritize areas that are well tested and documented or not. Then you're going to feed this into an AI model. So once you have the a and the metadata, the commit frequency, the test coverages, all that kind of stuff, we're going to selectively feed that into an LLM to generate the architectural
• 10:00 - 10:30 diagrams, the code explanations, and the tutorials and everything. The pricing would be really simple on this one. Something like $20 a month per onboarding user. Any tech startup worth its salt would absolutely pay that to avoid weeks of lost productivity from senior devs. Okay, this is part two in a video I've done before. So, I wanted to check in on the rules and also what happened last time. So, the last time we did this video, we got 178,000 views, but only 22 submissions and zero submissions for two project ideas. So, if you're feeling like there's no point
• 10:30 - 11:00 in working on it cuz someone else will beat you to it, like think again. I'll pay for the best service for each project and that'll be judged at the end of every month. Entries are based on the UI, like the functionality, and just generally what a real customer would want when purchasing something. If you're looking to work with me too after the submission, just let me know and and we can talk about it. To go over the rules one last time, if you create a submission, join the Discord below and there's a dedicated channel to each project. You can just paste it in there. The luck.