It’s hard right. You soon realise the more you wish for the harder it is to adequately describe.

Let’s keep it simple. Let’s say you wished for an elephant. Well what size of elephant? What colour should it be? (I mean the exact colour all over). How big should its ears be? How long should its trunk be? Even with a ‘simple’ elephant there are many details to consider.

This is what I am realising about Spec Driven Development. In theory it sounds great, just like being granted a magical wish, but in reality its doesn’t work out quite how you would expect.

Spec-Driven Development flips the script on traditional software development. For decades, code has been king — specifications were just scaffolding we built and discarded once the “real work” of coding began. Spec-Driven Development changes this: specifications become executable, directly generating working implementations rather than just guiding them.

I’ve seen the discourse online about Spec Driven Development. 

It seems to divide opinions pretty strongly

. I’m definitely falling into the camp of “

It sounds good on paper but in reality it’s a patchy approach at best

If you believe the hype then you write out a spec, point the AI to the repo and task board and then go away to the beach for the weekend and come back Monday morning and everything is done.

If you don’t believe the hype then you realise that software evolves over time, you can’t possibly write specs for all the scenarios and edge cases you will encounter in the wild and 

AI cannot be trusted for one task, let along a whole development cycle

The biggest issue is you have to declare what you want up front before anything gets built. It’s remarkably difficult, if not impossible, to envision and specify what you’re going to create in the very beginning. Just like describing your magical wish.

The level of detail and the level of edge cases that the app will encounter in the real world is mind-boggling.

That’s literally why we have code. It gives us just the right level of abstraction to be able to give enough detailed information and instructions to the computer whilst still being legible to humans.

I’ve been spending the last three years working on a few production apps, and the state the apps are in now versus what was envisioned in the beginning is light years apart.

On top of that is also the product lifecycle. 

Most Software Developers aren’t working on greenfield projects. They are working on existing projects within teams with some kind of project manager or project owner.

Personally I’ve spent the last six months fixing a whole bunch of bugs and very small edge cases on existing apps. These have only become apparent as the app has been used more and more in the wild. Specifying everything up front is impossible.

The majority of my time in a week is taken up with small bug fixes and incremental improvements that no person on earth could have predicted in the initial specifications.

Is Spec-driven development the way forward for Software Development?

Longer answer — It can be a useful approach in the right (very restrictive) circumstances and there are definitely some good practices to take from it.

I can theoretically understand the appeal of the Spec Driven Development approach. You write down clear instructions, you break it into small, digestible pieces, and then set to work. Whether thats you or AI is irrelevant. The point is everything is specced out so it’s crystal clear what needs to be done.

For a developer, that sounds really attractive. It’s a simple, straightforward approach. Everything is thought through and planned ahead of time before any implementation, which is usually a good thing in software and it allows one to take advantage of the benefits of AI.

It introduces guard rails and if done well, will limit the potential for “interpretations” by the AI.

However, having said that even with all the upfront work and guard rails in place there will still be things going wrong. Working in an existing codebase is never clean. AI may pick up bad habits from the existing code. Edge cases are likely to be missed. Changing one thing in one place could break some existing behaviour.

If used incrementally on small subsets of the codebase, then Spec Driven Development can be useful. 

I would even prefer it in some circumstances as it will force you to think things through carefully ahead of implementation time.

But it won’t solve everything. Even on small pieces of work things will be missed so as a developer you don’t want to come to rely on the spec. It will become a crutch and will limit you. As a developer you need to remain agile and alert.

As soon as I see anyone really pushing a development approach or practice, I always get suspicious. Anyone that is so hard-lined has probably not thought through the entire consequence of whatever they’re promoting.

Spec Driven Development is the same. It’s like anything, it certainly has it’s use cases, it brings some interesting ideas to the table and certain people will be more drawn to it than others.

I personally see it one more step towards completely sucking out the joy of coding. I don’t know one developer who is in the game so they can write more documentation. Don’t get me wrong, documentation is important but it’s not something someone jumps out of bed to write in the morning. Not any sane person anyway.

Is Spec Driven Development the future? I don’t personally think so. In anything more complex than a very simple CRUD app, holes start appearing in this approach and it ends up not being a very suitable tool for the job.

Spec Driven Development Sounds Perfect — Until You’ve Actually Shipped Software

It was at that point that I realised my social media consumption had to change. It was becoming all consuming. As I stood there an anger suddenly swelled up inside me. Where did that come from?

I think I was angry at letting myself get swept up in the social media wave. I thought I was above that. Obviously not.

I started thinking long and hard about the places it showed up in my life. I thought objectively of the true value it was bringing me. I asked myself difficult questions that I realised I had been avoiding for quite some time.

It was uncomfortable. My body felt physically awkward. I didn’t like my emotions showing their unwelcome faces.

I am a father of 3 children, I’m in my 40s and I live in the Czech Republic. I have thought long and hard about social media and the role it plays in our lives and I think I have an approach that works for me. Take from that what you will.

As I sat there I started to classify the different social media apps into a variety of groups. They didn’t fit nicely. This was going to be more complicated than I thought. After some time I came up with some rough groups:

Social media is difficult. To generalise and say it’s all bad is lazy. To be open to everything and consume whatever comes your way is dangerous and irresponsible.

Finding a path forward was going to be tricky.

This is the easiest group and for me there are two obvious ones that fit straight into this:

I never understood Tiktok and I am still amazed at how popular it is. I’ve never downloaded or used it. I have seen it a few times on friends phones and I started feeling anxious after just 10 seconds. I knew right away I didn’t need this in my life.

This one is a bit more nuanced. I did have it for a while on my phone and I was even posting fairly regularly. I soon realised however that I was spending so much of my “thinking time” pondering about what I could post next.

I was victim to the amount of likes my posts received and who in particular liked it. If certain people liked a post I created I could literally feel the dopamine flowing into my blood stream.

After some time though I finally realised that it wasn’t serving me. Most of the things people are sharing on there are false highlights and I was spending way too much time trying to please virtual strangers.

There is value in it though for many people and that’s where the complications occur

. If so many people gain value from the app then am I just not using it right? If I delete it will I be missing out on something?

Then I thought about what I value in my life. Not what other people value.

After much deliberation I decided that Instagram was not for me. I deleted the app. I deleted my account. 

I felt a pang of regret and something akin to grief.

 I found it incredible that this one app would cause so many levels of emotion inside me.

After a few days though, those feeling slowly started to subside and I thankfully haven’t looked back.

The obvious choices for me here are Twitter and Facebook.

Twitter is a bit of a strange one because I could make genuine valid reasons why my life would be better with more of it in my life. 

I work in tech as a Software Developer for my day to day and Twitter is the main place where developers hang out and share their thoughts. Whenever I do go on Twitter I usually learn about something new and interesting. So what’s the problem?

The main issue is the dark side of Twitter. The posts that annoy and enrage you. I don’t know if Twitter do this deliberately but they always manage to show me posts that get under my skin.

The last such instance of this being a photo of some protestors sitting on a main road blocking everyday commuters in order to highlight some cause they believed in. 

I remember not agreeing with the protestors and then getting involved with the comments. Never a good idea.

The only thing I can still really recall about the whole thing is the arguing, the anger I felt and the overall negative feelings I came away with.

Again, I had to ask myself: “What am I truly getting out of this?” — After much thought I realised I didn’t really have a good answer. The only reason I haven’t deleted this account too is the fact that I do very occasionally get some good out of it connected with my profession.

Facebook nowadays seems quite antiquated and tame compared to the other social media apps. It’s the eldest and it does show signs of that.

It’s actually the only social media app I have on my phone. I justify that mainly around ‘Marketplaces’ as in Czech it’s still a good place to sell your old stuff.

But I do also like to indulge myself sometimes and spend 5 minutes scrolling through the latest posts. We’re allowed a little indulgence in life. Right?

It’s mainly a few of my ‘friends’ posting some holiday snaps amongst a growing list of adverts.

 It’s mostly the same people posting too and honestly it’s all pretty harmless. I don’t feel any FOMO, I don’t feel any inclination to post anything.

The Facebook groups are good too. I often check the local town’s Facebook group for any updates on any local happenings around the town. 

But that’s probably just a sign of my stage in life. I’m more interested in the local planning permissions than what Lady GaGa (or whoever is popular these days) is wearing in her recent concert.

Then we arrive at my kryptonite — Youtube. Up until recently I would sing from the rooftops about Youtube. The diversity, the quality, the subject matter. It’s all there.

However, recently I’m starting to change my opinion. Like I said, social media is complicated. 

(And yes — I do still categorise Youtube as Social Media even though you could easily argue it’s just entertainment. It falls into that bucket of applications that are highly addictive, it’s social (in a way) and the content is managed by an algorithm. Therefore it must be monitored as such to avoid bad habits from forming.)

I don’t know if it’s because I’m a software developer but the algorithm has recently been showing me more and more clickbait titles that I know if I watch them I will get angry. 

“AI is taking your job — 5 habits to protect your career.”

“This one tool will change your life”

“Mr Beast is what Marx warned us about” — 

seriously?! Karl Marx? The guy who developed the theoretical framework of Communism. We should be worried about what he warned us about? Have people forgotten history already?!

This could just be Youtubers getting more savvy with their video titles too plus the influence of AI generating titles. I don’t like the direction it’s going.

However, I will probably spend at least 20–30 minutes on Youtube everyday. I still think this is fine but I am wary of it and I think that’s good. As long as you are aware of your compulsions then at least you’re not running on autopilot.

I’m keeping it for now as the net effect of this app on my life is still positive.

Social media is tricky. My relationship with it has changed over time but I think I have finally found a good place. For now at least.

It’s not as easy as removing it completely from your life but at the same time you need to have control over it otherwise it starts taking over.

I definitely lean on the side of what Cal Newport would call “Digital Minimalism” but I still think it can provide value in the right situation.

Apps like TikTok and Instagram I think could disappear tomorrow and I wouldn’t even notice. In fact I would notice because my 15 yr old son would maybe spend a bit less time doom scrolling.

Twitter (despite the horrendous negativity it can generate) has provided me value and Facebook indulges me on occasion.

My current sticking point is Youtube. The algorithm seems to be pushing more evocative content my way recently. This could be due to the impact of AI on Software Engineering or Youtubers learning the art of click bait. I’m not sure but I’m finding myself becoming anxious more and more after viewing some videos.

I will definitely need to keep an eye on it.

What’s your guilty pleasure when it comes to Social Media? Are there any apps you would outright ban? Do you think they are a fundamentally good thing for society?

I’m a Dad in My 40s Who Deleted Instagram (And Hasn’t Looked Back)

Then Anthropic released this feature — 

 — the impact this could have on my life scared me.

In itself, this seems like quite a cool little feature but there is something darker and sinister underneath all of this. I am seeing more and more of this recently and I don’t like the direction it’s going.

They push all the right buttons. It’s easier and easier to have access to your agents. Agents can do more and more with less prompting. Heavy users of AI are showing the same traits and behaviours as addicts.

Continue use despite negative impact on their life

Withdrawing from social and hobbies they used to enjoy

Rationalisation and justification of use — “Just one more feature”!

Denial — Refusing to admit they have a problem

Secrecy, lying or deceptive behaviour to hide how much they are using

When did building software become this? What the hell is going on?

It all starts with the promise of what AI can do for you. The lure and the potential of what AI can bring to your life is extremely hard to resist.

Who doesn’t want more for the same or less effort? It preys directly on human’s greatest weakness — 

AI can make your dreams come true. With a few prompts and a little business acumen you can now build that thing you always wanted to but never had the time and it will finally bring you the rewards you deserve.

Whether that’s recognition, wealth, popularity… it doesn’t matter. AI can help you get there. And fast.

Now with Claude Code Channels, you can reach that dream from where ever you are.

 No worries, ask you agent to implement SSO.

 No worries, ask you agent to implement that new api endpoint.

Out to dinner with my wife and she’s sharing a concern she has with the kids?

 No worries, add some extra test to make sure your app covers that weird edge case.

It’s easy to see this becoming a slippery slope. The same as 1 glass of wine at dinner every night soon becomes 2 and then 3.

There is also a whole bunch of genuine validation you have behind you on this.

“Once I build this app we can have some extra income to help with the kids college funds”

“Once I have built out this feature, then I can release to the wider world and get the recognition I deserve”

“Once I build this internal tool, I will get that promotion at work and we can finally start enjoying financial freedom”

These are all genuine and valid scenarios. Opposed to alcohol addiction which inevitably ends up in a bad please, being addicted to AI agents will ultimately bring rewards. Right?

This is where I would argue things get complicated. This extra time you spend with your AI agents will at some stage become too invasive in your life.

Whether it’s at the dinner table where you’re staring at your phone typing frantically or out with friends and they notice you are not really present.

Having constant access to your AI agents means your mind can never switch off.

Your mind will be constantly thinking about what the next thing is you could do. And don’t worry, your brain will constantly think of something else to do.

This thought will become your mantra — 

“When I just do this THEN everything will be good”

This behaviour has all the hallmarks of an addict. The scary thing is I am already seeing it in myself.

I have started a serious side project now I know I can do more than I previously could have thanks to AI. I spend many hours thinking about it. I catch myself when I’m away from the computer thinking about the next feature I need to implement. If I had an agent on my phone when watching Netflix with my family I know I wouldn’t be able to resist typing in that next prompt.

This is the main reason this new release from Claude code scares me. 

I can see how easy it would be to add something to my life that ultimately would make my whole quality of life worse

These are more instructions to myself than anyone else but I think others will benefit from this too.

I will only ever use agents at my computer.

I feel I should maybe print these off and stick them on my desk. The risk of falling off this path is so easy these days.

. 80% of the stuff people are building with AI right now is horse-shite. 19% will be good developers utilising AI well to build better software. Then there will always be that 1% who manage to find gold.

The problem with our current system is we always hear about the 1% and think that that could be us.

” is a horrible myth. This has never been true and never will be true. Those AI gangsters on twitter bragging about the number of lines of code they have produced with AI today should be ignored at all costs.

In hindsight the friction we had before to even start a project was actually a good thing. It separated out the trash ideas from the good ideas before you even started.

I would advise to find a project you really care about and take your time. Use AI to help you make better software but don’t become obsessed by it. It’s not going to solve all your problems even if it is successful.

If you think something will take you 1 week, give yourself 3 weeks. Not only will it take the pressure off but you are much more likely to make a much better piece of software. Building one piece of great software will always triumph 10 average projects. Whether using AI or not, building great software takes time and many iterations.

I fear that the infrastructure being built up around these AI tools will start intruding into our lives more and more. For the younger generations, I see this being especially problematic.

The human brain continually has thoughts and ideas and this has plagued our condition for centuries. These generative AI models and the tooling around them have tapped into giving an outlet for these thoughts and ideas. 

“If I just add this one thing… If I just build this one app… THEN everything will be right in the world and I will be happy.”

I have to constantly remind myself there are so many more aspects to being alive and all should be allowed their time to be explored and enjoyed. I have a beautiful family, a loving wife and wonderful friends. I take these for granted too often and end up spending much of my time trying to build something I think will make me happy. It’s like I have won the prize already but I don’t realise it.

I know my best ideas will likely come to me when I’m not at my computer anyway. It will come to me when I’m on a walk, doing the daily commute or taking a shower.

I think allowing more of these times into your life and focus a little less on the constant creation is honestly the way forward.

Claude Code Just Released a Feature That Genuinely Scares Me (And I Use AI Every Day)

I did this when Crypto development was going through it’s hype cycle. 

I soon realised that DAPPs were extremely niche and not that useful in real life.

I built a real project and soon realised Nextjs was a great framework with some really positive features

. Then the whole verson 14 came out with the apps folder and the massive over caching. Nextjs and myself soon parted ways. 

It’s 2026 and AI is here with the biggest hype cycle I’ve seen in my lifetime as a developer.

It was time to try out a real life project. I didn’t want to fully vibe code it as I think that’s just naive and ultimately going to lead to a whole world of pain. I did however really want to try pushing the AI Agent and see where it’s limits were and where it performs really well and where it performs badly.

My project was something I have wanted to build for a long time. 

Czech is a slavic language and all the language apps I have used over the years tailor to MANY languages so don’t deal with the specifics of the Czech language.

This is what AI can do for you right? Provide you the possibility to build that thing you never had time for before?

I learnt quite a bit over the course of building this app. It took me about 4 weeks in my spare time. I worked on it pretty hard, even when heavily utilising AI agents to build out the code base. I was also using Laravel on the backend and Tanstack start on the frontend (My favourite stack!) so I had a really solid foundation to start with.

What I have produced is a good MVP — still quite far from a finished product. I don’t subscribe to the model of get something out as soon as possible. I think you need to put some care and attention even into the MVP — first impressions count.

Approach one feature at a time. If you do too much at once it quickly gets messy and something bad will go unnoticed. 

I sometimes let two agents run at the same time but that’s it.

 I have no idea how people are running 4 or 5 agents at once. Pure madness.

 I‘ve learnt it’s much better to be more accurate than less in the prompts. It’s good to imagine you are talking to a real human. Would they be able to understand your request? If not, then rewrite it.

I definitely write more tests. It’s easy now and overtime the suite of tests you build up are really useful. As the app grows and the feature set grows, complexity does too. Complexity has a way of sneaking up on you.

 It really pays to keep a close eye on coding patterns and structure. AI keeps referring back to the codebase so if it sees patterns you have implemented then it will follow that in future. If you don’t control as you go you are more likely to end up with a mess. 

The old adage of “Go slow to go fast” really works here.

Have the AI write some summary docs for specific things as you go. These act as good resources in the future when you want AI to work on something related. You’re able to bring the new AI up to speed fast. 

Remember: Every new session the UI knows NOTHING about your repo. It’s very weird but you constantly need to remind yourself of this. AI is like a very knowledgable developer who suffers with amnesia.

If it does something good, let it know. It stores this in memory for future work and can refer back to it. Like a dog, feed that reward cycle.

It’s still really hard to build a good app. I am realising the design choices of both the app and data are the hard part of building a successful app.

Always use Git. It makes it very easy to undo, redo, stash etc. AI can go off the rails sometimes and you want to be able to backtrack easily.

AI doesn’t produce great React code. Even with all the React code in the world it still doesn’t follow best practices. It prefers long component files that consist of many different components. I had to refactor this out many times into reusable components and follow the more “React” way of doing things.

Decide the tech stack before you implement anything. If you aren’t sure have a good discussion before with AI. Don’t let AI blindly choose for you.

Overall your software should be better. I am aiming to make good MVPs now rather than the “bear minimum” before I release. Your expectations of what’s good enough should increase with AI usage.

I like watching the AI output. It’s good to see it’s workflow, how it does things and the order it does things in. It’s all quite logical and straight forward really and you can learn. 

AI is a master of the terminal. CLI tools for the future.

You need to know software to understand what the AI is suggesting or implementing. Knowing about migrations, schemas, models, workers, services vs traits, testing etc. I couldn’t imaging doing what I did if I didn’t have all the knowledge and experience I have. You still need a really good mental model of what is going on.

Did AI Speed Me Up? Yes and no. It definitely allows me to write more code. However, I think now the bar for quality of software should be much higher. It’s easier to test your application. It’s easier to get through technically difficult moments. It’s easier to make decisions. But that shouldn’t mean we just ship more code. It should mean we make higher quality software.

AI when used in producing code is definitely a game changer. I have really tried to utilise it over the past month and honestly, I can’t see myself going back to hand coding everything.

The tools you use are important and it’s clear even the main tools are still going through quite a bit of iteration. The models you choose also have a big impact. 

 I think it’s best to find one you like and stick to it, although keep exploring at the same time.

 It saves time ultimately. Keeping requests small is also really important. It’s about narrowing the scope of possibilities.

There is no hard or fast rule for this but just imagine if you gave the same plan and prompt to a human developer. Would they know what they have to do? If not, then the scope of the task is probably too large.

You have to have good mental models still. You can’t rely on AI to guide you on the bigger decisions, especially architecturally. By all means ask for its opinion but don’t follow it blindly. It may be quicker in the moment but ultimately it could cost you.

I really don’t like the multi agent approach. Two agents at once is more than enough to keep my cognitive load at maximum. This whole orchestration of agents — at this current time — just sounds stressful and I feel will lead you into a forest of mess and pain.

If you’re not using AI at the moment, I would strongly advise you to at least dip your toe in the water. It’s another tool at the end of the day and it can help you with some of those tasks you really dread — like writing extensive tests. However, keep a close eye on it. It just loves to go off the beaten path so it needs a tight leash. And don’t believe the hype. Never believe the hype. As in most things, the truth lays somewhere in the middle.

If you made it this far, here is the app I released: 

 — Still plenty to do but its up and working. If you want to learn Czech — give it a go!

14 Lessons From a Developer Who Actually Built Something With AI (Not Just Vibed)

However, I have one major issue with it. This issue is so confusing and frustrating that it is forcing me to look into other options.

This problem ultimately boils down to ‘value’. When you part with your money, you want to make sure you’re getting ‘value’. I feel with Cursor I am not getting as much ‘value’ as I could be getting.

Let’s dive into the specifics, see where the issue is and then I will show you some hopefuly alternatives I have been exploring.

It’s all to do with how many tokens Cursor seems to be burning through whenever I use it.

Over the past month I have been working on a personal project to really explore using AI heavily when writing code. I have heard enough hype and opinions that I realised it was time to give AI generated code a good go.

I currently use Cursor as my main editor (although this shifts quite regularly at the moment) and have been using primarily the 

Opus 4.5/Sonnet 4.6 and Composer 1.5 models

My general approach is if I am building out a new feature I will go into 

 with one of the Opus models and use that model to design out the feature. Then I would get the same model to implement the majority of the code.

If I am working on small UI improvements, writing tests or small updates to logic I will try to use Composer 1.5 first and if it’s not up to the job I will switch over to Opus.

When I started this project just over a month ago I was on the $20 Cursor subscription. I thought this should be enough for this project. I was wrong.

It didn’t take long, especially with the Opus models, to reach my limit. Once I had reached this limit I couldn’t even use Composer anymore. After some deliberation I soon realised I had no choice but to update to the $60 subscription. (

Obviously I also had the option to continue the project without using AI but come on! It’s 2026… if you’re not using AI then you’re a fool. Right?

This is when I started looking into how Cursor counts usage. 

Cursor splits up your usage into two buckets:

API (If you use Claude Opus 4.5 or Codex 5.3 for example)

Press enter or click to view image in full sizeAPI usage quickly surged up to 99%

As you can see, even with the $60, I was soon up in the 90% usage and I am now at 99% and still have some days left until my subscription renews. I am stuck using Composer or Auto.

This API usage seemed to go up really fast. I was making use of the models but I didn’t think I was doing anything excessive. I wasn’t letting it run neverending loops. My prompts were very precise in terms of what files to look at and what changes I wanted. I therefore went and looked into the “Usage”. Again, I was shocked.

Press enter or click to view image in full sizeRidiculous token usage like 4.4M or 1.4M

I was seeing some outrageous numbers — 4.4M or 1.4M tokens!! This was Composer which I only use for relatively basic frontend tasks — “

Hi, I'm Adam Drake

Featured Articles on Medium

Software Developers Will Be Thriving in 5 Years (Despite What Your YouTube Feed Says)

Claude Code Just Released a Feature That Genuinely Scares Me (And I Use AI Every Day)

It’s the Perfect Time to be a Senior Developer

Technical Skills & Tools