I got my degree last year in IT and haven’t been able to find a job. But, I have started making automation tools for my family’s service business for things like record keeping. This lets my family and I be able to do the work of multiple people with less effort. There are lots of ways to get creative with this and I love hearing about the topic.
"Start by building something that would be useful to you yourself. That way, if it doesn't work out, you still have the software that you built to use yourself." That's where the majority of our current tools came from anyway. All devs should strive to be nice to future devs (especially the ones with axes who know where you live 😃).
I could probably say that even after 15 years in programming I still find it challenging to write a software from start to finish even with GPTs that everyone is glorifying. I don't know about these cool kids who are claiming how AI are making them tons of stuff easily but for me writing a software takes time and I need to understand everything is written so I am sure I can maintain it and edit it whenever I want to. I just couldn't understand those tweets of someone who never wrote a line of code and says: "tech is over, I will soon write a prompt and a whole software will be written for me"
Yeah a lot of them are doing it purely for engagement. If you can drag together an app without writing any code it will fall apart the first time something won't work as you need it too.
You're not alone. A lot of people "still find it challenging to write a software from start to finish" even after programming for years. Largely because there's not a lot of opportunity to get experience doing that when you're working for someone else as part of a team. Which is why I advocate for doing your own thing - even if that thing, itself, doesn't end up making you any money
I would just say this. If you think your idea is dumb or no one would use it, but that is your only idea, still work on it. If anything, you will just get int the mode of working on your own thing and that act in and of itself may trigger a better idea. And also, the odds of making millions from any idea is so low it's not even worth calculating. It's all about starting.
I am currently trying to scrape some time to make a game in C++, but not for money, just to not lose my skills in the language as I work in Python in my day job.
“Most of them are full of SHIT”! So true. I love the clickbait titles - Create a Full Stack Web App With One Prompt! Good luck with deploying and maintaining that!
Carl. I want you to know - your channel is just really great. I run a scaling up tech company that is shifting to leverage more a.i. tools internally and customer-facing. And my son (13yrs) and I are building a different company using A.I., Claude and Cursor. I have been following a.i. since before chatGPT (from the likes of Philip from Two Minute Papers and Dr. Alan Thompson) and I love how your videos provide that reasoned and rightfully skeptical approach. I very much respect your opinion and views, and regularly share your videos with our product manager and our engineering team. So I just want you to know - you're very much appreciated and keep up the great work! PS - I read The Lean Startup 20+ times and can probably recite it 🙃
In a universe of content that constantly pivots between extreme AI hype and dev doom and gloom, your content is an oasis of sanity and relevance. Thank you Carl!
Came up with a webapp idea about a month ago on language learning about a month ago. Engineer is currently building it to spec. Decided I want to learn this stuff to help contribute to its maintenance. Having a project you’re passionate about is a crazy incentive to learn something. Im addicted
I‘ve spend a few hours over the last few months starting a side project. Can‘t wait for other videos on this topics. Thanks a lot and keep up the good work!
I wish you luck with that (sincerely). Shifting careers is challenging, although it's better to do it than to continue in a career you regret. I hope you find something you like and make the transition smoothly.
@@InternetOfBugs It is not that I don´t like coding, but right now it is really harsh and I want to have a back up plan. Still hopeful that it will workout as developer, because I invested last 3 years on that.
Three years isn’t a lot of time. The minimum apprenticeship for a trade in Aus is four years. It takes 6 years minimum to become a doctor, including a year of internship. Software is hard. You should expect it to take you a long time. If you had done a traditional education path, you would only have just finished your BSci in CS (assuming a three year undergrad). Idk what your journey was but a lot of “learn to code and get a six figure job in six months” type training providers set extremely unrealistic expectations. Three years in means you’re still extremely green. I’d encourage you to push through if you really think it is what you want to do.
I’m an early career developer and have always dreamed of getting real traction on one of my projects/ideas. I’m very excited for your series! Thank you for doing this.
i have also started building a product of my own and currently in the development process but i got an offer from a startup with very promising role and now I won't be able build it with speed, but I'll definitely slowly keep it building.
Just keep the consistency and make it a habit. I worked at a startup from FFF round up until series B and still worked on my side-projects - it's totally possible.
This is what I am doing since gpt-3.5 came out. building my own rapid AI development tools. out of these I can build some useful applications. If this does not work I have build up AI dev knowledge and have a framework for consulting. So I leveled up on multiple aspects. About the website iteration idea: I would build a basic static website with all information and put it into a RAG. Then let the real website be build by an interactive LLM following visiters interests ( site by site creation ). So the website experience will constantly change.
I hate that I just never have ideas that software can solve that is not extremely personal to me. Like I set up my own vps for my blog so I could publish with emacs. Or I contributed to open source projects that I used directly. But other than that, I don't have any problems that software can solve that isn't solved by existing software that I don't have any ideas to improve that the useful software I do use. Sucks.
Try taking a step back. Software development isn't a great market, because there is so much competition - so try to avoid "problems" related to that. What else do you do? Think about issues you have with your commute (or the issues you have with the process of working from home). Think about what you do to get food, or for fun, or exercise, or other non-work things. Those are the kinds of things that are more likely to be markets. But the big thing is not to sit at your desk and think about what you want to fix with software. The thing is to try to keep in the back of your mind that you're looking for problems to solve as you go about your day, and see what you notice that you hadn't thought of before.
I have side projects, just not programming. I just can't, after work I am too exhausted to do more programming. It's just how it is Recession is guaranteed in the next 1-2 years, you are correct. Things aren't great, and programmers will be cut again
What I need most is a full batteries-included, soup-to-nuts full stack web app starter template. It needs frontend/backend/database. All the basic AAA of course. Self-hostable on Kubernetes with Helm charts. CI/CD tooling. User self-registration with various options (require user to have a specified e-mail domain, etc.). Configurable e-mail and SMS notifications to drive 2FA handshakes. Configurable OAuth providers. Built-in code generation scaffolding. User change password. OIDC support (Active Directory, Keycloak, etc.) User forget/reset password. Installers for Mac, Windows, Linux (for desktop versions of the app). Some starter end-user documentation. It is a lot of stuff to do which has nothing to do with writing your first line of app-specific business logic. If I were starting from scratch on a new project, I guess THAT itself would be the project, since that is what I need.
@@Braindouchedotnet Maybe, I don't know much about Rails. I write mostly in C#. I know Rails is really good, but I seriously doubt it does ALL of what I just described by running a single command to install a template. Sure, you can code your brains out after you get your project set up, just like you can code your brains out in C# after you run one of the supplied dotnet templates. You're right though, something like Rails for C# is kinda what I'm thinking, but it needs more than what the basic template provides I think.
@@kdietz65 you might want to look at Django (Python on rails) and Laravel (php on rails) and in particular Laravel Forge and similar. I doubt any of them give you everything you're looking for but you can get pretty close.
Don't forget about renting office space with metal detectors to deter BYOD, and have a self-hosted or cloud based telephony system in place to receive two-factor-authentication codes, and off-shore server racks across multiple continents incase of court seizures under an autonomous system providing CDN from the Maldives or Seychelles. I know a guy who knows a guy that can hook you up with basic infrastructure in Romania. This is all allegedly and haven't tested it myself.
I didn' start watching because I saw the obvious click bait. I started watching because I saw you had a new video. And by the way, you are ruthless. "Use AI to ... write text for your website and then A/B test" 🤣. That was cold and cruel man. When I saw the title I thought they got to you and paid you money.
It's been 2 years since I couldn't find a job as a programmer. In the meantime I started a personal project that I think has commercial potential and I'm working on an MVP, but even that requires a lot of work. During all this time I also had a job in construction out of an urgent need, but I still couldn't find a programming role. I currently refuse to commit to another role just for the money and live in abject poverty in the hope that something will appear on the horizon. I'm ashamed to ask my family for more money, but I'm afraid that if I take a job elsewhere, I'll be stuck there, unable to attend interviews or program to maintain my skills. I've done this before and all I "earned" was some money to survive and severe depression due to the meaninglessness of the job, so the shame of asking my family for money is easier to bear than that situation. I've gotten to the point where I don't even want to work as a programmer anymore and I think the entrepreneurial path is what I want. Now I just have to convince someone to fund me. :)
I just watch this as someone who made his own product so I can feel better about my decision. I had to put it on hold to work and build up some savings.
Great video, and very appliable to non-technical jobs too A friend started a business and he learned a ton about his field but also about other hard and soft skills and it made him grow a lot
great video and definetly agree and glad to say I just launched my first product, and yes if it doesnt work it’s at least useful to me and have already been using it!
I read in a comment of yours that (paraphrasing) you believe AI is useful in cases where there are many right answers, like marketing copy. My first thought when you mentioned AB testing using AI-generated marketing copy in the video is that you might be committing the same error of non-skilled software developers when they say AI will replace software development, i.e. marketing copy is something you are not skilled at, and so AI's production of it looks good enough to you. I'm not saying this is 100% the case, and can see your point about marketing having a lot more correct answers than a coding problem, but I just wondered if you had any comment about this? Thanks for the interesting video
The trick here is using A/B testing. By being able to test different AI outputs against each other, you can get quite good output (eventually). In theory, the same kind of thing could be done in the programming space, if (and only if) you had a clear fitness function that you could use as a test. For example, if you were trying to optimize code to run faster, you could, in theory, have a bunch of AIs spit out a whole bunch of different variants of the implementation, throw out all the ones that didn't get the right answer, and then test them against each other to see which was fastest. In code, though, it's rare that you have that kind of clear fitness function (except for benchmark-type stuff where you already know the answer). With marketing, copy, though, knowing what to test for (and how) is clear, so it's a better use case for AI (in my opinion).
I use ChatGPT all the time to help me with the copy, especially, with micro-copy like CTA button label etc. I asked it to rephrase my awkwardly written version or I give it a paragraph describing something and ask it to make it concise. Unlike with unskilled programmers, I am able to assess if the end result is good.
I wasn't getting any good work in my company and I was bored with doing Android development Started building my own pet project with my own backend server and everything
How important in your opinion is the economic venues for that idea? Say I make a new program that helps me do some common mundane task that either there aren't many competing software in that space, or they all suck (good opportunity). It might be useful to me, and some other subset of users, but how is this supposed to make any money? Thank you for the video, and for working against these insidious trends we all see of extensive demoralization from actually programming
I've seen that advice a lot, but I don't think it applies in this case. I think the "create an audience first" advice is mainly aimed at people who are trying to avoid having to spend money on development costs for ideas that aren't going to make back the money they spent on development. For a non-technical founder, the ability to pivot to a different business case often involves starting over from scratch (or nearly so), causing a lot of wasted dev time. For a technical founder, the incremental cost of pivoting is much less (in my experience), because we can understand what parts of the stack we can reuse, and what the minimum marginal effort of pivoting from one failed MVP to the closest possible MVP in the new direction. The other issue with us is, if you do the "put up a marketing site first, and don't start on development until you think you found a good niche" thing, you're likely to spend a ton of time iterating on marketing copy instead of programming, from which you won't learn anything about being a better programmer.
Amazing content. Thanks for sharing. What about the "create an audience" first? I think getting visibility in a crowded market it's the most difficult part. On top of that, it's a common mistake to create business or products without domain knowledge. I have ideas to build stuff but all of them are software engineering related and are hard to build specially in terms of time.
Anything that you can think of right now and think "there should be a service for this", already exists or did exist at some point. If it doesn't exist there are good reasons it doesn't exist. It's either far too difficult to actually implement, or has limited monetization potential due to said difficulty or size of user base. If you can somehow bridge that gap then there's potential. Do you have the niche knowledge necessary to bridge that gap? It's definitely possible if you do. I've seen several folks that have like ~50k followers across all social media platforms making a quarter million per year selling courses and the like. But it did take them ~5 years to get to that point. I had an idea to find a way to monetize offline twitch chat as the amount of times I've seen said offline chats jump on something and actually spend money was way too much for there to be nothing there. Plus there are close to a thousand of them with a sizable user base (~25k). Essentially a hyper targeted ad platform you can sell to ad agencies/coordinators to get a cut of revenue. Issue is the data that is there ended up being fundamentally useless (similar with google stuff), with nowhere near enough users to make those razer thin margins worth chasing. But I did gather a ton of stats on the user base which led to the further conclusion that the overwhelming majority are flat broke hence why running ads on twitch is as close to a waste of money as you can get. Most are high school/college folks/unemployed. I've seen said unemployed people get a job and the time they spend in an offline chat plummet by 99%. Neat idea but fundamentally zero money in it. Maybe if I could cut costs down by a factor of 20x it would be worth it, but that takes not insignificant upfront investment. Again have to find a gap you yourself can fill that nobody else thought was worth pursuing.
I think the "create an audience first" advice is mainly aimed at people who are trying to avoid having to spend money on development costs for ideas that aren't going to make back the money they spent on development. For a non-technical founder, the ability to pivot to a different business case often involves starting over from scratch (or nearly so), causing a lot of wasted dev time. For a technical founder, the incremental cost of pivoting is much less (in my experience), because we can understand what parts of the stack we can reuse, and what the minimum marginal effort of pivoting from one failed MVP to the closest possible MVP in the new direction. The other issue with us is, if you do the "put up a marketing site first, and don't start on development until you think you found a good niche" thing, you're likely to spend a ton of time iterating on marketing copy instead of programming, from which you won't learn anything about being a better programmer.
"Create an audience" first doesn't work for everyone. IIRC "create an audience first" as a dogma started when Joel Spolsky advertized Trello to his followers (I don't recall if it was a newsletter or Twitter or both) and he got loads of signups overnight because of it, so then people turned this into a myth and dogma. Also, the domain gotta support writing a lot about it, so if your project is a tool that does one simple job extremely well (like generating backgrounds for websites) - it's hard to build an audience for this topic. The reverse is also possible - think of all celebrities attaching their names to random things like Rihanna starting a cosmetics brand etc. Highly recommend the book "Traction: How Any Startup Can Achieve Explosive Customer Growth" by the DuckDuckGo founder.
Great video, solid message. I'm curious about the point on AB testing tho. Any analysis without a big enough sample size will likely be underpowered. The approach that VCs use as explained in this video is to run adds which gets a large sample and you can hopefully use to iterate. I'm curious what your thoughts are when doing this without a VC. I don't understand the "it takes time" message in the video
VCs spend money to get a lot of clicks on a site in a short time to get a large sample size. Without that advertising, you can still get people to come to your site (through SEO, through posting links on Social Media, through mentioning the site in blog posts or in interviews, etc), but you'll get a lot fewer people that way. So, instead of getting enough traffic for an A/B test to be significant in a few hours with ads, it might take weeks to drive enough organic traffic to get decent data.
Getting a VC round is a full-time job, and in this market - it's an impossible job without revenue and/or a proven track record and connections. One of the links @InternetOfBugs posted was to Microconf - it's a microsaas community where most (all, really) founders are bootstrapped, i.e. they already built and launched something. Rob Walling created this community not only to have a community of bootstrappers and share tips and knowledge but also it's his vehicle to find companies to invest - he invests smaller rounds and it's more "friendly" than the typical VC route. With regards to testing a hypothesis - there multiple ways, you don't have to build a page and run ads or attract traffic. You can hang out where your audience hangs out and ask questions. Even places like Reddit (just don't violate the etiquette). Also highly recommend reading the book "The Mom Test: How to talk to customers & learn if your business is a good idea when everyone is lying to you"
Can you please make a new video on the findings that inference, scaling, and synthetic data are actually improving LLMs. Take for example gpt-o1, we have not even come close to maxing out the abilities of LLMs alone, assuming no more AI breakthroughs I do not see how this will not disrupt everything. Beyond improving reasoning ability, we can make them more efficient too, so they are cheaper and integrated into everything. I think its time all the people who buried their head in the sand accept their fate.
Genuinely knowledgeable programmers are actually pretty rare in the grand scheme. Most people would Google and then copy pasta. AI has really only made that problem worse, but it was already there in the culture before AI came along.
I agree. But, in my opinion, the best way to get more knowledgeable is to build things. Most developers won't ever do that, but the ones that do tend to differentiate themselves from the pack that way.
It’s never been easier to start a Software Project but it’s also never been more competitive. Don’t go into it with the expected outcome of income, you’ll (probably) be really disappointed.
This really appealed to me because the end goal was never to work for someone. It is to be so useful to others, that they have no choice but to pay me. Hence why i never labeled myself as a "developer". I'm a problem solver.
I'm not a dev really but I'm getting into coding. What blows my mind is how "developers" are so obsessed with going to work at some shitty soul sucking corporation. It sounds like jail to me. How about you have fun with building and build something you want to see in the world? I'd rather put a fork in my eye than have to deal with 1% of the corporate world.
Doesn’t pivoting quickly and repeatedly lead to low quality, just-enough-for-now, kind of programming? I love the idea of doing side-projects, but I don’t see how it helps getting better at programming on the technical side.
It leads to embracing the YAGNI principle. Also, would you rather find what to build by pivoting and iterating rather than building something that nobody needs but with high quality code?
You are wrong. I can point out many samples from the video, but the easiest is the books you recommend. TARGET AUDIENCE FOR The Lean Product Playbook are Product Managers Startup Founders UX Designers & Engineers Business Analysts Agile Teams TARGET AUDIENCE FOR The Lean Startup are Entrepreneurs & Startup Founders Product Managers Business Leaders Investors Innovation Teams Which book you think is for software engineers ? I know you don't know, but none of them :)
I saw the title and thought, isn't that the guy who I've seen criticize ai stuff? this must not be a trick, I honestly do want to do something like that, specifically in fiction writing, preferably without AI but i'm only the first minute in so lets see what you have to say
I try to be balanced. There are things that AI is awful at (primarily things that need to be "correct" or have a right answer) and things it's helpful for, which to me is stuff where there is a huge range of potentially valid responses that vary in quality - like marketing copy or stock images.
I used it yesterday, it's better than Claude but it's still not blowing me away. The problem with these things is you can ask them to do a task that is nonsensical and it will attempt to do it. For example, ask it to create a property decorator for a class in typescript that initializes the property outside of the constructor. It will do the task eventually as you work with it, but this particular task achieves nothing because properties are already initialized outside of the constructor lol. I would expect models to be able to detect BS tasks like this and say: this is a bad idea and it shouldn't be done this way. I've also discovered something else related to this: if you ask a question about why it's doing something a certain way, it will immediately "agree" with you and try to find a different way to do it, even though you never even said that something was wrong, in fact there may have been nothing wrong. The idea that these things will work autonomously seems like complete nonsense to me.
Agreements you sign with employers where you seemingly sign over your creative life are highly unlikely to stand up in court unless the product is in direct competition with your employer, and was produced on their time. Most employers won’t care anyway, and for those that might, there are PLENTY of ways to hide things ahead of time. Be careful, but I would not let this discourage you.
@@InternetOfBugs understand that can happen sometime, but it would be a very rare occurrence, especially if you’re working on things unrelated to the LOB of your employer. My agreement specifically indicates the work must be related to said business. At any rate, I wouldn’t let that hold you back, as there are many, many ways to make sure things are never found out. Large employers are unlikely to waste time and resources on it, again, unless it’s a serious direct threat.
Thinking of an idea that is not already offered is really hard, IMO. I'd love to do this, but I'm not sure i have any novel business ideas. I'm just a second-rate coder, then, I guess. 😢
A few weeks ago I was feeling some kind of way because I felt like I was failing to come up with some "novel" or "good" idea. After listening to some other sources I decided to just lean into solving my own problems, so I made a free chrome extension to fix an issue I had at work. It wasn't very complex and I almost felt silly for publishing it in the Chrome Web Store. It's not a lot, but in less than a month I got 5 users and a 5 start review with someone calling it one of the best inventions he's seen. I have two more ideas for chrome extensions, one paid and another free one, so I'll see where that takes me.
It's here: ruclips.net/video/l9Z7KPnP0-c/видео.html It should be linked on the end-screen of this video (at least if the platform you're on has support for end-screens). I forgot to put it in the description didn't I? Thanks for pointing that out. I'll go fix that.
But surely now with o1 out it's just a matter of days before all our jobs are replaced. It's as intelligent as a whole phd I've heard. I am not as smart as a phd. So my career is over. Matter of fact I have seen somebody make flappy bird with it using just one prompt, in 5 minutes. Who can compete with that? Anyways. My current Sideproject is Coding Snake in x86 Assembly running in Dosbox. I don't think it will get me a lot of job offers, but it is a lot of fun.
That's an interesting question. I wouldn't mind doing that, but it's so different than my normal content, I'm guessing a lot of people wouldn't like it, so it would penalize the channel's stats. Let me see what I can figure out
The best part of starting a business is, you have no choice but to learn to identify BS vendors aka Grifters and Shills. If you do not gain anything technically (unlikely), you would gain a lot of life wisdom.
Don't make your channel beholden to youtube analytics. Or AB testing for that matter. If you go down that path all the way, you end up creating Mr. Beast videos with your mouth open on the thumbnail. Your strength IS the technical deep dives, so lean into that, instead of shying away to maximize views.
Info at ruclips.net/video/l9Z7KPnP0-c/видео.html on my ideas about where the channel is going.
my king, you forgot to change the filter after the BSOD. I'm still waiting for it. Help.
You're like a cool programmer uncle that I've never had.
who's name is Bob
I got my degree last year in IT and haven’t been able to find a job. But, I have started making automation tools for my family’s service business for things like record keeping. This lets my family and I be able to do the work of multiple people with less effort. There are lots of ways to get creative with this and I love hearing about the topic.
Love this! I made a calendar tool for my girlfriend's Airbnb business.
This is awesome! You gotta turn it into a product! Otherwise, you can also do the same for others via professional services model.
I absolutely love the intros to your videos. Classic "and it's SHI *ding* BSOD, rebooting." Takes me back.
"Start by building something that would be useful to you yourself.
That way, if it doesn't work out, you still have the software that you built to use yourself."
That's where the majority of our current tools came from anyway.
All devs should strive to be nice to future devs (especially the ones with axes who know where you live 😃).
I could probably say that even after 15 years in programming I still find it challenging to write a software from start to finish even with GPTs that everyone is glorifying. I don't know about these cool kids who are claiming how AI are making them tons of stuff easily but for me writing a software takes time and I need to understand everything is written so I am sure I can maintain it and edit it whenever I want to. I just couldn't understand those tweets of someone who never wrote a line of code and says: "tech is over, I will soon write a prompt and a whole software will be written for me"
Yeah a lot of them are doing it purely for engagement. If you can drag together an app without writing any code it will fall apart the first time something won't work as you need it too.
You're not alone. A lot of people "still find it challenging to write a software from start to finish" even after programming for years. Largely because there's not a lot of opportunity to get experience doing that when you're working for someone else as part of a team.
Which is why I advocate for doing your own thing - even if that thing, itself, doesn't end up making you any money
as zozin said “so I installed emacs, and bought programming socks..... where money? money when?”
There are other ways to get paid by wearing programming socks...
😏@@CEOofCulturalMarxism
@@gauravtupe1799 🥺
I would just say this. If you think your idea is dumb or no one would use it, but that is your only idea, still work on it. If anything, you will just get int the mode of working on your own thing and that act in and of itself may trigger a better idea. And also, the odds of making millions from any idea is so low it's not even worth calculating. It's all about starting.
Exactly! A dumb idea is still a better playground to develop skills than no playground.
I am currently trying to scrape some time to make a game in C++, but not for money, just to not lose my skills in the language as I work in Python in my day job.
@@DagarCoH ruclips.net/video/rX0ItVEVjHc/видео.htmlsi=h0nTjaXUptdMQ4jc
“Most of them are full of SHIT”! So true. I love the clickbait titles - Create a Full Stack Web App With One Prompt! Good luck with deploying and maintaining that!
Carl. I want you to know - your channel is just really great. I run a scaling up tech company that is shifting to leverage more a.i. tools internally and customer-facing. And my son (13yrs) and I are building a different company using A.I., Claude and Cursor. I have been following a.i. since before chatGPT (from the likes of Philip from Two Minute Papers and Dr. Alan Thompson) and I love how your videos provide that reasoned and rightfully skeptical approach. I very much respect your opinion and views, and regularly share your videos with our product manager and our engineering team. So I just want you to know - you're very much appreciated and keep up the great work! PS - I read The Lean Startup 20+ times and can probably recite it 🙃
In a universe of content that constantly pivots between extreme AI hype and dev doom and gloom, your content is an oasis of sanity and relevance. Thank you Carl!
intros from this guy are gold ... and the content too
I was going to say working for a small company where you are the one helps. It was nice to see it was mentioned.
Came up with a webapp idea about a month ago on language learning about a month ago. Engineer is currently building it to spec. Decided I want to learn this stuff to help contribute to its maintenance. Having a project you’re passionate about is a crazy incentive to learn something. Im addicted
I‘ve spend a few hours over the last few months starting a side project. Can‘t wait for other videos on this topics. Thanks a lot and keep up the good work!
I have started new project - find another field for work.
I wish you luck with that (sincerely). Shifting careers is challenging, although it's better to do it than to continue in a career you regret. I hope you find something you like and make the transition smoothly.
@@InternetOfBugs It is not that I don´t like coding, but right now it is really harsh and I want to have a back up plan. Still hopeful that it will workout as developer, because I invested last 3 years on that.
Whats the alternative ur thinking about?
@@musashi542 I have license for a mountain guide, also I was warehouse manager before learning how to code.
Three years isn’t a lot of time. The minimum apprenticeship for a trade in Aus is four years. It takes 6 years minimum to become a doctor, including a year of internship. Software is hard. You should expect it to take you a long time. If you had done a traditional education path, you would only have just finished your BSci in CS (assuming a three year undergrad). Idk what your journey was but a lot of “learn to code and get a six figure job in six months” type training providers set extremely unrealistic expectations. Three years in means you’re still extremely green. I’d encourage you to push through if you really think it is what you want to do.
Sir, there are very few people I really respect. You one of them!
Long time viewer & first time commentor. I enjoy your videos because they are well thought out. Please do more on this topic!
Can’t wait for the next videos! Thank you
You are my favorite programming channel. I look forward to your more in depth technical content.
I read Lean Startup a couple times, and its philosophy is ever-present in my head. Seeing that on your thumbnail brought me here.
keep it up we need more people like you in the world, you provide good inspiration for my career
I’m an early career developer and have always dreamed of getting real traction on one of my projects/ideas. I’m very excited for your series! Thank you for doing this.
My favorite video on career for programmers,
Please guys, see the power on your hands to build stuff, your own Stuff!
Glad you'll be making more videos about this topic, I'd like to hear more about that!
i have also started building a product of my own and currently in the development process but i got an offer from a startup with very promising role and now I won't be able build it with speed, but I'll definitely slowly keep it building.
Just keep the consistency and make it a habit. I worked at a startup from FFF round up until series B and still worked on my side-projects - it's totally possible.
This is what I am doing since gpt-3.5 came out. building my own rapid AI development tools. out of these I can build some useful applications. If this does not work I have build up AI dev knowledge and have a framework for consulting. So I leveled up on multiple aspects.
About the website iteration idea:
I would build a basic static website with all information and put it into a RAG. Then let the real website be build by an interactive LLM following visiters interests ( site by site creation ). So the website experience will constantly change.
Great video. I recently started to think about all my projects I am building as products vs code projects.
I hate that I just never have ideas that software can solve that is not extremely personal to me. Like I set up my own vps for my blog so I could publish with emacs. Or I contributed to open source projects that I used directly. But other than that, I don't have any problems that software can solve that isn't solved by existing software that I don't have any ideas to improve that the useful software I do use. Sucks.
Try taking a step back. Software development isn't a great market, because there is so much competition - so try to avoid "problems" related to that.
What else do you do?
Think about issues you have with your commute (or the issues you have with the process of working from home). Think about what you do to get food, or for fun, or exercise, or other non-work things. Those are the kinds of things that are more likely to be markets.
But the big thing is not to sit at your desk and think about what you want to fix with software. The thing is to try to keep in the back of your mind that you're looking for problems to solve as you go about your day, and see what you notice that you hadn't thought of before.
@@InternetOfBugs yeah it's good advice. Probably one of the drawbacks of being a pretty chill dude is I usually don't encounter problems. 🤷♂
This video came in right time when I am considering to build some project on my own
i like you telling storys of development more then all your AI videos
so thanks for the video, hugs from Brazil
I have side projects, just not programming. I just can't, after work I am too exhausted to do more programming. It's just how it is
Recession is guaranteed in the next 1-2 years, you are correct. Things aren't great, and programmers will be cut again
This. I can't just live with only programming in mind.
Recession is here, they just give us bad statistics.
Personal take, I am writing down ideas in notes from last 3,4 years. But never really worked on any of them. Looks like its time to put things to work
Good luck
Thank you so much ! You're amazing and your content is amazing !
Love the channel, and the BSOD that appears whenever you say something naughty really cracks me up.
Also, where can I buy that mug?
What I need most is a full batteries-included, soup-to-nuts full stack web app starter template. It needs frontend/backend/database. All the basic AAA of course. Self-hostable on Kubernetes with Helm charts. CI/CD tooling. User self-registration with various options (require user to have a specified e-mail domain, etc.). Configurable e-mail and SMS notifications to drive 2FA handshakes. Configurable OAuth providers. Built-in code generation scaffolding. User change password. OIDC support (Active Directory, Keycloak, etc.) User forget/reset password. Installers for Mac, Windows, Linux (for desktop versions of the app). Some starter end-user documentation. It is a lot of stuff to do which has nothing to do with writing your first line of app-specific business logic. If I were starting from scratch on a new project, I guess THAT itself would be the project, since that is what I need.
Isn't that just rails?
@@Braindouchedotnet Maybe, I don't know much about Rails. I write mostly in C#. I know Rails is really good, but I seriously doubt it does ALL of what I just described by running a single command to install a template. Sure, you can code your brains out after you get your project set up, just like you can code your brains out in C# after you run one of the supplied dotnet templates. You're right though, something like Rails for C# is kinda what I'm thinking, but it needs more than what the basic template provides I think.
@@kdietz65 you might want to look at Django (Python on rails) and Laravel (php on rails) and in particular Laravel Forge and similar. I doubt any of them give you everything you're looking for but you can get pretty close.
@@kdietz65 Django, Laravel or Laravel Forge, spring boot.
Don't forget about renting office space with metal detectors to deter BYOD, and have a self-hosted or cloud based telephony system in place to receive two-factor-authentication codes, and off-shore server racks across multiple continents incase of court seizures under an autonomous system providing CDN from the Maldives or Seychelles.
I know a guy who knows a guy that can hook you up with basic infrastructure in Romania.
This is all allegedly and haven't tested it myself.
You are blessed, I need this message, this video in this very moment
thanks for making this video, was streaming myself building a project on twitch.
Thanks for showing codecrafters! "Leetcode that is not useless."
Jokes aside, these two can't be even compared.
I didn' start watching because I saw the obvious click bait. I started watching because I saw you had a new video.
And by the way, you are ruthless. "Use AI to ... write text for your website and then A/B test" 🤣. That was cold and cruel man.
When I saw the title I thought they got to you and paid you money.
It's been 2 years since I couldn't find a job as a programmer. In the meantime I started a personal project that I think has commercial potential and I'm working on an MVP, but even that requires a lot of work. During all this time I also had a job in construction out of an urgent need, but I still couldn't find a programming role. I currently refuse to commit to another role just for the money and live in abject poverty in the hope that something will appear on the horizon. I'm ashamed to ask my family for more money, but I'm afraid that if I take a job elsewhere, I'll be stuck there, unable to attend interviews or program to maintain my skills. I've done this before and all I "earned" was some money to survive and severe depression due to the meaninglessness of the job, so the shame of asking my family for money is easier to bear than that situation. I've gotten to the point where I don't even want to work as a programmer anymore and I think the entrepreneurial path is what I want. Now I just have to convince someone to fund me. :)
The opening was great 😂
I just watch this as someone who made his own product so I can feel better about my decision. I had to put it on hold to work and build up some savings.
Hey much love from India good to have people like you in this community
Thank you
Great video, and very appliable to non-technical jobs too
A friend started a business and he learned a ton about his field but also about other hard and soft skills and it made him grow a lot
Thank you for this info!
I would warn developers that there are ideas that seems like cool and useful, but in fact they have no potential in money making
Thanks!
That was great unc! Thanks for sharing your intelligence
great video and definetly agree and glad to say I just launched my first product, and yes if it doesnt work it’s at least useful to me and have already been using it!
Thank you! That was very helpful.
I read in a comment of yours that (paraphrasing) you believe AI is useful in cases where there are many right answers, like marketing copy. My first thought when you mentioned AB testing using AI-generated marketing copy in the video is that you might be committing the same error of non-skilled software developers when they say AI will replace software development, i.e. marketing copy is something you are not skilled at, and so AI's production of it looks good enough to you.
I'm not saying this is 100% the case, and can see your point about marketing having a lot more correct answers than a coding problem, but I just wondered if you had any comment about this?
Thanks for the interesting video
The trick here is using A/B testing. By being able to test different AI outputs against each other, you can get quite good output (eventually).
In theory, the same kind of thing could be done in the programming space, if (and only if) you had a clear fitness function that you could use as a test.
For example, if you were trying to optimize code to run faster, you could, in theory, have a bunch of AIs spit out a whole bunch of different variants of the implementation, throw out all the ones that didn't get the right answer, and then test them against each other to see which was fastest.
In code, though, it's rare that you have that kind of clear fitness function (except for benchmark-type stuff where you already know the answer).
With marketing, copy, though, knowing what to test for (and how) is clear, so it's a better use case for AI (in my opinion).
@@InternetOfBugs Thanks, that's a nice explanation
I use ChatGPT all the time to help me with the copy, especially, with micro-copy like CTA button label etc. I asked it to rephrase my awkwardly written version or I give it a paragraph describing something and ask it to make it concise. Unlike with unskilled programmers, I am able to assess if the end result is good.
Thank you so much for your content!
Thanks for the advice, much appriciated.
Do you recommend any other books on testing and error handling other than Working Effectively with Legacy Code?
Love your channel dude, keep it up
I wasn't getting any good work in my company and I was bored with doing Android development
Started building my own pet project with my own backend server and everything
How important in your opinion is the economic venues for that idea? Say I make a new program that helps me do some common mundane task that either there aren't many competing software in that space, or they all suck (good opportunity). It might be useful to me, and some other subset of users, but how is this supposed to make any money?
Thank you for the video, and for working against these insidious trends we all see of extensive demoralization from actually programming
Thanks for the video, it was great.
The first step is always to talk to potentatial customers to see if there is a market for your idea. 20-30 interviews will give you a good idea.
I've seen that advice a lot, but I don't think it applies in this case.
I think the "create an audience first" advice is mainly aimed at people who are trying to avoid having to spend money on development costs for ideas that aren't going to make back the money they spent on development.
For a non-technical founder, the ability to pivot to a different business case often involves starting over from scratch (or nearly so), causing a lot of wasted dev time.
For a technical founder, the incremental cost of pivoting is much less (in my experience), because we can understand what parts of the stack we can reuse, and what the minimum marginal effort of pivoting from one failed MVP to the closest possible MVP in the new direction.
The other issue with us is, if you do the "put up a marketing site first, and don't start on development until you think you found a good niche" thing, you're likely to spend a ton of time iterating on marketing copy instead of programming, from which you won't learn anything about being a better programmer.
Amazing content. Thanks for sharing. What about the "create an audience" first? I think getting visibility in a crowded market it's the most difficult part.
On top of that, it's a common mistake to create business or products without domain knowledge. I have ideas to build stuff but all of them are software engineering related and are hard to build specially in terms of time.
Anything that you can think of right now and think "there should be a service for this", already exists or did exist at some point. If it doesn't exist there are good reasons it doesn't exist. It's either far too difficult to actually implement, or has limited monetization potential due to said difficulty or size of user base. If you can somehow bridge that gap then there's potential. Do you have the niche knowledge necessary to bridge that gap? It's definitely possible if you do. I've seen several folks that have like ~50k followers across all social media platforms making a quarter million per year selling courses and the like. But it did take them ~5 years to get to that point.
I had an idea to find a way to monetize offline twitch chat as the amount of times I've seen said offline chats jump on something and actually spend money was way too much for there to be nothing there. Plus there are close to a thousand of them with a sizable user base (~25k). Essentially a hyper targeted ad platform you can sell to ad agencies/coordinators to get a cut of revenue. Issue is the data that is there ended up being fundamentally useless (similar with google stuff), with nowhere near enough users to make those razer thin margins worth chasing. But I did gather a ton of stats on the user base which led to the further conclusion that the overwhelming majority are flat broke hence why running ads on twitch is as close to a waste of money as you can get. Most are high school/college folks/unemployed. I've seen said unemployed people get a job and the time they spend in an offline chat plummet by 99%. Neat idea but fundamentally zero money in it. Maybe if I could cut costs down by a factor of 20x it would be worth it, but that takes not insignificant upfront investment. Again have to find a gap you yourself can fill that nobody else thought was worth pursuing.
I think the "create an audience first" advice is mainly aimed at people who are trying to avoid having to spend money on development costs for ideas that aren't going to make back the money they spent on development.
For a non-technical founder, the ability to pivot to a different business case often involves starting over from scratch (or nearly so), causing a lot of wasted dev time.
For a technical founder, the incremental cost of pivoting is much less (in my experience), because we can understand what parts of the stack we can reuse, and what the minimum marginal effort of pivoting from one failed MVP to the closest possible MVP in the new direction.
The other issue with us is, if you do the "put up a marketing site first, and don't start on development until you think you found a good niche" thing, you're likely to spend a ton of time iterating on marketing copy instead of programming, from which you won't learn anything about being a better programmer.
"Create an audience" first doesn't work for everyone. IIRC "create an audience first" as a dogma started when Joel Spolsky advertized Trello to his followers (I don't recall if it was a newsletter or Twitter or both) and he got loads of signups overnight because of it, so then people turned this into a myth and dogma. Also, the domain gotta support writing a lot about it, so if your project is a tool that does one simple job extremely well (like generating backgrounds for websites) - it's hard to build an audience for this topic.
The reverse is also possible - think of all celebrities attaching their names to random things like Rihanna starting a cosmetics brand etc.
Highly recommend the book "Traction: How Any Startup Can Achieve Explosive Customer Growth" by the DuckDuckGo founder.
Thank you for this video porgrammer my american programming sensei
Thanks! Could you talk about >self ideas invest< ? Nothing of quality content related with this ideas and your perspective is brilliant and simple.
I'm not sure what you mean by ">self ideas invest
This is realy excelent. Great experience on show here - and even greated advice.
Great video, solid message. I'm curious about the point on AB testing tho.
Any analysis without a big enough sample size will likely be underpowered. The approach that VCs use as explained in this video is to run adds which gets a large sample and you can hopefully use to iterate.
I'm curious what your thoughts are when doing this without a VC. I don't understand the "it takes time" message in the video
VCs spend money to get a lot of clicks on a site in a short time to get a large sample size.
Without that advertising, you can still get people to come to your site (through SEO, through posting links on Social Media, through mentioning the site in blog posts or in interviews, etc), but you'll get a lot fewer people that way. So, instead of getting enough traffic for an A/B test to be significant in a few hours with ads, it might take weeks to drive enough organic traffic to get decent data.
Getting a VC round is a full-time job, and in this market - it's an impossible job without revenue and/or a proven track record and connections. One of the links @InternetOfBugs posted was to Microconf - it's a microsaas community where most (all, really) founders are bootstrapped, i.e. they already built and launched something. Rob Walling created this community not only to have a community of bootstrappers and share tips and knowledge but also it's his vehicle to find companies to invest - he invests smaller rounds and it's more "friendly" than the typical VC route.
With regards to testing a hypothesis - there multiple ways, you don't have to build a page and run ads or attract traffic. You can hang out where your audience hangs out and ask questions. Even places like Reddit (just don't violate the etiquette). Also highly recommend reading the book "The Mom Test: How to talk to customers & learn if your business is a good idea when everyone is lying to you"
Thanks. It was a great video.
That's one of the best intros ever lmao
Fantastic content!
Subbed let's keep going~
Can you please make a new video on the findings that inference, scaling, and synthetic data are actually improving LLMs.
Take for example gpt-o1, we have not even come close to maxing out the abilities of LLMs alone, assuming no more AI breakthroughs I do not see how this will not disrupt everything.
Beyond improving reasoning ability, we can make them more efficient too, so they are cheaper and integrated into everything.
I think its time all the people who buried their head in the sand accept their fate.
Thank you for this
Genuinely knowledgeable programmers are actually pretty rare in the grand scheme. Most people would Google and then copy pasta. AI has really only made that problem worse, but it was already there in the culture before AI came along.
I agree. But, in my opinion, the best way to get more knowledgeable is to build things. Most developers won't ever do that, but the ones that do tend to differentiate themselves from the pack that way.
It’s never been easier to start a Software Project but it’s also never been more competitive. Don’t go into it with the expected outcome of income, you’ll (probably) be really disappointed.
This really appealed to me because the end goal was never to work for someone. It is to be so useful to others, that they have no choice but to pay me. Hence why i never labeled myself as a "developer". I'm a problem solver.
I'm not a dev really but I'm getting into coding. What blows my mind is how "developers" are so obsessed with going to work at some shitty soul sucking corporation. It sounds like jail to me. How about you have fun with building and build something you want to see in the world? I'd rather put a fork in my eye than have to deal with 1% of the corporate world.
sir make a video on new model from openai
Do you think self taught devs have a place in the current economy, or a degree is the way to go. I would love to hear your opinion
I did an interview about that. ruclips.net/video/f9bO9aTXog0/видео.html
@@InternetOfBugs thank you sir
Doesn’t pivoting quickly and repeatedly lead to low quality, just-enough-for-now, kind of programming?
I love the idea of doing side-projects, but I don’t see how it helps getting better at programming on the technical side.
It leads to embracing the YAGNI principle. Also, would you rather find what to build by pivoting and iterating rather than building something that nobody needs but with high quality code?
I want to grt better and understand what needs to happen to reduce the bugs and make things work, how?
You are wrong.
I can point out many samples from the video, but the easiest is the books you recommend.
TARGET AUDIENCE FOR The Lean Product Playbook are
Product Managers
Startup Founders
UX Designers & Engineers
Business Analysts
Agile Teams
TARGET AUDIENCE FOR The Lean Startup are
Entrepreneurs & Startup Founders
Product Managers
Business Leaders
Investors
Innovation Teams
Which book you think is for software engineers ?
I know you don't know, but none of them :)
That mug move? Smooth af
LOL
I saw the title and thought, isn't that the guy who I've seen criticize ai stuff? this must not be a trick, I honestly do want to do something like that, specifically in fiction writing, preferably without AI but i'm only the first minute in so lets see what you have to say
I try to be balanced. There are things that AI is awful at (primarily things that need to be "correct" or have a right answer) and things it's helpful for, which to me is stuff where there is a huge range of potentially valid responses that vary in quality - like marketing copy or stock images.
What do you think about openai o1
Hoping to take a look at it this week. Stay tuned.
@@InternetOfBugs waiting for your view on it thanks a lot 👍
I used it yesterday, it's better than Claude but it's still not blowing me away.
The problem with these things is you can ask them to do a task that is nonsensical and it will attempt to do it. For example, ask it to create a property decorator for a class in typescript that initializes the property outside of the constructor. It will do the task eventually as you work with it, but this particular task achieves nothing because properties are already initialized outside of the constructor lol.
I would expect models to be able to detect BS tasks like this and say: this is a bad idea and it shouldn't be done this way.
I've also discovered something else related to this: if you ask a question about why it's doing something a certain way, it will immediately "agree" with you and try to find a different way to do it, even though you never even said that something was wrong, in fact there may have been nothing wrong.
The idea that these things will work autonomously seems like complete nonsense to me.
Agreements you sign with employers where you seemingly sign over your creative life are highly unlikely to stand up in court unless the product is in direct competition with your employer, and was produced on their time. Most employers won’t care anyway, and for those that might, there are PLENTY of ways to hide things ahead of time. Be careful, but I would not let this discourage you.
It's not about it standing up in court, it's about getting bled dry by the fight. Even if you win, you've lost.
@@InternetOfBugs understand that can happen sometime, but it would be a very rare occurrence, especially if you’re working on things unrelated to the LOB of your employer. My agreement specifically indicates the work must be related to said business. At any rate, I wouldn’t let that hold you back, as there are many, many ways to make sure things are never found out. Large employers are unlikely to waste time and resources on it, again, unless it’s a serious direct threat.
Thinking of an idea that is not already offered is really hard, IMO. I'd love to do this, but I'm not sure i have any novel business ideas. I'm just a second-rate coder, then, I guess. 😢
i am struggling with the same idea
OK, I have an idea now. It's ridiculously nichey, but it would help improve an annoying process in my field of work.
A few weeks ago I was feeling some kind of way because I felt like I was failing to come up with some "novel" or "good" idea. After listening to some other sources I decided to just lean into solving my own problems, so I made a free chrome extension to fix an issue I had at work.
It wasn't very complex and I almost felt silly for publishing it in the Chrome Web Store. It's not a lot, but in less than a month I got 5 users and a 5 start review with someone calling it one of the best inventions he's seen.
I have two more ideas for chrome extensions, one paid and another free one, so I'll see where that takes me.
These MBAs need competition
where is the video about membership?
It's here: ruclips.net/video/l9Z7KPnP0-c/видео.html
It should be linked on the end-screen of this video (at least if the platform you're on has support for end-screens).
I forgot to put it in the description didn't I?
Thanks for pointing that out. I'll go fix that.
But surely now with o1 out it's just a matter of days before all our jobs are replaced. It's as intelligent as a whole phd I've heard. I am not as smart as a phd. So my career is over. Matter of fact I have seen somebody make flappy bird with it using just one prompt, in 5 minutes. Who can compete with that? Anyways. My current Sideproject is Coding Snake in x86 Assembly running in Dosbox. I don't think it will get me a lot of job offers, but it is a lot of fun.
"if i gave you $100k and six months, what would you make?" Free software. How does that make you feel?
Depends on the software. It's not about business acumen or profit. It's about thinking through a problem.
@@InternetOfBugs you, specifically, aren't the one I'm worried about lol.
Hello, saw your podcast where you talked about finding out that you have ADHD, can we have a video on that please?
That's an interesting question. I wouldn't mind doing that, but it's so different than my normal content, I'm guessing a lot of people wouldn't like it, so it would penalize the channel's stats.
Let me see what I can figure out
Where have you been all these years
The best part of starting a business is, you have no choice but to learn to identify BS vendors aka Grifters and Shills. If you do not gain anything technically (unlikely), you would gain a lot of life wisdom.
Clark Jessica Hall Sarah Gonzalez Kenneth
This info is good but we don't need to hear you drink water. It's OK to edit out pauses.
Don't make your channel beholden to youtube analytics. Or AB testing for that matter. If you go down that path all the way, you end up creating Mr. Beast videos with your mouth open on the thumbnail. Your strength IS the technical deep dives, so lean into that, instead of shying away to maximize views.
I"m not smart enough to start my own business.
AI will replace programmers.