I have a business. Sometimes I buy things on Amazon. At tax time, I search for the big business items and categorize them in QuickBooks until I give up.
You know who doesn't give up? Claude, who doesn't seem to realize how shitty and slow QuickBooks Online is. It's even happy to click 19 times to do a split because I dared to add a router and shampoo to the same Amazon cart in a weak moment of convenience one night back in February 2025.
Some so-called skills are worth letting atrophy. Whatever I used to do with a janky browser plugin and two monitors full of Excel windows at tax time is one of them.
I wonder if you could get claude to serve up amazon in a reasonable way:
- the search terms you asked for
- filter out results you want to remove
- no sponsored results
- no recommendations
for example, say I don't want "smart" <something>. I can't get rid of them from the results. "not smart" uses smart as a search term. Not only that, amazon devices always show up, I drown in sponsored results, and a good part of the page is my shopping cart, recommendations, and sometimes ask alexa popups.
Amazon does not want this. The more stuff you wade through, the more ad revenue they get. The more impulse sales they get. The more amazon devices they sell. (and then amazon devices become a surveillance cash register sitting in your house)
I agree that ecommerce interfaces (with the exception of McMaster-Carr) are all terrible. However, the next phase will probably leapfrog "make the interface better" to have a concierge agent do the shopping, or at least present a tight list of options and pros/cons.
By running a Tax Summary report and looking at it, just like I always did. Except this time, with fresh eyes and in a good mood. It's still just a bookkeeper and frankly does a more thorough job than I did.
notably, in my experience book keepers have been doing a terrible job since COVID, often committing what seem to be highly negligent mistakes. so i’m really not sure that is a compliment. but does seem to be somewhat useful?
the only way I found out w/ my bookkeeper was reviewing individual entries
> Whatever I used to do with a janky browser plugin and two monitors full of Excel windows at tax time is one of them.
Yeah the amount of time the taxman has wasted in my life is beyond belief. I automated a lot of it myself over the years but now having the help of LLMs to fill the bullshit, pointless, Brazilian (the movie, not the country) crap to please mister taxman is a godsend.
What I think this was intended to say was that a lot of nonsense work can be done by AI, so let’s eliminate the nonsense work. Valid, but consider that you may be protected too by the amount of paperwork it takes to hurt someone.
> this was intended to say was that a lot of nonsense work can be done by AI, so let’s eliminate the nonsense work
I'm not sure this makes sense, since there is no getting away from "nonsense work", which is shorthand for low value return on time and cognitive overhead.
I understood the post to mean something different...with my own added context.
The digital age has increased the complexity of life and various processes by orders of magnitude. Programming is at the forefront, having had to deal with implementation for decades. The complexity necessarily grows every day, as the easiest problems are solved first, before tackling more nuanced or difficult ones (very different). Understanding how to identify these, track the tradeoffs, and choose specialized approaches, grows in every vertical. There are many subjects that a person can study for their entire life and not fully comprehend the totality. We have been living in a time where specialists must be increasingly specialized, for decades, due to the breadth of human knowledge and social progress.
One of the best ways to deal with this complexity, has been to be smarter. Have a better memory. Have high cognitive function. This leaves an ever-growing amount of the population in uncertainty. Families all have (or had) a "smart" resource they tapped to answer the tough questions and find paths through seemingly-intractable legalities, kafka-esque bureaucracy, medical emergency, financial advice, et al. AI has given many of the less-gifted (the majority of the population) a "wise" resource that seems to be informed enough to provide solutions and even offer opinions that can be reasoned, for some value of "reason". AI has also granted easily accessible automation. This will make the situation worse. A form of specialized logical processor would have been necessary at some point (mentat or AI). It came earlier than many expected, but this is the form and it's here now.
instead of me using my brain to figure out the web of emails and contacts for an integration project, I can connect [AI] to my email inbox and say "who the fuck do I need to talk to to get an answer for where this project is”
Huh? What kind of project are you working on where you don’t even know who you’re working with?
This specific situation was an EMR multi step implementation that included multiple hands from billing, sales, tech support, PMO, and someone named Trudy who “usually knows where things are in stuff like this”.
I don’t entirely disagree with the premise persay - but then I think this warps the incentives, we’ll spend more and more on tokens because there will be more and more forms. We’ll add microtransactions to every form AI parses.
It all seems like an improvement in this world, but wait until the enshittifiers figure out all the ways they can abuse this process. They can add new micro taxes you won’t even know about!
As an SRE, I think this question comes up which is like would you ever trust AI to manage your infrastructure, and I think capability wise maybe, but long term I think it’s going to lead to some wacky incentives/interaxtions.
I’m not the person who reads every line of code, but we are in a golden age of AI right now where “I don’t have to understand anything I can just let the AI do it all” feels like a possible future.
> certain keyboard shortcuts are off-limits. Press the Escape key and you are thrown out of fullscreen mode. There’s been no way to prevent this — until now.
> The Keyboard Lock API lets you request access to specific keys when your application needs it. A game can now use the Escape key for its own menu system.
I cannot recommend StopTheMadness [0] enough against this kind of user-hostile "feature".
I have a business. Sometimes I buy things on Amazon. At tax time, I search for the big business items and categorize them in QuickBooks until I give up.
You know who doesn't give up? Claude, who doesn't seem to realize how shitty and slow QuickBooks Online is. It's even happy to click 19 times to do a split because I dared to add a router and shampoo to the same Amazon cart in a weak moment of convenience one night back in February 2025.
Some so-called skills are worth letting atrophy. Whatever I used to do with a janky browser plugin and two monitors full of Excel windows at tax time is one of them.
AI's superpower might be it's ability to serve as versatile and easy to use interface on top of systems with terrible UI, inconsistent data, etc.
DELIBERATELY terrible UI
I wonder if you could get claude to serve up amazon in a reasonable way:
- the search terms you asked for
- filter out results you want to remove
- no sponsored results
- no recommendations
for example, say I don't want "smart" <something>. I can't get rid of them from the results. "not smart" uses smart as a search term. Not only that, amazon devices always show up, I drown in sponsored results, and a good part of the page is my shopping cart, recommendations, and sometimes ask alexa popups.
Amazon does not want this. The more stuff you wade through, the more ad revenue they get. The more impulse sales they get. The more amazon devices they sell. (and then amazon devices become a surveillance cash register sitting in your house)
I agree that ecommerce interfaces (with the exception of McMaster-Carr) are all terrible. However, the next phase will probably leapfrog "make the interface better" to have a concierge agent do the shopping, or at least present a tight list of options and pros/cons.
How do you know it’s doing it correctly and not walking you into tax fraud?
By running a Tax Summary report and looking at it, just like I always did. Except this time, with fresh eyes and in a good mood. It's still just a bookkeeper and frankly does a more thorough job than I did.
notably, in my experience book keepers have been doing a terrible job since COVID, often committing what seem to be highly negligent mistakes. so i’m really not sure that is a compliment. but does seem to be somewhat useful?
the only way I found out w/ my bookkeeper was reviewing individual entries
Indeed. I can't imagine I'm the only one signing amended returns this week after having LLMs take a peek at the last three years of tax filings.
> Whatever I used to do with a janky browser plugin and two monitors full of Excel windows at tax time is one of them.
Yeah the amount of time the taxman has wasted in my life is beyond belief. I automated a lot of it myself over the years but now having the help of LLMs to fill the bullshit, pointless, Brazilian (the movie, not the country) crap to please mister taxman is a godsend.
What I think this was intended to say was that a lot of nonsense work can be done by AI, so let’s eliminate the nonsense work. Valid, but consider that you may be protected too by the amount of paperwork it takes to hurt someone.
> this was intended to say was that a lot of nonsense work can be done by AI, so let’s eliminate the nonsense work
I'm not sure this makes sense, since there is no getting away from "nonsense work", which is shorthand for low value return on time and cognitive overhead.
I understood the post to mean something different...with my own added context.
The digital age has increased the complexity of life and various processes by orders of magnitude. Programming is at the forefront, having had to deal with implementation for decades. The complexity necessarily grows every day, as the easiest problems are solved first, before tackling more nuanced or difficult ones (very different). Understanding how to identify these, track the tradeoffs, and choose specialized approaches, grows in every vertical. There are many subjects that a person can study for their entire life and not fully comprehend the totality. We have been living in a time where specialists must be increasingly specialized, for decades, due to the breadth of human knowledge and social progress.
One of the best ways to deal with this complexity, has been to be smarter. Have a better memory. Have high cognitive function. This leaves an ever-growing amount of the population in uncertainty. Families all have (or had) a "smart" resource they tapped to answer the tough questions and find paths through seemingly-intractable legalities, kafka-esque bureaucracy, medical emergency, financial advice, et al. AI has given many of the less-gifted (the majority of the population) a "wise" resource that seems to be informed enough to provide solutions and even offer opinions that can be reasoned, for some value of "reason". AI has also granted easily accessible automation. This will make the situation worse. A form of specialized logical processor would have been necessary at some point (mentat or AI). It came earlier than many expected, but this is the form and it's here now.
This specific situation was an EMR multi step implementation that included multiple hands from billing, sales, tech support, PMO, and someone named Trudy who “usually knows where things are in stuff like this”.
I don’t entirely disagree with the premise persay - but then I think this warps the incentives, we’ll spend more and more on tokens because there will be more and more forms. We’ll add microtransactions to every form AI parses.
It all seems like an improvement in this world, but wait until the enshittifiers figure out all the ways they can abuse this process. They can add new micro taxes you won’t even know about!
As an SRE, I think this question comes up which is like would you ever trust AI to manage your infrastructure, and I think capability wise maybe, but long term I think it’s going to lead to some wacky incentives/interaxtions.
I’m not the person who reads every line of code, but we are in a golden age of AI right now where “I don’t have to understand anything I can just let the AI do it all” feels like a possible future.
https://www.etymonline.com/word/per%20se ;)
> I look at Apple crushing decade old bugs and technical debt
Specifics on this? There is zero evidence of this.
If you have followed their latest OS releases you would see an uptick trend on stability and consistency, instead of new.
This is one of the first things I saw back in March that began that trend in my eyes. https://9to5mac.com/2026/03/24/apple-details-safari-26-4-wit...
> certain keyboard shortcuts are off-limits. Press the Escape key and you are thrown out of fullscreen mode. There’s been no way to prevent this — until now.
> The Keyboard Lock API lets you request access to specific keys when your application needs it. A game can now use the Escape key for its own menu system.
I cannot recommend StopTheMadness [0] enough against this kind of user-hostile "feature".
[0] https://underpassapp.com/StopTheMadness/