Yep, I realized this back in February, you can remake basically anything in a drastically shorter span of time with a single frontier model and patience and test thoroughly.
Yes, anything, not just software. I wonder if monopolies will somehow start falling or if they will try to defend themselves, are we on a path of continuous decentralization?
Genuinely curious: what weird quirks have you encountered in Excel? Back 10 years ago when I was using Excel daily I didn’t think anything was weird about it (it was actually my favorite tool)
A very small group of companies have the infrastructure required to train and serve a frontier model. They have a stranglehold on the compute/resources required for a new participant to break in.
And an overwhelming majority of enterprises running AI in production do so via one of two providers.
We'll like have an open-weight opus 5.5 class model we can run on (fancy) local hardware in a year or two at full perf, especially with AI optimized M7 macs. That's pretty distributed IMO.
Because we might have 2 year old models running locally on the most expensive consumer hardware in 2028?
I'm optimistic about open weight/local but that's not my definition of distributed :) Also, this only applies to personal use. As soon as we are talking about building AI features into products and serving to moderate traffic, we're back to the big 2.
I'd rather wish people put effort into LibreOffice.
I'd hoped there would be significant improvements there as there is a huge scope of bug/compatibility fixes. But I guess it is not sexy enough.
Do they even allow AI contributions? I'd love to contribute if they do.
Some people tell me why not donate instead, and the reason is the same as me giving people actual presents instead of gift cards: I take the time to choose and understand an issue, spend time on it, just like I spend time picking a present. Both are riskier than giving money directly, but the euphoria of having my PR get merged or present being appreciated is addictive :)
This is a really interesting possible difference. Legacy open source seems to be generally trending anti-AI (although I don’t know the exact stance of LibreOffice). These new vibe-coded Adobe/MS copies are obviously AI-friendly.
I'd rather not. LibreOffice is terrible and borderline unfixable. It's certainly better than these vibecoded alternatives, but not where the effort should go. Perhaps we need a modern reboot of open source office products.
Just curious… why do you think that? We have been using LibreOffice in our company for 20+ years, exchanging files with customers and vendors who use MS Office in all its versions during this time. We never had a problem…
Lawyer, translator and professor here. I have an Office 365 subscription and I got native Word and other MS Office apps installed on my windows partition. But, since I use Linux on the desktop, and MS chose not to offer their office suite as Linux native apps, and they make it near impossible to use the Windows apps under Wine, I use LibreOffice.
And it works really well. I can use it and exchange documents with people using Word without problem in the enormous majority of the cases.
And the fact that LibreOffice is a real GUI app built using the WIMP paradigm instead of that stupid ribbon interface that forces me to use the mouse, is a very significant net positive for me.
People who say that LibreOffice has a bad UI or that is a bad piece of software are just more vocal.
I'm sure you can use it with no problems. But the software is stuck in the 90s in usability, accessibility and overall look and feel. The way the dish looks is important, you know?
When you look at top tier open source projects like Blender or Godot, you see the difference. These applications are competitive if not superior in most fronts versus competitors. They feel like modern commercial software. Libre Office feels like the stuff you use just because it is free.
The media response to this omits the fact that these vibe coded "reverse engineered" apps hardly work and we've had open source alternatives to the Microsoft and Adobe products for years. The narrative they're pushing is a frontier model can make a perfect clone of any app, and the reality is it makes a ui that looks somewhat similar but none of the functionality actually works.
In the case of Microsoft Office native desktop applications the LLMs can figure out everything underneath through a mix of decompiling the binaries, reading the product documentation, and exploratory testing. Some SaaS is more opaque.
This is not true, most software can't be reverse engineered even if you had all of the source code, much less if you just have the binary.
Functionality piles on top of each other, products are such a garbled mess with fixes built on top of shaky grounds all over.
20% of software is code, 30% is design/strategy/marketing/telemetry/usage, rest of the 50% is lots and lots of stuff baked into software from user feedback over the decades.
Any software product that isn't in a similar balance is just crap to use, for instance Windows it's 20-30% code, 60-70% marketing/ads and the rest 10-20% bug fixes..
There is 0 user feedback being addressed.
In my opinion, AI is the thing that should be a wake up call for software companies to actually build on top of feedback proactively and with sharp intent.
I don't think you've been paying attention to what's happened in the last few months. The LLMs have gotten way better at reverse engineering and design. They might not be great at building a whole new product from scratch but cloning an existing mature application will be no problem within a year at most.
I don’t trust anything made by DHH now. He merged a PR that rewrote ttfx from rust to Claude generated asm (despite contributor objections) and then turned around and merged another that rewrote all of that into rust.
Which is to say that a normal organization would not be changing things so quickly based on the whims of a single individual.
Maybe I’d be ok to use omarchy just for fun, but could never use it seriously in any capacity.
Yeah though FWIW, the overall impact of companies realizing maybe they shouldn't be spending $100k a year on a document editor is likely to raise all boats for alternatives.
It remains really wild people think Office is worth $30/month/user. Anything that erodes that belief is a net positive.
i dont think anyone thinks it's worth that on its own, but you need some kind of user management (entra id in this case), some kind of cloud storage (sharepoint in this case) and some kind of standard document format (docx, xlsx, pptx in this case) in any company.
of course you can do all that with open source stuff but office is just easy to buy and hard to get rid of once you have bought in. the bigger your company the harder it is to migrate.
Entra ID is separate if you don't want logins from Nigeria. One of the most insane-to-me things is Office 365 is basically "open access by default" and you have to pay $6/month/user to fix it. Most people don't realize just how much attack surface they open up moving their data to the cloud, and that it costs a lot to even start to get back towards a secure environment like you had on-premise.
But yeah, the lock-in effect is huge. Microsoft cloud products deliberately don't work well with anything but more of their own cloud products. And as soon as you start a migration to it based on what you need, you suddenly realize how much other stuff you also need to pay them for. And you can't escape without replacing all of it.
I think a lot of people forget about the labor costs of migrating. The $30/mo looks cheap compared to all of the time spent trying to teach end users how to use something else
Part of the potential of these vibe coded clones is that the UI and workflows will be very similar to the genuine Microsoft products. So if they're actually completed and debugged then a migration won't necessarily require end user retraining.
Paying for software is perfectly fine. If people don’t want to pay for office there isn’t a lack of alternatives at all. Google doc is right here for them to use of they want a free cloud version.
The idea that companies with subscriptions are extorting users is cancerous nonsense
You've fallen prey to the "Everyone is stupid except me" fallacy.
By the time you add cloud-storage space integration (SharePoint), live collaborative editing, support hotline, support for 30 year old macros, and Active directory user integration, you'll be charging the same per seat.
Also wake me up when there’s a cloud version that works in an even vaguely similar way.
The Office advantage is a suite of apps and integrations with their cloud offerings. That’s their big moat against vibe-coded apps, because few vibe-coders want to maintain a SaaS service for Office files and even fewer CIOs want to commit their company to use that service when Microsoft exists and is proven to work.
The narrative has a point. The LLMs are getting very good at this task. I got 95% of functionality by using LLM to port our Qt desktop app to thin web client and web server. Astra medium, two prompts, 2 hours of LLM processing. Each of two LLM plans had about ten steps. The missing step for porting of these apps is to test against the original (ex. Ms office) app in the plan.
The alternatives have always been crap and unusable and incompatible with actual professional software and this is coming from someone who loathes Windows/Adobe and tried very hard for more than a decade to not rely on Windows-only products. I know, Microsoft doesn't respect its own specs, the maintainers do a heroic job, it's free, etc. Still doesn't change the usability issues, and they've been going on for literal decades.
Funnily enough, two common retorts against criticism of user-centric open source software are:
>stop complaining and feel free to open a PR if you're unsatisfied
Seems like many maintainers are now getting what they wished for and they are somehow unhappy about it. Ooops
>feel free to start your own project
Seems like the artcraft dude literally just did that and people are somehow unhappy about it. Ooops
My thinking is that cloning existing apps is missing the point. It still takes a disproportionate amount of time to get all of the functionalities and quirks of the original right.
But, it takes no time to get the handful of things you need working, on your system. That is where I see the most potential: there is little need for massive pieces of software that do everything, I can build the components I need, customized for my use case, and move on.
Does that actually work with Office tools? Like yes you can make your own documents and spreadsheets. But what about when you receive something from someone else? Or need to send it back after doing your edits? Do the files still work correctly?
Big part of the value is in true interoperability...
I don't think the best path towards interoperability is to run after those who based their entire business model on making interoperability a problem in the first place.
If I were to create an office suite that cares for interoperability, it would be based on markdown and CSS, viewable in the terminal, one could grep and sed through it, LLMs could read and write it natively, ... But I'm afraid it'll take another 40 years until we're there.
I've been thinking about this these days, this is the final revenge of open source. After having ingurgitated all the open source projects there exist in the world, AI is making basically all software un-copyrighteable (yes, you can't copyright a picture made by AI as you can't copyright a piece of software made -entirely- by AI).
It should be doable to vibe code all Microsoft Office apps by using Apache OpenOffice as input. The Artcraft team seems to have successfully been able to vibe code all of Adobe's creative suite https://getartcraft.com
We are really speed running destroying the software market, the long term effects won’t be nice. There is a reason FLOSS isn’t about cloning 1-1 existing applications. That’s something we could always have done if we really wanted. What we will end up with will be corporations pushing everything to the cloud, pushing for even stricter copyright/trademark protections to fight against that type of clones, stricter controls over software distribution.
Instead of using the power of LLMs to create genuine new products we instead get clone factories just to avoid paying or switch to an alternative. What a really dumb future we are building here, where trying to create novel software is a risk because any 14yo with a Claude subscription will clone it, distribute it for free. At least with keygen someone had to be pretty skillful and the distribution was niche…
This would be interesting if it was based on markdown and Jupyter notebooks and all that good stuff we have today.
But just replicating what was useful some 37 years ago to get office workers accustomed to the new digital world? Why?
Never mind, just checked it out... it has absolutely nothing. I guess this is nice for people who work with excel as a fancy digital scrap paper, for which indeed many alternatives already exist.
I've been waiting 26 years for a Visio replacement since M$ bought it, hopefully they can do it too. No other app or web-based copies are good enough still (sorta omnigraffle if you use a mac [ew, no]), and I've tried most over the years to find one as I still use Visio weekly in 2026. It's the only reason I'll actually buy a copy of real windoze and visio for keeping around a VM, I can still buy real licenses overseas for both visio and project for ~$20usd each instead of the ~$500-1000usd actually IN the US.
I don't buy the hype on these vibecoded apps, but I think an interesting use of this methodology would be to revive proprietary applications that were axed by their owners.
For example, what about a new clone of Macromedia's Free Hand? or Adobe's Page Maker? Or MS Front Page, vibecoded to be compliant with the latest W3C standards?
but if you have an old PC with Frotz to play with, create some game by hand (no AI used, at all) with Inform6 or better, Puny Inform and try doing things by yourself. You will be surprised of what you can do with very little...
72 comments
[ 2.3 ms ] story [ 49.6 ms ] threadThere are plenty of stories, some of them:
Pasting old dates: https://www.journalofaccountancy.com/issues/2015/aug/excel-d...
Low row limit: https://support.microsoft.com/en-us/excel/excel-specificatio...
Conversions: https://pmc.ncbi.nlm.nih.gov/articles/PMC4994289/
And an overwhelming majority of enterprises running AI in production do so via one of two providers.
I'm optimistic about open weight/local but that's not my definition of distributed :) Also, this only applies to personal use. As soon as we are talking about building AI features into products and serving to moderate traffic, we're back to the big 2.
Some people tell me why not donate instead, and the reason is the same as me giving people actual presents instead of gift cards: I take the time to choose and understand an issue, spend time on it, just like I spend time picking a present. Both are riskier than giving money directly, but the euphoria of having my PR get merged or present being appreciated is addictive :)
Time will tell which wins.
And it works really well. I can use it and exchange documents with people using Word without problem in the enormous majority of the cases.
And the fact that LibreOffice is a real GUI app built using the WIMP paradigm instead of that stupid ribbon interface that forces me to use the mouse, is a very significant net positive for me.
People who say that LibreOffice has a bad UI or that is a bad piece of software are just more vocal.
When you look at top tier open source projects like Blender or Godot, you see the difference. These applications are competitive if not superior in most fronts versus competitors. They feel like modern commercial software. Libre Office feels like the stuff you use just because it is free.
Maybe that's many are convinced AI can replace developers, they can only see the surface of software and think there's not much more underneath it...
Functionality piles on top of each other, products are such a garbled mess with fixes built on top of shaky grounds all over.
20% of software is code, 30% is design/strategy/marketing/telemetry/usage, rest of the 50% is lots and lots of stuff baked into software from user feedback over the decades.
Any software product that isn't in a similar balance is just crap to use, for instance Windows it's 20-30% code, 60-70% marketing/ads and the rest 10-20% bug fixes..
There is 0 user feedback being addressed.
In my opinion, AI is the thing that should be a wake up call for software companies to actually build on top of feedback proactively and with sharp intent.
Good luck and please notify us about the progress.
Which is to say that a normal organization would not be changing things so quickly based on the whims of a single individual.
Maybe I’d be ok to use omarchy just for fun, but could never use it seriously in any capacity.
It remains really wild people think Office is worth $30/month/user. Anything that erodes that belief is a net positive.
of course you can do all that with open source stuff but office is just easy to buy and hard to get rid of once you have bought in. the bigger your company the harder it is to migrate.
But yeah, the lock-in effect is huge. Microsoft cloud products deliberately don't work well with anything but more of their own cloud products. And as soon as you start a migration to it based on what you need, you suddenly realize how much other stuff you also need to pay them for. And you can't escape without replacing all of it.
The idea that companies with subscriptions are extorting users is cancerous nonsense
By the time you add cloud-storage space integration (SharePoint), live collaborative editing, support hotline, support for 30 year old macros, and Active directory user integration, you'll be charging the same per seat.
But hey, proof me wrong please. Code it.
Have you... ever tried talking to Microsoft support?
The Office advantage is a suite of apps and integrations with their cloud offerings. That’s their big moat against vibe-coded apps, because few vibe-coders want to maintain a SaaS service for Office files and even fewer CIOs want to commit their company to use that service when Microsoft exists and is proven to work.
Not to mention it's single binary barely 80 MB and super fast to start and very snappy.
Have you tried that or that PDFCraft?
Funnily enough, two common retorts against criticism of user-centric open source software are:
>stop complaining and feel free to open a PR if you're unsatisfied
Seems like many maintainers are now getting what they wished for and they are somehow unhappy about it. Ooops
>feel free to start your own project
Seems like the artcraft dude literally just did that and people are somehow unhappy about it. Ooops
LibreOffice is actually fine for 99% of all paper pusher workloads. The only office app that can claim superiority is Excel.
Big part of the value is in true interoperability...
I would not trust slopcoded tax software and I would not trust slopcoded CAD tools in any way.
Has the author even opened the apps, let alone tried to use them?
> GridCraft on getartcraft.com
> WordCraft on getartcraft.com
> 404 This page could not be found
(though, "to be fair", not existing is "far from what they intend to replace")
Instead of using the power of LLMs to create genuine new products we instead get clone factories just to avoid paying or switch to an alternative. What a really dumb future we are building here, where trying to create novel software is a risk because any 14yo with a Claude subscription will clone it, distribute it for free. At least with keygen someone had to be pretty skillful and the distribution was niche…
For example, what about a new clone of Macromedia's Free Hand? or Adobe's Page Maker? Or MS Front Page, vibecoded to be compliant with the latest W3C standards?
but if you have an old PC with Frotz to play with, create some game by hand (no AI used, at all) with Inform6 or better, Puny Inform and try doing things by yourself. You will be surprised of what you can do with very little...