As a software developer I was appalled by their answers. The last answer in particular predicts a very lonely future.

I think we have seen in the last 12 months what AI can do in software. You can produce an app now in a few days and then have it rolled out to users.

I think in 12 months WE (as in we humans) will be set tasks by AI as they can go through data when we sleep and come up with tasks that need doing.

AI is making me so productive that I don’t need to hire nearly as many people now. Companies in the future will be much smaller.

I don’t think we’re that far away from the first “$1 billion single man company”

stirred up a whole load of emotions inside me

. This is a recurring theme for me nowadays. In this article I will deep dive into each statement and give my thoughts.

I am seeing a sentiment, especially in management and people who aren’t in the weeds of Software everyday, that AI can essentially code now. 

 and the like spouting rubbish that all startups in Silicon Valley are now vibe coding their apps in record time.

Building a crappy little MVP in Nextjs with Tailwind

that looks like it does the thing you want is not the same as building a scalable piece of reliable software that can really support a business.

But there is a growing gap between what business people THINK AI can produce and what it actually can do in reality.

This isn’t helped by every AI model company CEO telling the masses that AI can produce 60, 70, 80 or even 100% of your code and everything will be hunky dory.

Let me tell you, as a software developer who has been using these Generative AI tools from day one, it‘s nothing like that.

AI is a rampant text generator and fantastic pattern matcher. It seems to know everything right up until the point when you tell it that it made mistake. Then it tells you 

 like you’re a god damn genius and then continues to churn out more text. 

(Oh and I am saying the word “text” deliberately because thats ultimately what it is)

Have you seen the way AI generators produce code? Top to bottom.

 No developer ever writes code top to bottom. You write it in logical steps and jump around a file to fill in the details as you work through the problem.

The AI is not thinking. It has no intelligence. It just spouts out patterns it’s seen over and over again. Don’t get me wrong, it’s an incredibly sophisticated pattern matcher and it’s remarkable how far it can get with this skill, but it’s certainly not ‘thinking’.

This one depressed the hell out of me. AI is going to become our task master? We’re literally going to become slaves to the machine? The person stating this fact seemed quite happy about this future. I was in disbelief.

A task master is a person who imposes a harsh or onerous workload on someone. I wonder if AI will be a hard task master?

Are we really pursuing the pinnacle of all technology — AGI — at breakneck speed just so it can increase our already bloated TODO list? That seriously can’t be it. Right? Right?!?

This is very topical right now and depending on 

I don’t think it’s quite as obvious as some people like to make out but whether you like to admit it or not, 

AI is definitely having some impact in the work place

From my own experience and speaking to people within my social circle I can see pretty much everyone is using generative AI to some extent. The feedback I hear varies greatly but regardless, it’s being used.

That’s not so easy to answer because it can be very subjective. Just because you can create more words in a given time doesn’t necessarily mean you are being more productive. What about all the extra noise that is being created now? Summaries after summaries. Notifications galore. All the extra code that is now being produced? 

Is this a productivity shift or just more slop?

From what I have read and understood, power users of AI are seeing tangible benefits but for the average Joe, not so much.

software world that the interviews now expect AI to be used

 as part of coding challenges and recruiters want to know how you are fitting AI into your workflow.

Will companies be much smaller? I’ll get into that in the next section.

Whereas AI becoming our Task Master depresses me, this prediction saddens me.

Is this really what we are aspiring to as a society? To work alone with our AI agents to produce extraordinary wealth for the individual? Is this really the dream for an up and coming tech entrepreneur?

To work alone in a room building up a huge pile of cash

 to ultimately do what? To sit alone in your huge new mansion surrounded by your material wealth anxiously checking you bank balance on your phone every 5 minutes?

How is this even a goal for a human being? Is this how far we have fallen as a society?

I know we hold aloft the star individual. The clinical CEO entrepreneur who created a transformative company and changed the ‘

’ with their incredible latest innovation. We all aspire to be the next ‘big thing’ and we’re learning how to reduce everything to ‘first principles’. But are we now aspiring to do all this alone?

Please lets stop this madness and learn that this journey we’re on should not be taken alone.

The future of AI is extremely difficult to predict. The tools we’re using everyday and constantly evolving and changing. Things you weren’t even dreaming of 6 months ago are now a reality.

Having said that, some of the predictions I heard at this conference really shook me in a negative way. It all seemed so… bleak. 

Yet again we are building a technology to isolate ourselves and promote the individual.

 To push more power into the individual and hold that aloft above everything else.

Why do we continually do this as a society? Why do we think our problems will be solved once we are all isolated and ‘rich’? I sincerely hope we don’t go down this road but as history likes to show us, we inevitably will.

The Lonely Future We’re Building With AI

Most developers feel somewhat uncomfortable when asked, and why wouldn’t they. It’s a tough thing to go through. 

It’s a whole load of extra work, your skills will be under really close scrutiny and you’re not guaranteed to get anything out of it.

I’ve been on the other side of coding challenges for a while now. I’ve had the pleasure of 

. I’m always on the lookout for a few things. It’s nothing surprising either.

Yet I am still amazed how many people simply don’t cover the basics.

Here is a list of things I personally look for when reviewing code 

(It’s frontend/React related but the tips still apply to any coding challenge).

I’m interested on your opinions on the last tip (AI related)

. I share this in the hopes that it could help you succeed on your next coding challenge. Or at least do slightly better.

This may sound obvious but about 50% of the code that I review doesn’t work on startup.

Issues range from the likes of environmental variables not being set, docker containers configured incorrectly, an import missing, a type error… 

In these challenges you want to get off to a good start. If the reviewer has to start debugging right away when using your app then you’re already in trouble.

Tests have a couple of purposes. They make sure a piece of code does what it’s supposed to do under different conditions. 

But it also gives new developers to the codebase a much better idea of what a certain chunk of code is supposed to be doing

By writing tests in your submission, not only are you verifying your code is working you are also helping the reviewer see what you think the code 

At least have some unit tests in there. Use 

. If you have time throw in some e2e tests with 

. Nothing overly complex but cover the basics. With AI these days, this is a no brainer.

It shows you don’t just build something and hope for the best. 

You are the type of developer that likes to put the odds in your favour as much as possible

 and drives me crazy in code bases. Why is this code commented out? Will it be needed in the future? Did the developer forget to uncomment it? Did they forget to delete it?

It just raises questions and shows a lack of care about the codebase you are working on. 

Don’t do it. Uncomment it and use it or delete it.

I received a code submission to review recently for a React app. All the code was declared in the App.tsx file. 17 declarations of useState , svg icons, handler functions. You name it, it was all in that one file.

Not only does this make a long file and make it difficult to read, it also doesn’t scale well at all. If you want to reuse one of the icons, you can’t without some refactoring. If you want to use a piece of that state somewhere else in the codebase, you can’t without some refactoring.

I would argue whatever codebase you are working on, build it out with the likelihood that it will scale in the future.

Calling an api? Put that in a separate function in a services folder. This way if another part of the app needs to use the same api in the future it can just import this function and be done. Have some piece of state that might be needed somewhere else in the future? Then put it in some context or a global state library like Redux or Zustand.

The point is apps naturally grow over time, and you want to show you are thinking ahead. 

You don’t want to be refactoring every time you want to add a new feature.

 Show that you are the type of developer who takes this into consideration.

This connects in some way to the last point. 

Don’t put all your logic in one big function.

 Don’t combine clearly separate parts of the data.

If you’re going to use a global state library, don’t put everything into one big piece of state, separate it out. Maybe you have users and products. Deal with the logic for handling state updates separately. Make it clear what are the “groups” of state that are important in this app.

I’m not saying you have to go down a functional programming rabbit hole but separate logic out where it makes sense and show the reviewer this is something you care and think about.

The reviewer’s time to review your code is likely to be short. You want to make it as easy as possible for them to understand your code and how to get it up and running.

You can even elaborate here stating your approach, why you made certain decisions and what you think is the most interesting part of the submission to focus on.

Don’t write a novel but put yourself in the reviewers’ shoes for a moment and write what would be useful when viewing your code for the first time.

We live in 2025 and we have established tools to check both linting and types. Use them!

 I am still amazed how many different code bases I see where the existence of basic linting or types is nowhere to be seen. 

Use Typescript btw, not plain Javascript.

Unnecessary errors due to lack of typechecking should not be occurring.

This is just basic 101 stuff for a developer in this day and age. Reviewers shouldn’t be having trouble reading your code due to formatting. I would advise to use 

 as it’s the goto library most teams use these days.

Remember, you are applying for a job to work in a team and having these basic tools in your workflow really help a team work together. You want to show reviewers you care about this stuff.

If you have the luxury of time then I always suggest you wait a day between when you completed the coding challenge and before you submit. This gives you time to take a breath and then review the code one last time.

I can guarantee you will find something that can either be fixed or improved.

When you are in the middle of doing the code challenge you are concentrating so much on the challenge it’s easy to miss some details.

By giving yourself that little gap you can come back with flesh eyes, do a little clean up and make sure you didn’t miss anything obvious. These small tweaks in the clean up phase can make all the difference.

This is probably the most relevant at the time of this writing. More and more companies are allowing or even requesting you use AI during your code challenge.

It makes total sense as in your average job these days, a developer is making some use of AI.

The important thing here though is transparency of usage. It can be very revealing to the reviewer how you are thinking by the questions/prompts you ask the AI.

Be honest with the reviewer which prompts you used, why you used them and what you did with the results.

The reviewer wants to gauge as much as possible how you will be using AI were you to get the job. Are you asking basic things you should already know? Are you offloading everything and vibing your way to a result? How clearly can you express yourself? 

How good is your technical communication?

There are many aspects to a successful job application and plenty of those aren’t related to coding and are outside of your control. However, when it does come to things you can control you want to do your best to put the odds in your favour.

Covering the basics in the coding challenge is one of those things. I’m not saying by covering the basics you will pass this part of the application but I am saying if you don’t, then it could very easily eliminate you.

All these practices are good practices to have in your day to day job anyway. These are things that a company expect from future employees. Cover your basics.

Stop Making These Coding Challenge Mistakes (From Someone Who Reviews Them)

Who’s Zed? Zed’s dead baby. Zed’s dead. — Butch and Fabienne — Pulp Fiction

Luckily for us, Zed is definitely not dead. Zed is alive and well and seems to be thriving.

The trusted Code Editor is at the centre of any developer’s workflow. Some obsess over it like a frenzied DJ constantly twiddling knobs whilst others simply use the default settings with whatever editor seems to be most popular within the community.

, which has been great, especially with their AI integration. It’s been the smoothest experience when it comes to getting into a flow with an AI assistant.

 on it recently. I read the comments on this video and knew I had to give it a try.

I Switched from Cursor to Zed for a Week — The Results Surprised Me

Once you have gotten over all the emotional trauma this shift in the industry is causing then lets jump in to some of the more positive sides of using AI on the daily when working with Software.

Let’s be honest a second. How many times did you have the opportunity to write some tests when working on a feature but chose not to because of “time constraints” or “just going for the MVP”? 

How many times did that come back and bite you?

Nowadays, there is no excuse not to write tests.

For unit tests especially, generative AI in Cursor or Claude Code is an absolute god send.

 “Please write unit tests for this function @functionNameGoesHere and test edge cases” and BAM! 

 Edge cases accounted for. All tests have clear descriptions and titles. All in a couple of minutes.

Let’s Talk About the Positives of AI in Software

 — maybe even created it? Not sure… google it) and he was talking about the way he had updated his keyboard layout to avoid “finger gymnastics” when typing code. 

I suddenly became very aware of the weird and wonderful contorted movements I was doing with my hands on the keyboard when I was coding.

Holding ‘shift’ whilst twisting my hands and wrist slightly to reach the ‘curly brackets’. Stretching a little more to reach the ‘normal brackets’. Hands jumping all over the keyboard just to write a small function.

I instantly realised I was using my main set of tools — 

 (yes and my brain… I get that’s the most important tool but lets not get bogged down in semantics)- in a horribly inefficient way.

Switching to Vim Motions in Cursor Completely Changed How I Navigate Code

It also makes me highly skeptical of those “Gurus “ online who tell you it’s so easy to make money on a side hustle. Maybe it is for some people but it certainly isn’t for me.

Recently though, and maybe a consequence of getting older, I am settling into a stack that I think gives me everything I need for a personal web project. That stack is:

These are the main reasons I like this stack:

The Frontend uses React which I have used for many a year. I really like React and I can be pretty productive in it. 

 gives you all that React goodness plus all the Tanstack goodness!

My New Favourite Stack — Tanstack Start, Laravel, Postgres and Mantine

I have seen AI as more of another tool to use in a developers’s toolkit.

 It can be a very useful tool in some circumstances but a tool none the less. Therefore, I like to make small incremental steps when introducing any AI into my workflow.

. I have been quite impressed so far and think it can help in my existing workflow and make my daily developer life a little better.

The README is really well documented and it was easy to get it setup in Cursor. I got it connected to our Jira board that we use at work all within a few minutes.

Using the Atlassian MCP server in Cursor

However, over the last couple of months it has been dawning on me that if you are a experienced, seasoned or senior developer (whatever you want to call yourself) you are in a beautiful position to utilise this amazing new tool that has forced itself into our everyday lives. I want to make the case that AI is not a threat to our jobs but a big boost. However, (and it’s a big HOWEVER) it all depends on how you use it.

Senior Developers Have The High Level View

 from the talented TJ DeVries and he talks about using AI to learn. Himself and (The) Primeagen were building an app with Laravel, which they both didn’t know, and found the LLM predictor to be very useful. It was providing patterns for them in a language they weren’t familiar with and they took it 

This is the massive advantage that experience Developers have right now. They know enough theory, patterns and programming fundamentals that they can quickly determine if the output from the LLM of choice is a Hit or Miss. They also know the right questions to ask or right prompts to give which I have found to be very important and not spoken about much.

It’s the Perfect Time to be a Senior Developer

The persistent inability to believe that one’s success is deserved or has been legitimately achieved as a result of one’s own efforts or skills.

Which seems to make sense in a highly cognitive environment where Software Developers find themselves.

You see people sat at their desks working fastidiously on some complicated UI on the monitor

I’m always amazed in office environments at how professional and serious everyone appears to be at first glance. When you go into an office for the first time, be it an interview or a new position, you see people sat at their desks working fastidiously on some complicated UI on the monitor. They’re dressed seriously, their expression is one of seriousness and concentration and there is that low level noise of keys being tapped and mouses (mice?) being clicked and 

It’s very easy to be convinced these people are geniuses who are operating at a super human level and that you have no right to be amongst these rarified supreme individuals

It’s even more compelling in an office full of developers. Their screens are much more elaborate and impressive. Screens filled with dark squares placed precisely into strategic compartments with rows of different coloured text flashing continuously across them. Terminal windows printing rows and rows of obscure text and different windows popping up here and there and then quickly disappearing again. It’s very easy to be convinced these people are geniuses who are operating at a super human level and that you have no right to be amongst these rarified supreme individuals.

As you peer closer the illusion starts to shrink away. That terminal window you thought they were using to hack into some top secret government server is actually being used to ssh into a AWS server to check the right docker image is running. That rows and rows of different coloured text is actually some HTML which is used to render the ‘About Us’ section on the company website. That unrecognised text editor one developer is using is actually just Neovim… wait! That guy is a genius! I jest… but my point is, if you understand what these developers are actually doing its all very logical and usually not so complicated.

I almost felt guilty for not knowing things and if anyone was to find out I didn’t know certain things then that’s it! The game is up.

I Rarely Asked Questions as a Junior Developer

I remember very distinctly the feeling I felt when starting out as a developer. I started with 4 other developers of different skill levels at exactly the same time. I was the developer with the least technical experience. In all honesty I think I was lucky to get the job from a technical perspective. I had other skills from my previous working background but my technical skills were distinctly lacking.

I knew this and I couldn’t shake the thought that I ‘

’ (I’ve later come to realise that ‘

’ is a very dangerous word to tell oneself). I almost felt guilty for not knowing things and if anyone was to find out I didn’t know certain things then that’s it! The game is up. My head will be on the chopping board and I would be sent packing with my tail between my legs. In order to avoid this potential scenario I adopted the best strategy available to me. 

I chose this strategy because I figured if I don’t say anything, then I can’t say anything stupid. 

If I were to say something stupid then my colleagues would quickly figure out I am a fraud and I will be exposed.

 If I didn’t say or ask anything then my colleagues will just assume I already know whatever they happen to be talking about. I will be the silent developer in the corner. The mysterious quiet one. Or so I thought.

This strategy however had one major downside. I couldn’t ask questions. It turns out (and this might come as a massive shock) that questions are really useful for finding out things. If you don’t know something and you’re speaking to a person who does know something, by asking questions you can find out what they know. Incredible right!

But choosing a strategy of silence makes the strategy of asking questions difficult. 

. I therefore had to resort to finding things out by other means. Reading Stackoverflow (it was a site used before ChatGPT to find answers to specific programming questions), searching through Google (it was a site used before ChatGPT to find links to specific websites that might provide insight to whatever specific problem you have) and watching tutorials (do people still watch tutorials?).

Looking back, maybe this actually had some benefits. I learnt to be self sufficient. I learnt how to be resourceful even in the face of seemingly insurmountable challenges. I learnt from other developers albeit through reading instead of talking.

Adopting this strategy meant I missed big opportunities to learn from those around me. This is one of the beautiful things about working in a team. You get to speak and learn directly from colleagues who will all have different levels of experience. Many would have toiled and suffered to learn certain things that they can then teach you in 5 mins. And vice versa, those things you have learnt the hard way might be just what your colleagues need on a certain project.

I am not sure exactly when the penny dropped for me, but it was after many years of developing. Maybe the fact I hadn’t been kicked out the team yet and had actually received a promotion started to alleviate fears that I was a complete imposter.

I had also read Jordan Peterson’s book where he speaks about ‘

. I’m paraphrasing but he basically said if you are ignorant and you ask questions, first off the person you are talking to can understand where your gaps in knowledge are so they can focus on those parts in their explanations and secondly, if you ask 1000 questions pretty soon you’re not going to be ignorant anymore.

This strategy does come with a caveat though. You can’t just constantly be asking questions and expect all the answers to come your way.

For some reason this really clicked with me so I decided to drop my pride and when I didn’t understand what someone was talking about I would ask questions. Even if I thought I ‘

’ know the answer. Turns out this is actually a better strategy than silence.

I started getting a better understanding of problems coming my way. I started learning things that I had been too afraid to ask about previously. I started to better understand different developers’ points of views.

This strategy does come with a caveat though. You can’t just constantly be asking questions and expect all the answers to come your way. Have everyone else do the hard work of finding stuff out so you don’t have to. You have to still put effort in and do some work to find answers yourself so then when you do have questions, it’s obvious to the person you’re speaking with that you have done all you can on your side.

An example of a bad question would be coming into a meeting and asking ‘What’s this meeting about?’ when the agenda has been clearly defined in the invite. This just comes across as you being lazy and unprepared which are not good traits in a professional working environment. Your colleagues will silently notice this.

Another example of a bad question is asking questions which have been answered or addressed in the recent past. Say you’re in a meeting and someone presenting is explaining what a particular section of code does. If you then ask 2 minutes later what that particular section of code does then it just looks like you weren’t listening. By all means ask questions to clarify certain aspects of what they are explaining but don’t just ask lazy questions because you weren’t listening or paying attention.

There is so much to know and learn in the world of Software Development and you will never know everything. To that end it’s OK to not know. It’s therefore ok to ask questions. In fact asking questions helps clarify to the other person what you understand about what they’re saying and what you don’t. It gives them hints as to what they should focus on elaborating in order for whatever is being discussed is understood by all.

Questions in team settings can also really help because if you don’t know something then chances are someone else in the room also doesn’t know so you’re doing you and that other person a favour by asking questions.

However, don’t be lazy with your questions and don’t ask questions about absolutely everything. This can have the opposite effect to what you may have hoped.

Should you ask questions as a Software developer? What are the benefits of asking questions?

Hi, I'm Adam Drake

Featured Articles on Medium

It’s the Perfect Time to be a Senior Developer

Using the Atlassian MCP server in Cursor

My New Favourite Stack — Tanstack Start, Laravel, Postgres and Mantine

Technical Skills & Tools