It may sound weird, but it's based on my 20y observations in the industry (including teaching noobs).
Given the goal, you don't want to bee strategic about learning technical skills.
It is a botomless pit - the more you dig for the sake of improving a skill - the more there is to dig. And the worse you feel about yourself, the more unrealistic your goal seems.
The best thing I would do in your place is stay focused on the business goal - i would try a couple of courses, books that is think would get me to shipping some specific product in mind ASAP. You don't have (and even better not) to be a good coder to launch a product.
You only need to be good enough to launch and get your first client. Then the users/market would hint you specific directions and highlight your gaps pretty well.
If your product is complex - I would dumb it down to some school class grade prototype and attack it first.
When you just want to play a cool song on a guitar — you learn the song. You don't spend years of your life becoming a decent guitarist, unless this (or procrastination) is you real goal.
Specific piece of advice: Ruby on Rails is your friend.
All batteries included, noobs friendly, time saving framework designed for one-man shows / quick prototyping. Yet mature enough to scale with you out of your prototyping phase.
Countless of startups has been started (and most continued) as Ruby on Rails web app.
Even though I code for a living and mostly enjoy the process, I’ve spent the last few nights with Cursor, migrating my old C# project to Python. I decided to do it without providing my old code to the tool and without coding at all. Surprisingly, it was amazing—just like having a super-efficient, eager programmer at your service. In just a few sessions, I managed not only to rebuild my old app but also to add many new features that would have taken me days to develop. (I decided to use PyQt6 for the GUI, despite having zero experience with it.)
So, if you know how to run code and understand the absolute basics, just start with Cursor. The free version should be good for a start, and then it’s just $20 per month. You can learn in the meantime by asking questions, etc. However, I would focus on understanding high-level concepts because mundane coding seems to be a solved problem now.
It’s a bit sad from a coding enthusiast’s perspective, but on the other hand, it makes you super productive when you have an idea for an app.
Realistically very few solo developers are able to make sellable products.
Even fewer make a net profit. You're probably going to lose money here unless you want to work for a corp.
I'd go with Python or JavaScript. It depends on your first project. Python is basically very good with data science and okay at everything else. JavaScript is almost the only way to build a website and okay with other things.
Write something simple. Like a sales tax calculator with Python. Then slowly build yourself up to other more challengingly things.
Start with something that already does mostly what you need and you can tweak or fix the way you like. (Really everything is content management, data capture, and workflow right?)
As you grow in skill and familiarity you will add features and configure more pieces to work together.
After all of this (in years) you will hate everything you’ve done and have had to do to get things working (and they have.) You will dream of a day when you can scrap it all for a new from scratch exhibition of all that you have learned.
Someday you will.
Until then, just hobble things together so they work in the environment you are most confident in, like everybody else.
There is an underlying question to answer for yourself - do you really want to learn to code, or do you want to learn to launch products on your own? As you said, AI is changing, as are the skills required to be a one-man software shop.
I would not waste time with today's prompt engineering. The results are iffy, and the tech is evolving. I'd focus more on understanding how complex apps are structured. When AI hits its next big inflection point, it will be able to handle everything shy of the actual architecture of your app. I'd focus skills on being able to lead a team of coders, and then you can do so whether they are digital or human, expanding your capabilities in tandem with AIs improvements.
I'd also focus on determining the right use cases for AI. Right now, people are throwing it at everything, including processes and problems that need deterministic answers. LLMs are non-deterministic, so some people are not using them correctly yet. They are trying to pound in nails with screwdrivers.
All that to say that if I were you, I'd get good at understanding the current AI capabilities (and not limited to LLMs), applying it to the correct product and dev features, and being able to lead both AI and humans through the gaps, selecting the correct team member to solve the problems which cater to their strengths.
codeclub.org Python. Pulling data from APIs and showing it. Using pyautogui to do silly automation. Reading about denial of service attacks from cloud flare blog posts, then googling "send UDP packet python", sticking it in a for loop and a while loop, and crashing the router. Circumventing the school firewall in increasingly elaborate ways.
Subscribing to every computer related subreddit possible and learning through osmosis.
Building things, breaking things, and learning as I go. It's been good fun.
There is exactly one way to learn programming, just like there is exactly one way to learn a language. You need to do it, you need to actually program things, you need to want to make something.
Here are some things which you should not worry about:
- Which language
- Which editor
- Which course
- Which book
- Which library
- Which operating system
LLMs are also very good at answering questions and explaining things. If you don't know something ask an LLM.
I'd ignore advice like "Ask ChatGPT to make you python course". Get a high-rated course on udemy for $10 where they are building projects/products, get Think Python, and build something on your own later. Bonus if you can get it deployed.
Find the top rated short book in the language you're interested in. If you don't know which language, since you eventually want to get into AI, Python probably would be best bang for buck. IMO the more pictures, the better. Also, be aware that sometimes setting up the environment for coding can be half the grunt work, and to avoid that, often something like using Google Colab works just fine. [0]
2. If you have some coding knowledge and want to improve that first before thinking about AI? Here are some tricks I would try that I think are less common:
a. Find a youtuber who meets these 3 criteria:
- 1. is a great coder
- 2. is a great communicator
- 3. has posted videos of them solving the Advent of Code challenge series while they think out loud about their reasoning and solving.
Simultaneously try to solve the AoC problems yourself, while also typing out their solution. Observing tacit knowledge is very effective for skill transfer.
Pause the video whenever you think you know the way forward etc. so you get a chance to try yourself.
b. Get very used at quickly digging through codebases. Best way to do that is to find a lot of codebases to click through.
Github is pretty good for this and the code browsing feature actually lets you jump through code reasonably well enough to traverse codebases just by clicking on any symbol that looks interesting to jump to its usages or definition.
Look for anything that sounds cool on Github (browse trending repos etc), and try to figure out how it works. This is a quicker way to introduce yourself to a variety of patterns without having to whip everything from scratch - you can learn those planning skills at a later time.
3. For getting into AI, once you feel comfortable with coding generally, try the Deep Learning For Coders youtube lecture series from fast.ai
[0] I just point this out to give you a heads-up that basically it's quite easy to run into situations where for some strange reason if you just keep installing stuff, things can get borked, and trying to solve that saps energy and educational momentum
I'd start with ESP32 if I had nothing else to do. Get two of them working with each other and convince some neighbors to join a neighborhood mesh network.
C# using Visual Studio for APIs. Deploy to Azure. Simple, plenty of help from AI. Not sexy but you are trying to get things done. Stay away from MAUI...far away. Flutter is a big chew but the AI assistants can barf out working code easily to get you started.
> launching my own products
It may sound weird, but it's based on my 20y observations in the industry (including teaching noobs).
Given the goal, you don't want to bee strategic about learning technical skills.
It is a botomless pit - the more you dig for the sake of improving a skill - the more there is to dig. And the worse you feel about yourself, the more unrealistic your goal seems.
The best thing I would do in your place is stay focused on the business goal - i would try a couple of courses, books that is think would get me to shipping some specific product in mind ASAP. You don't have (and even better not) to be a good coder to launch a product.
You only need to be good enough to launch and get your first client. Then the users/market would hint you specific directions and highlight your gaps pretty well.
If your product is complex - I would dumb it down to some school class grade prototype and attack it first.
When you just want to play a cool song on a guitar — you learn the song. You don't spend years of your life becoming a decent guitarist, unless this (or procrastination) is you real goal.
Specific piece of advice: Ruby on Rails is your friend.
All batteries included, noobs friendly, time saving framework designed for one-man shows / quick prototyping. Yet mature enough to scale with you out of your prototyping phase.
Countless of startups has been started (and most continued) as Ruby on Rails web app.
Even though I code for a living and mostly enjoy the process, I’ve spent the last few nights with Cursor, migrating my old C# project to Python. I decided to do it without providing my old code to the tool and without coding at all. Surprisingly, it was amazing—just like having a super-efficient, eager programmer at your service. In just a few sessions, I managed not only to rebuild my old app but also to add many new features that would have taken me days to develop. (I decided to use PyQt6 for the GUI, despite having zero experience with it.)
So, if you know how to run code and understand the absolute basics, just start with Cursor. The free version should be good for a start, and then it’s just $20 per month. You can learn in the meantime by asking questions, etc. However, I would focus on understanding high-level concepts because mundane coding seems to be a solved problem now.
It’s a bit sad from a coding enthusiast’s perspective, but on the other hand, it makes you super productive when you have an idea for an app.
Realistically very few solo developers are able to make sellable products.
Even fewer make a net profit. You're probably going to lose money here unless you want to work for a corp.
I'd go with Python or JavaScript. It depends on your first project. Python is basically very good with data science and okay at everything else. JavaScript is almost the only way to build a website and okay with other things.
Write something simple. Like a sales tax calculator with Python. Then slowly build yourself up to other more challengingly things.
Start with something that already does mostly what you need and you can tweak or fix the way you like. (Really everything is content management, data capture, and workflow right?)
As you grow in skill and familiarity you will add features and configure more pieces to work together.
After all of this (in years) you will hate everything you’ve done and have had to do to get things working (and they have.) You will dream of a day when you can scrap it all for a new from scratch exhibition of all that you have learned.
Someday you will.
Until then, just hobble things together so they work in the environment you are most confident in, like everybody else.
There is an underlying question to answer for yourself - do you really want to learn to code, or do you want to learn to launch products on your own? As you said, AI is changing, as are the skills required to be a one-man software shop.
I would not waste time with today's prompt engineering. The results are iffy, and the tech is evolving. I'd focus more on understanding how complex apps are structured. When AI hits its next big inflection point, it will be able to handle everything shy of the actual architecture of your app. I'd focus skills on being able to lead a team of coders, and then you can do so whether they are digital or human, expanding your capabilities in tandem with AIs improvements.
I'd also focus on determining the right use cases for AI. Right now, people are throwing it at everything, including processes and problems that need deterministic answers. LLMs are non-deterministic, so some people are not using them correctly yet. They are trying to pound in nails with screwdrivers.
All that to say that if I were you, I'd get good at understanding the current AI capabilities (and not limited to LLMs), applying it to the correct product and dev features, and being able to lead both AI and humans through the gaps, selecting the correct team member to solve the problems which cater to their strengths.
The way I first learnt.
Scratch from a young age.
codeclub.org Python. Pulling data from APIs and showing it. Using pyautogui to do silly automation. Reading about denial of service attacks from cloud flare blog posts, then googling "send UDP packet python", sticking it in a for loop and a while loop, and crashing the router. Circumventing the school firewall in increasingly elaborate ways.
Subscribing to every computer related subreddit possible and learning through osmosis.
Building things, breaking things, and learning as I go. It's been good fun.
There is exactly one way to learn programming, just like there is exactly one way to learn a language. You need to do it, you need to actually program things, you need to want to make something.
Here are some things which you should not worry about:
- Which language
- Which editor
- Which course
- Which book
- Which library
- Which operating system
LLMs are also very good at answering questions and explaining things. If you don't know something ask an LLM.
I'd ignore advice like "Ask ChatGPT to make you python course". Get a high-rated course on udemy for $10 where they are building projects/products, get Think Python, and build something on your own later. Bonus if you can get it deployed.
Depends on where you're starting from
1. Absolute zero?
Find the top rated short book in the language you're interested in. If you don't know which language, since you eventually want to get into AI, Python probably would be best bang for buck. IMO the more pictures, the better. Also, be aware that sometimes setting up the environment for coding can be half the grunt work, and to avoid that, often something like using Google Colab works just fine. [0]
2. If you have some coding knowledge and want to improve that first before thinking about AI? Here are some tricks I would try that I think are less common:
a. Find a youtuber who meets these 3 criteria:
- 1. is a great coder
- 2. is a great communicator
- 3. has posted videos of them solving the Advent of Code challenge series while they think out loud about their reasoning and solving.
Simultaneously try to solve the AoC problems yourself, while also typing out their solution. Observing tacit knowledge is very effective for skill transfer.
Pause the video whenever you think you know the way forward etc. so you get a chance to try yourself.
b. Get very used at quickly digging through codebases. Best way to do that is to find a lot of codebases to click through.
Github is pretty good for this and the code browsing feature actually lets you jump through code reasonably well enough to traverse codebases just by clicking on any symbol that looks interesting to jump to its usages or definition.
Look for anything that sounds cool on Github (browse trending repos etc), and try to figure out how it works. This is a quicker way to introduce yourself to a variety of patterns without having to whip everything from scratch - you can learn those planning skills at a later time.
3. For getting into AI, once you feel comfortable with coding generally, try the Deep Learning For Coders youtube lecture series from fast.ai
[0] I just point this out to give you a heads-up that basically it's quite easy to run into situations where for some strange reason if you just keep installing stuff, things can get borked, and trying to solve that saps energy and educational momentum
> Find a youtuber who meets these 3 criteria: > - 1. is a great coder
This never happens (i'm happy to be mistaken, please share example).
Great coders are busy being coders. Youtubers are busy being youtubers, different game they playing.
What you mean I suppose "is (was) real professional programmer", that and #2 is a good enough mix for most of the noobs to start.
> transition into launching my own products
Programming is glorified building work. The building you create is the product. Work out what you want to create and then find what skills you need.
Making something you believe in is by far the best way to learn, albeit sometimes quite painful.
I'd start with ESP32 if I had nothing else to do. Get two of them working with each other and convince some neighbors to join a neighborhood mesh network.
C# using Visual Studio for APIs. Deploy to Azure. Simple, plenty of help from AI. Not sexy but you are trying to get things done. Stay away from MAUI...far away. Flutter is a big chew but the AI assistants can barf out working code easily to get you started.