AI Isn't Just Replacing Jobs, It's Replacing Software Too

SirTop

Elite Member
Jr. Executive VIP
Jr. VIP
Joined
Aug 23, 2017
Messages
2,628
Reaction score
7,069
I don't know if y'all saw the recent news about HubSpot laying off nearly 660 people.

It could easily be seen as another massive AI layoff story but I really think there's more to why some of these software companies might be laying people off.

I think part of it might be that customers no longer see the same value in what they're paying for with some of these software tools

Especially when most of these companies just keep increasing their prices.

Some teams and businesses are now building internal tools themselves because AI has made it much cheaper to do so

Or they're simply switching to cheaper software alternatives.

And once people start asking why they should pay so much for a SaaS tool when they can build something they actually own
Those big subscription prices start looking unreasonable.

But here's where I think it gets interesting.

As more customers start cutting back on subscriptions or moving to cheaper alternatives
These software companies could start feeling the pressure on their revenue and budgets.

That pressure could lead to layoffs as they try to reduce costs and at the same time, some might decide to increase their prices to make up for lost revenue or protect their profits.

But then those price increases make even more customers question whether the software is worth paying for, pushing them towards cheaper alternatives again.

And the cycle just keeps going.

giphy.gif



Yes, AI is still taking jobs but at the same time, it's also making it easier to replace a lot of expensive software with cheaper alternatives.

So maybe some of these layoffs aren't just about AI replacing employees.

They could also be a result of AI changing how much people are willing to pay for software in the first place.

What do you think? : )
 
Its not that its replacing software. (actually on reflection, it is thats a bigger picture! ) Fundamentally its replacing processes within that software. You can also access adobe cloud processes via new google tech, and build in all the same stuff.

There is already an open source adobe replacement creative cloud suite, not sure how long it will last, since its all adobe's IP? Surely they'll sue.

We have to assume things will go like steam from pc game fame. It will be which platform/software has the best in built service, and is willing to change the fastest to meet the demand of ever increasing and switching time periods of software. People will use your software in jan feb, then try something else for a couple of months, come back to you, then try something different again, lol.

(remember back in the day we all used to use napster, and the things that came after that to share files and music, there was no loyalty, we went were the community went. ) Now its happening at such a micro pace, that the macro overall model is having to adapt.
 
Its not that its replacing software. (actually on reflection, it is thats a bigger picture! ) Fundamentally its replacing processes within that software. You can also access adobe cloud processes via new google tech, and build in all the same stuff.

There is already an open source adobe replacement creative cloud suite, not sure how long it will last, since its all adobe's IP? Surely they'll sue.

We have to assume things will go like steam from pc game fame. It will be which platform/software has the best in built service, and is willing to change the fastest to meet the demand of ever increasing and switching time periods of software. People will use your software in jan feb, then try something else for a couple of months, come back to you, then try something different again, lol.

(remember back in the day we all used to use napster, and the things that came after that to share files and music, there was no loyalty, we went were the community went. ) Now its happening at such a micro pace, that the macro overall model is having to adapt.

You have just described exactly how most people behave now and it happens to me quite often as well : D
 
https://boristane.com/blog/self-improving-software-is-inevitable/

You might find this interesting @SirTop I just read.

Self-improving software is inevitable. Applications that read their own signals, decide what to change, ship the change, and check whether it worked. And once software can improve itself, it can improve the machinery that improves it: better observability, sharper detection, deeper investigations, better fixes, each loop tightening the next one.

For that to happen, we have to give up being in charge. Engineers, PMs, designers, all of us. We have to stop being the meat proxies who decide what gets fixed next, and when.

1791810611022.png

1791810625462.png

This is where the "puck is going" it seems. Self improving software. This will be cool, i can imagine working like this.

At the moment, i'm a defo meat proxy! lol

This is a market anyway, we can look towards, and hire out services, based on ai. We just need some more good AI tutorial links we can follow that might help us with this sort of thing about building local agents in software for people?

@xReminisce you know of any good sources High Sorceress? You seem to be doing all types of strange ai agent stuff...any good resources or tutorials of how we can build self improving apps?
 
Partly true, but most teams aren’t rebuilding a full CRM in-house. They’re just cutting seats and dropping add-ons. Same revenue hit though.
 
I agree with this. I see it in my own work too. These days often just build something with AI and automation. It's cheaper and does exactly what I need.

But theres another side to this. It's also getting harder to build new product that people will pay for when someone else can replicate most of its functionality in matter of days
 
True. Instead of learning more software, people are learning how to use AI better.
 
@xReminisce you know of any good sources High Sorceress? You seem to be doing all types of strange ai agent stuff...any good resources or tutorials of how we can build self improving apps?
Yes and No
AI isn't inherently smart
AI is just a numbers guessing game.
you can rig it to lean more towards a certain way and that's about it.
Skills and Rules goes a short way but for something true as self improving...
I really don't think we are there yet.
1791829091915.png
---
Although, that being said
i seen ways of where people are trying to invent a new AI only language to completely compress long form writing into much smaller amount of tokens
take a look https://github.com/MemPalace/mempalace
 
The SaaS-pocalypse is real, but the pendulum will swing back at some point.

Speaking from my own experience, the more inhouse tools I build for myself, either because SaaS pricing was ridiculous or the tool simply didn't exist, the more time I end up wasting on maintenance.

That’s the moment you realize you didn't actually save money; you just started paying with your own time instead.

A few examples:
1. I coded a slick, playlist-centric plugin for the IINA media player (Winamp-style, everything within reach, super handy). Then IINA 1.5.0 dropped, broke it, and I had to waste some time just adapting my code to the new version.
2. Same with a custom AI agent IDE I built: nearly 400 commits in less than six months. Every Claude update I have to update it too.

Nobody is paying me to maintain this stuff :)

Eventually, you hit a wall and start thinking: "Honestly... I’d happily throw cash at a paid tool just to make the maintenance someone else's problem."
 
https://boristane.com/blog/self-improving-software-is-inevitable/

You might find this interesting @SirTop I just read.



View attachment 551771

View attachment 551772

This is where the "puck is going" it seems. Self improving software. This will be cool, i can imagine working like this.

At the moment, i'm a defo meat proxy! lol

This is a market anyway, we can look towards, and hire out services, based on ai. We just need some more good AI tutorial links we can follow that might help us with this sort of thing about building local agents in software for people?

@xReminisce you know of any good sources High Sorceress? You seem to be doing all types of strange ai agent stuff...any good resources or tutorials of how we can build self improving apps?

This is really interesting, what you've shared here.

I think we will naturally get to a stage like this eventually.

I don't even know whether to call it advancement or laziness but since AI is already doing so much for us, we might as well just let it take full control.

I actually experienced something similar recently. I'm working on a project where I still have to handle the final stage myself.

It's a simple process that probably takes less than a minute but I've gotten so used to the convenience of AI handling everything else that I even tried getting it to handle that final stage too.

Unfortunately, it refused because of security restrictions.

And that just reinforces your point. If we could, I think most of us would happily sit back and let AI handle everything without our involvement.

But I'm sure security is still one of the biggest reasons people are skeptical about this, alongside the restrictions AI companies already have in place for certain tasks.

And at this point, Rem is already AI herself. She's always up to something and is already part of the clan

Sorry to say, but we've already lost one human to AI : D


Partly true, but most teams aren’t rebuilding a full CRM in-house. They’re just cutting seats and dropping add-ons. Same revenue hit though.

Yes, you're right about people cutting seats and downgrading plans or even negotiating custom plans with some of these software companies.

Not every part of a tool is useful to everyone and the idea of still paying full price for features you're not even using just doesn't make much sense anymore.


I agree with this. I see it in my own work too. These days often just build something with AI and automation. It's cheaper and does exactly what I need.

But theres another side to this. It's also getting harder to build new product that people will pay for when someone else can replicate most of its functionality in matter of days

It almost feels like there's nothing entirely new anymore. And even when someone comes up with something new, someone else could literally build something similar in no time.

I think the catch now is who can market what they've built better.

Or people might just focus on building things that work for their own needs, rather than necessarily creating something to sell or release for public use.


True. Instead of learning more software, people are learning how to use AI better.

Yup yup... people are now learning how to use AI to get those same things done.


Yes and No
AI isn't inherently smart
AI is just a numbers guessing game.
you can rig it to lean more towards a certain way and that's about it.
Skills and Rules goes a short way but for something true as self improving...
I really don't think we are there yet.
View attachment 551838
---
Although, that being said
i seen ways of where people are trying to invent a new AI only language to completely compress long form writing into much smaller amount of tokens
take a look https://github.com/MemPalace/mempalace

I know this is a meme but knowing you, Rem, this is something you would have actually done with Claude just to prove to it that Grok is better than it : D


The SaaS-pocalypse is real, but the pendulum will swing back at some point.

Speaking from my own experience, the more inhouse tools I build for myself, either because SaaS pricing was ridiculous or the tool simply didn't exist, the more time I end up wasting on maintenance.

That’s the moment you realize you didn't actually save money; you just started paying with your own time instead.

A few examples:
1. I coded a slick, playlist-centric plugin for the IINA media player (Winamp-style, everything within reach, super handy). Then IINA 1.5.0 dropped, broke it, and I had to waste some time just adapting my code to the new version.
2. Same with a custom AI agent IDE I built: nearly 400 commits in less than six months. Every Claude update I have to update it too.

Nobody is paying me to maintain this stuff :)

Eventually, you hit a wall and start thinking: "Honestly... I’d happily throw cash at a paid tool just to make the maintenance someone else's problem."

You know, I was actually talking to someone about this same thing yesterday and they brought up the issue of maintenance too.

But they didn't explain it in as much detail as you did and I can see how that could become an issue really quickly.

But then, thinking about the self-improving software Panther shared earlier, don't you think that if something like that eventually becomes possible, issues like the ones you've explained wouldn't be much of a hassle anymore?

What I told the person I was talking to yesterday was that someone in-house could probably handle maintenance for a team or business.

But obviously, it's a different situation when you're just one person using the tool and also have to be the one handling all the maintenance yourself.

I can imagine how tiring that would get, even just thinking about it now.
 
Yeah, I guess we'll eventually get to a point where AI becomes a super-assistant that creates software tailored to user. But for now,the most important thing is marketing - making it better than competition
 
Grok is good at nothing
disagree, drop a cartoon image in and ask it to animate, and it does an ok job, quick enough, at least it allows you to do such a thing, the others don't i don't think. This what i was using it for. it can do that 1 thing well i need. Thats it, 1 thing! lol All the rest can do everything better.
 
disagree, drop a cartoon image in and ask it to animate, and it does an ok job, quick enough, at least it allows you to do such a thing, the others don't i don't think. This what i was using it for. it can do that 1 thing well i need. Thats it, 1 thing! lol All the rest can do everything better.
astra prob does it better
opus 5.5 for sure does it better
lol
 
Back
Top