I am in the process of attempting to have AI run my business. I'm actually making very good progress, but it's happening in pieces - I document some task and have it take over, or I give it something to handle while staying in the loop and providing feedback until it's in good shape. At this point it's handling large swathes of my operations, marketing and finance.
The experience of getting it there makes me pretty skeptical of the idea of a general business agent like this. First, because I still find myself having to review some categories of work for errors. These are decreasing over time, but they're still there. I fully believe that as models get better, errors will decline, but I am somewhat surprised to see some of the errors current models make given their intelligence. A common set is having Claude take some product photos and turn them into lifestyle ones using ChatGPT via Claude in Chrome. It's a pretty well-honed workflow at this point, but it'll still return images where the product is obviously not correct and seemingly not notice them.
Anyway, even if the agents are "perfect" in terms of their ability to execute tasks, there's still just an enormous amount of nuance and context in each business that takes a ton of time to convey. I've been at this for a couple of years now, and I'm still clarifying things. Now maybe an agent starting a business from scratch would have a better time since it's not inheriting all of this, but to have one run an existing business requires a very extended handoff, even if the agent is objectively amazing at all aspects of running a business.
> I am somewhat surprised to see some of the errors current models make given their intelligence
They have no intelligence. These are very very very refined prediction engines.
> A common set is having Claude take some product photos and turn them into lifestyle ones using ChatGPT via Claude in Chrome. It's a pretty well-honed workflow at this point, but it'll still return images where the product is obviously not correct and seemingly not notice them.
Obvious to you or I, or someone with actual intelligence. But things like this slip by a frontier model in the same way AI from a few years ago would generate an image with seven fingers. They do not count. They do not understand. They do not consider.
No matter how good these models appear to be at intelligent tasks, it's foolish to give them "a company" to run, because they cannot understand when they've made a mistake the way even the least competent human can.
> They have no intelligence. These are very very very refined prediction engines.
Silly and pointless criticism. Why do the semantics of the word intelligence matter?
Also, you need to understand that language evolves over time, and people often use a word to describe a thing that is newly discovered or invented that's similar to the word being used, because it's a helpful way of describing it that the listener will understand better than a longwinded technical description. The purpose of language is to communicate thoughts, not engage in pedantry.
> But things like this slip by a frontier model in the same way AI from a few years ago would generate an image with seven fingers.
Yes, you make an excellent point here that over time, the capabilities of these models to recognize certain categories of problems have increased dramatically. I expect this will continue!
> it's foolish to give them "a company" to run, because they cannot understand when they've made a mistake the way even the least competent human can.
The words of someone who has not spent time with the least competent human, or anything close to it.
> Silly and pointless criticism. Why do the semantics of the word intelligence matter?
It matters because, we're still sorting out what intelligence means for an AI agent. As you pointed out, language evolves over time. The question remains whether or not attributing intelligence to the current iteration of models is correct. This is not settled and I don't see why it's wrong to bring it up.
It's not wrong to bring it up, but the other comment did not "bring it up". It purported to correct someone by categorically stating "it is not intelligence, you are wrong and I am right", while not once saying what, then, intelligence is.
One thing is clear: LLMs at least already are capable of 1) not making the same mistake that person made, and 2) clearly seeing why the other person's post was "wrong" in both reasoning and tone.
Being good in chess was (and still is) associated with being intelligent. But if a computer does it, it is just a calculator (and it is).
Then go, the great game for intelligence, too complex for calculators to have a chance against intelligent humans. Until it was solved.
And then text, the original Turing test solved, AI capable enough to fool humans. And now already replacing humans in jobs strongly associated with intelligence - programming.
I find it hard to debate, that we don't have created artificial intelligence, by the way we used to use the word "intelligence" before.
So if AI is really understanding something?
Likely not in the way we use that term. But it definitely shows intelligent behavior and actions.
When people talk about solving a game like go or chess, they mean proving mathematically if there exists a way to guarantee a win, or if the second player can always guarantee a draw, among related questions. Current game engines have no such knowledge, they just pick statistically what the think the best move is and hope for the best.
No, that is not what "people" in general mean, this is only what some people mean.
Other people know there is no solving go with calculating, but using statistics to achieve the goal of becoming better at humans. And they are. (With a recent unexpected exception unlikely to be repeated more often)
I don't think people "in general" refer to games as "solved" or "not solved" at all. Among people who do, chess and go are definitely not referred to as (fully) solved.
"to have a chance against intelligent humans. Until it was solved."
But this was the original statement - so solved references "chance against intelligent humans". And this is clearly solved. No comment on that there cannot be better go engines, but no matter how painful it is, they beat the best humans. (And it was painful, they did cry)
That's all the more reason to stop talking about it and actually discuss concrete capabilities or lack thereof.
We're all well aware that we have different thresholds for what we consider intelligence, and that these thresholds are constantly changing.
Even if we agree that "LLMs are AI!!!" or "LLMs are not AI!!!" in this thread, all we've established is that this particular set of commentors share a similar enough definition at this point in time.
So let's say we make up a new word, machilligence to serve as a parallel term to intelligence but strictly for machines. How is the world different in that case vs. if we call it intelligence?
Or I guess to put it another way, if we want to coin a new term for the intelligence-esque thing that AI has, but the key differentiator is that it's an AI thing and not a human thing, then what linguistic value does the new word have? If I said "Claude is intelligent" then the fact that we're talking about AI "intelligence" is already captured in the sentence anyway; no new word needed
I think "intelligence" carries baggage. When most people hear it, they're thinking of the constellation of things: judgment, consistency, moral reasoning, the ability to decide something and stick with it. An intelligent person has an internal model of the world, values they apply consistently, and the capacity to learn from mistakes in a meaningful way.
LLMs don't do any of that. They perform statistical pattern matching on text at an extraordinary scale. They're shockingly good at mimicking the surface features of intelligent behavior.
I think the overuse of the word intelligence is something to criticise as grandma is not across the tech details.
It matters the same way that calling a dog or a cat intelligent matters. It humanizes the thing, even if that isn't your intention. Right now that's not a big deal since most people agree that machines should not have rights, but I wouldn't take that for granted.
It's not the word I'm taking issue with, it's the concept. From your post you are surprised that the model makes mistakes and does not recognize things, and these are only surprising if you imagine the models to be thinking about things.
We can argue about words like thinking and intelligence all day long, and so can an LLM. But at the end of the day the LLM is only mimicking the processes you or I use.
Last week I tried out gpt sol 5.6 and asked it to count the letter r in a massive string of letters, without using an external app. It succeeded until I made the garbage sentence suitably large and then it consistently, confidently failed. Each time the "thinking" showed that it was teething to find a "gotcha" each time. "Ah, the first time I forgot to count the letters in the instruction itself" etc. at no point did it just understand that it had miscounted. It seems to be incapable of considering that it just made a regular mistake. Even a six year old child would just try again the same way and end up with the right answer
"Judgement" is the #1 thing missing from AI if you ask me.
You give AI an input, and then an output is created...a very coherent and deep output in fect...but, there is very little understanding of alternatives or downstream implications in my experience.
I think its a very good text/image generator on any subject. But I am not finding much in the way of understanding tradeoffs or downstream implications in a non-pro's and con's list. It's like a human that has a particular type of brain damage...they can make an infinite pro/con list, but not reasonable decision is reliably made.
> They have no intelligence. These are very very very refined prediction engines.
"Very refined prediction engine" is not a bad working definition for intelligence. It is not the only component you need, but it might be the most important over-all capability.
If its making lots of errors, its not doing a very good job of predicting the outcome of its actions.
It's a terrible definition of intelligence. Intelligence is far more than just predicting things based on a massive set of similar things. We do do pattern recognition, but I don't need to have seen 10k dogs to be able to recognize a dog. And it's not a difference of degree, either. "Meaning" is something I am able to derive from my experiences, and it is not something an LLM does or will ever be able to do (nor is training data even analogous to having experiences in any way shape or form.)
I didn't say anything about training methods. I said an intelligence can predict things accurately, most particularly if it can predict the outcomes of possible actions it can take then it can do planning. Its especially important that it should be able to predict outcomes in cases that are not exactly the same as things it has seen before, but how it gets to that capability level is not important for understanding what the capability unlocks.
This is one of the most distinctive qualities of human intelligence compared to many animals - is our ability to adapt to new situations and make accurate predictions of the outcomes of our actions. Other animals can also do this but over very narrow time horizons and situations.
AI agents are better at it than any animal, and better at it than humans in some domains.
Is "meaning" inherent to intelligence? Is emotion and the ability to have a subjective experience inherent to meaning? Genuine questions that I'm not sure we have good answers for yet.
Your other point that humans need very few examples for learning vs. what LLMs need is an interesting one. A few researchers talked about this very thing on the most recent episode of Dwarkesh's podcast.
My own view is that it seems unfair to compare an LLM's training with just what a human gets over the course of a lifetime, because our brains have been trained by a billion years of evolution for pattern matching. I'm not at all confident that AI models won't catch up.
> I am in the process of attempting to have AI run my business. I'm actually making very good progress, but it's happening in pieces
Very cool! What's been the hardest part? Have you successfully automated non-trivial communications? (for example with prospects, customers, vendors, partners, etc.)
> maybe an agent starting a business from scratch would have a better time
This describes the project I've been working on since last year, with agents in company roles, deciding on strategy, collaborating, and making progress (admittedly nonlinear). Some parts of the platform are stronger than others. So far the company agents have developed a company strategy and plans for executing it, launched a website and blog, and built and operate two products, one free and one paid.
But I would be wary of using a third party like Pion as a platform for running a business. It's one thing to know that your prompts and responses will be used to train models in the future, by companies that have a huge sea of relatively unstructured data. But it's another to hand over to another company every aspect of your strategy and operations, available to that company in real time and structured in such a way it's quick and easy to understand what you're up to and where you are going. Especially if it's being offered for free and at scale. With a free offering, value creation will likely come from either customer data or from escalating prices once customers are locked in. And even if neither of those happen, you now have a huge single point of failure for your entire organization. Seems like there are lots of strategic risks in there for the users.
Cool that he's writing about this publicly, but for someone that's so business-minded I find it odd that he never renders any of these AI experiments in terms of ROI. Hard to find any signal in those posts on whether it's profitable to automate these processes with AI vs. alternatives.
That's a fair criticism for some of it, but in a lot of cases I try to automate things that don't have alternatives because they're unique to me (e.g. all the brands I own in my fund have one bank account but need to be tracked separately for investor payback purposes, so I have a Claude skill that takes all of my transactions and assigns them to the various Google Sheets ledgers I have).
In other cases I'm doing stuff that's complex enough that there's no real "alternative" at this point that doesn't involve hiring one or more people and working with them for an extended period. My next set of posts is going to be about my latest acquisition and how AI redesigned a Shopify store then designed and launched Meta ads, all of which only took a couple of days but is going to easily push revenue up by something like 40% this year. The money made is great, as is the money saved on people I would've hired to do this sort of thing, but the real cherry on top is that even hiring someone would've required me to spend waaaaay more time than I did here. So big ROI on the money but also the time, which is arguably more important, since time savings permit me to acquire more brands.
Ah, so it’s ”No Silver Bullet” (1986) [1] all over again.
Thinking out loud here...
AI mainly helps reduce accidental complexity. It can help one understand essential complexity, but essential complexity must still be paid.
You describe doing the essential, irreducible part. And that’s specific to your needs, depends on the problem you’ve chosen, understanding reality of your domain, evaluating tradeoffs, and being accountable for the outcome.
For anything visual, they are still effectively blind, right? Still just working off the image embedding, sometimes using scripts to actually inspect individual pixels.
Blind seems too far here - Claude's increasingly able to diagnose visual issues on its own. I used to have to check every single image it generated, but now it's at the point where it'll catch most bad ones and regenerate them before it gets to the review step. Still misses some, though.
no. I just used Gemini to update a website I'm managing for a charity, and one of those steps involved Gemini, on its own, offering to scan the website for one of our beneficiaries, specifically the header images, to see if there was more information to include in our writeup about the beneficiary. And it did it quite well.
Mostly Claude Code and a lot of internal tools that it's built. I've honestly still not had a chance to play with those kinds of harnesses yet, but my impression is they're just a better way for you to instruct an assistant to take care of things in order to get them done. My goal isn't to be managing an assistant, it's to get things automated without my input (and where my input is needed, have the assistant escalate to me).
And for the record, I only buy brands that sell high-quality products. Marketing and operations I can fix, but if the stuff they sell is no good, there's no point.
How do you inspect the quality of the products you're selling? From your blog it appears you just read through their sales information, and don't ever see or handle the product. And yet we all know Amazon sellers juice their sales metrics and ratings with scammy practices, and now I see selling their brand is part of the incentive.
And more, you are AI-generating listings and images for your acquired brands. You're filling Amazon with slop, and automating the process.
You should seriously reflect on your role in polluting these marketplaces.
Of course I do. No one buying a business would do it without that information. I get their P&L before I even get on a call with them, and during the due diligence period I get access to Shopify, Amazon, etc. to get first-party data to validate their claims.
I suspect you mean this as an insult, but it's not too far off. That was one of the main points of the business - each one is simple to run, so I can manage a lot of them. And that was before AI got really good.
Anyway, I'm sure the app you makes that lets you get a phone number that can send image messages is very complex, and I think that's very nice.
Open to sharing what workflow you're using on the lifestyle image creation? I've tried the Higgsfield MCP, and direct Figma integrations - but keep running into hallucinations on product dimensions, etc.
One-man show here - I own a dozen small e-commerce brands. I'll assume you're just taking away AI, but not the tools I've built with AI.
I've got all kinds of useful tools that do helpful things, like check to make sure Amazon hasn't lost a bunch of my inventory or suddenly deindexed my product from search. These sorts of things happen without warning from time to time and are trivially fixable if you catch them, but you have to be constantly monitoring 10 different things to catch anything that could go wrong in a timely fashion.
I've similarly got an inventory/cashflow management system that's deterministic that would be fine, but some of my other tools, like one that scrapes my Alibaba/WeChat/Whatsapp/Gmail supplier convos to keep my order status board updated, is finicky. I wouldn't be able to repair that once it broke.
For some tasks, I'd have to go back to freelancers. At this point, when I take over a new brand and want to refresh images or want to A/B test new ones with an existing brand, I can do that entirely with image gen. No AI would just send me back to Upwork.
For other tasks, I'd just have to do them myself. Right now when I acquire a brand, I have a Claude skill that runs through all of the Amazon search and ads reports to find the key search terms, analyzes the text of the listing and proposes updates to make sure we're hitting the ones that matter. I used to do that myself, but it takes somewhere between 2-10 hours depending how many products we're talking about.
The biggest loss would be the inability to expand. I started out just buying Amazon-only brands because Amazon provides a crazy amount of infrastructure that enables me to run a lot of them by myself. My most recent acquisition came with a Shopify site that had minimal sales plus a Meta ad account the guy who sold me the brand had done a little experimentation with. A year ago, I would've just discarded those, since my time is the biggest bottleneck. It takes a lot of time to run your own site and Meta ads, so it makes more sense for me not to bother and instead just make more Amazon-only acquisitions.
But now I'm having a great time really pushing Claude and ChatGPT to do most of the work. I redesigned the site more or less by just having ChatGPT do research on competitors and best practices, then having Claude run with that. That took maybe eight total hours of work.
I'm spinning up an enormous amount of creative to test on Meta, and I had ChatGPT research and spec all of the kinds of landing pages that are worth trying. Claude implemented them and set up the ads using Meta's ads APIs. I got all of that done yesterday.
So yeah, basically if I couldn't use AI any more, I'd get back to the original thesis of my business, which is buying small brands at reasonable prices and operating them as leanly as possible. It still works, but it'd be a lot less fun and profitable.
I started an e-commerce brand on a Shopify site. I swore to myself I would never put up one of those stupid things that pops up "Someone bought X product an hour ago!" messages in the corner of the screen.
I ended up trying it. Boosted conversion rate meaningfully. Worth the price I pay in mild self-loathing.
That's the real problem. Being aggressively annoying works. Dark patterns work. Popups that grab your attention work. Flashing text conveying urgency works. Thumbnails with open-mouth YouTube-face work. Moving buttons around so people mis-click works. Tiny [X] buttons that make people accidentally click an ad work. You're never going to convince someone to stop doing something that is making them money.
Does it work because you're taking advantage of a group of people who are extremely vulnerable to manipulation? People who already struggle with impulse control, who are prone to making bad financial decisions? The elderly, kids?
We have to escape this mentality that anything that makes money is valid, that the money itself is the validation of "rightness".
Some of us, just some think that maybe enterprises that prey on the vulnerable just don't deserve to be in business. Otherwise everything might as well be payday loans and online gambling. I'm not saying OP is definitely in that category, but I would encourage them to think long and hard about weather or not using dark patterns to goose their sales is really the kind of world we want to be fostering.
Well, if you own an online casino, I guess your moral compass is already so skewed that you simple don't care. On the other hand, if you own an online mom & pop store that sells clothes for puppies, you rationalize that the impact of those manipulations is not that big of a deal. You're not ruining anyone's live by the single act of persuading them to purchase $100 worth of apparel for mr. pancake.
I agree with you too, but I don't think it's necessarily a gross over-exaggeration: it's possible that the majority of the buying population are extremely vulnerable to manipulation.
"Average" doesn't necessarily mean half way between two extremes - the average human might actually be all of:
- extremely vulnerable to manipulation
- struggling with impulse control
- prone to making bad financial decisions
- the elderly / kids (<16/>60)
Based on age demographics, rampant consumerism and social media addiction, especially in those age demographics, I'd be more inclined to guess the average human is all of those things.
> We have to escape this mentality that anything that makes money is valid, that the money itself is the validation of "rightness".
I agree. But.
I run an online business. It's tough. I scrape by. GP noted that the skeezy popup "Boosted conversion rate meaningfully". So that's real money in GP's pocket that they presumably use to have a nicer life. At what cost? Some popup? I mean…
Would I take more money in my pocket? Boy, I'd like that. So far I haven't gone the skeezy popup route, but the day I try, and more money ends up in my pocket, it's gonna be hard to turn that thing off.
Yeah, I mean at the end of the day there is clearly a spectrum of conversion rate optimizing stuff you can do, some of which is fine and some of which is unethical. A/B testing the color of your add to cart button and using the one that converts best? I think it'd be a little crazy to argue that's unethical manipulation. Lying to customers about the product to get them to buy? Obviously bad.
Silly obnoxious social proof popup is clearly on the ethical side in my view (my issue was never ethics, just that I personally find them very annoying). I'm not misleading anyone (it does in fact show real purchases) or forcing anyone to do anything. If that's the difference between you buying or not, I still feel perfectly comfortable that you made the choice of your own free will.
On the extreme side, things like blackmail and fraud are generally illegal, even if they can make you money. Most of these dark patterns that "work" tend to follow a similar pattern: by taking advantage of the target, one can extract more money from them.
I would possibly be a terrible salesperson, because these all give me the ick. Your product should provide an offer of genuine value.
That's the real problem: selling drugs works! Forming a cartel to produce and distribute addictive products, killing anyone who gets in your way? It's effective! You're never going to convince someone to stop killing people and dumping their bodies into culverts when they're making money. It's just not reasonable to expect that, or do anything about it.
(edit; not directly responding only to your comment, also to the parent)
Advertising is shitty but how would you realistically even make it illegal without violating free speech.
Also, advertising does have a good-faith purpose in functioning markets.
Failures in regulation IMO are mostly at entirely different levels.
E.g. monetization of public utilities and public space. Noise, brightness, and of course deceptive tactics could all be better regulated as well.
But advertising, including its diffusion into general pop culture and entertainment, is older than the printing press, and I think for a broad sense of "advertising", even a lot older.
Corporations are not natural persons and do not have constitutionally guaranteed free speech rights that can be violated. Any individual shilling for a corporation can be held personally liable for damages.
And we already don't allow everything that is technically speech. Harassing people is already illegal, unless you're a corporation. Taking advantage of the vulnerable is already generally illegal, unless you're a corporation.
Thank you for the follow up link, it brings up good points. I think the 1st Amendment angle is too quickly brushed off by the author and not sure how this would be settled in a legal dispute--especially considering the means and depth of advertising and tech companies. The article is worth reading and great food for thought.
Isn't this a part of legally excluding under-18s from using social media?
Seriously. I did not consent to these companies to be manipulated into giving them money. It's an invasion of the public sphere, and worse it is highly authoritarian which should always be resisted on principle.
Very much agree, but also (not saying you're saying otherwise) stealing, lying, entitlement (which is possibly just stealing and lying), and scamming "work." These are checked in large part by a person's aversion to it not just by its lack of working. Of course dark patterns work. If what one says increasingly inches towards 100% misrepresention (but shy of it), people will "mistake" what's being said. "Mistake" doing a lot of blame shifting.
"You're never going to convince someone to stop doing something that is making them money." Reasonable, but I would soften from "never". There was less of it at one point—and it seems logical to guess we have less today than we'll have tomorrow. The main reason is likely that we simply didn't know about these tricks yet, but somewhere below that on the list of the reasons is that some people dropped off from doing it at lesser forms of misrepresentation. Or they made the case against it at work resulting in them either winning (and their projects perhaps did less well) or them losing and being overrun by those more willing. With losing possible also leading towards leaving, not getting promoted, or getting fired. This is just a way to say that there are people who do forgo money, they just might not be around or visible for various reasons. And, as implied by the difficulty of convincing people to not make money, their (former) coworkers prefer that on some level even if they don't believe they agree with stealing or lying. But losing or earning less money is not the same as having no choice.
There used to be less data to optimise from. Doing ordinary business is a low risk strategy, compared to dark patterns. Now we have data to enable us to do dark patterns without risk.
This reminds me the clickbail title and thumbnail on YouTube.
One of my favorite channel apologized for it but explained that the difference compared to NOT doing this is phenomenal and they can't afford stopping.
If this is the only way to get people to watch your videos, maybe your channel sucks. But glad you are making money tricking people into wasting their time so you can get one more ad impression.
Sounds like you've let your feed fill with shitty channels, but no, not all channels do this. I mostly refuse to watch videos with those cringe clickbait thumbnails, and I've still got more stuff to watch that I've got time for. And it's mostly gaming and variety streamers, or music/animation artists; not serious stuff.
Now, I'll grant you that these may not become as successful as the clickbaity ones. Then again, the latter channels might as well not exist to me and like-minded people.
Just an hour ago I unfollowed a streamer precisely because he's been getting clickbaity. We've got to take some responsibility over the people we choose to watch.
I don't even know what you're trying to convey: I'm countering the perception that every channel does this. But it isn't true, they've simply allowed themselves to have a disgusting feed.
The "players" can do whatever, I don't care, because I don't watch them; they barely even show up on my algorithmic feed anymore. No one has to watch them, they simply chose to and then pretend they didn't have a choice. I just want people to own it, these channels do clickbait because many people like it.
My point is simply that you might have habits that made you filter out those channels, but the stats show that using clickbait thumbnails and titles has an impact that tremendously change the views.
Youtube creators are spending sometimes 1/3 of the time of production of a video on the thumbnail, with a dozen of variations in A/B testing. They have to optimize because the basic user decides in the first 5 seconds if they will continue watching. And Youtube punishes content creator if they notice that those users don't stay more than those few seconds.
I believe the vast majority of the creators (the "players") hate having to resort to those, but they just don't have a choice if they want it to be sustainable.
Out of all the annoying website things, always thought that one was pretty tame. Almost feels like a spiritual successor to the visit-o-meter. Obviously as long is it doesn't put a (1) on my tab or makes a noise and doesn't steal mouse focus.
Gosh, I always assumed they were factual. Now I'm wondering the factualness of the old "15 people have this in their shopping cart" and "This is a popular item - limited stock left".
I always assume they're lies. In fact, I default to assuming every marketing tactic purely presents lies and the goal is to get my money by any means possible.
Entirely factual. It's "Someone from <city> bought <product> <period of time since purchase> ago!" with a picture of the product. All purchases are real.
Is there a law of society / business where everything we hate is on average something that made the society able to function ? it's similar to chesterton but it's not a fence, it's the tree we live on.
I actually kind of like those, though I've always been suspicious that those events are fabricated because that's the sort of thing marketing people would do.
The last time somebody made this thread, I posted about https://the-waterline.com/, a site I launched as a side project/eval to see how well Claude can do a decent-sized project (mostly) independently. It pulls public government data about water levels in the western US, updates the website and sends out a newsletter.
It started getting some search impressions and clicks, so I decided to launch the next 9 of the top 10 ideas that ChatGPT came up with for this type of site - won't post all of them here, but https://the-dwell.com/ (data on ship and truck dwell times/port congestion) is doing the best thus far from a traffic perspective.
First I had it send me a weekly analysis of the GA4/GSC data to see what could be improved about the site, but I've since automated that - each week it reviews traffic and search queries, then updates articles or creates new ones to meet demand. So now in theory I have a full engine to launch these sites and have them self-improve over time. Kinda neat to see all of that being done autonomously.
The thing that these sites really need are some backlinks, so I'm thinking through how to get those. I had ChatGPT generate some lists of places that might find real value in linking to these sorts of things, but I'm unwilling to let an AI agent start emailing people. I'm going to try sending some emails to some of the places that seem like a really good fit, see what the response is, and if it's good try to figure out how to do that on a larger scale without AI email spam.
Pretty cool apps! One thing i'd love to see for both is a historical view of the data - how do things change over time? Where is today relative the past five years for a given place? ten years? etc.
Thanks! There are actually five year charts for most of the data, but you have to click into the page for a specific dataset (e.g. https://the-dwell.com/anchored/east-coast).
But you're right that it'd make sense to have an aggregate view on the level above (e.g. https://the-dwell.com/anchored). I shall go command Fable to figure out all the places where this would be appropriate and add it.
"The study, authored by Capraro with Chiara Marcoccia of École Normale Supérieure and Walter Quattrociocchi of Sapienza University of Rome, deliberately used questions where AI models typically fail: visual details from films, such as the colour of a team’s uniform in Bend It Like Beckham."
I get why they used questions where AI models fail, but it also really reduces the value of this study. Nobody is really asking AI the color of a team's uniform in a movie, and if they do and confidently get it wrong, it just doesn't matter at all.
Asking trivial questions also feels like it would affect the rate at which people are willing to confidently say things that are wrong. If you ask me some question of pointless trivia and I ask ChatGPT, I'll probably just repeat the answer because who cares. If you ask me something even mildly important and I ask ChatGPT, I'll either verify the information before I repeat it to you, or I'll qualify that I looked it up with ChatGPT and didn't verify. But some things are just so unimportant that they don't even warrant the disclaimer.
Yeah good god OP please start writing a Substack about this! Engineer who buys a run-down bowling alley and cleverly fixes problems is exactly what the internet needs more of.
I don’t even care that much about the email thing, and most newsletter sites make it easy to have your articles available on the web, but Substack specifically is a horrible company that needs to die. I refuse to read anything hosted on Substack on principle.
I heard an episode of the Odd Lots podcast about HayWire (haywireag.com), a site that pulls public data from government PDFs + APIs, uses LLMs to parse it and turns it into an easily readable website that has all of the latest info on hay prices.
The host made an offhand mention that there's probably a bunch of other similar sites that could be created with all the of useful but difficult-to-access government data out there. That sounded interesting, so I thought I'd give it a whirl!
All pretty fascinating topics to learn about, plus it's been interesting to see how much of the website setup I can fully delegate to Claude. With Cloudflare to buy domains and put the sites up, a Google Service Account with access to Google Search Console and GA4 to create those properties and a Buttondown API key for weekly email sending, it's almost all hands off for me. Though it refuses to take control of the browser and create a new Buttondown account, which I was surprised is a red line.
Thanks! I am definitely not an expert in either, and I've run the content by both Fable and GPT-5.6 with instructions to make sure it's written in such a way that it would read normally to people in the industry. They assure me the wording makes sense in that context, but we'll see if the sites actually get traction or not.
Not an email newsletter service but a contact form. I recently added the ability for LLMs to sign up to my service https://www.simplecontactform.org autonomously via API. Curious to hear your experience if you can ever make use of something like it.
The Limitations section at the bottom certainly has a lot of limitations:
> This paper is a review, meaning it synthesizes and interprets existing research rather than presenting new experimental data. The authors themselves note that current visual tests for susceptibility to discomfort are subjective and poorly standardized. They also acknowledge that the proposed mechanism (that discomfort is the brain’s response to overwork) has not been fully tested, particularly the hypothesis that colored tints reduce discomfort by steering visual stimulation away from overactive brain areas. The relationship between the brain’s excitatory and inhibitory chemical signals and visual discomfort also remains, in their words, “unsettled.” Several key research questions are flagged as unresolved, including how to best quantify the real-world impact of visual stress on people’s lives and how to objectively measure susceptibility.
Flickering lights are about the only thing I saw in here that seem like they'd be a problem in the long term. Everything else your brain just adjusts to over time and stops noticing. Maybe the first few days in an office with bright colors would be slightly distracting, but after that you just stop seeing them. I would guess that a lot of the studies they reviewed probably tested people's reactions to these things when they saw them one time, not the hundredth time.
The article does explicitly state that the brain doesn't adapt to this.
From the article:
"And when the brain encounters something it can’t process efficiently, it doesn’t simply adapt. Brain imaging studies cited in the review show it generates stronger neural responses in visual areas, consumes more oxygen, and in some people produces pain, distortion, or worse."
> And when the brain encounters something it can’t process efficiently, it doesn’t simply adapt. Brain imaging studies cited in the review show it generates stronger neural responses in visual areas, consumes more oxygen, and in some people produces pain, distortion, or worse.
If the studies are of a person's initial exposure to these sorts of conditions, then that doesn't tell us anything about whether people adapt over time (and to be clear I have not read all the studies, but given the limitations listed I'm comfortable assuming they're not incredibly robust until someone tells me otherwise). I suspect the article's use of the word "adapt" is not the same as mine; from the context when they say the brain doesn't adapt they just mean that it shows a response at the time of the particular exposure they're measuring.
I think there were studies on this, leading to, among other things, painting control rooms seafoam green to reduce visual fatigue. This implies that people don't simply adjust (or that the studies were too limited).
I own a dozen Amazon brands that are probably largely of the kind that OP would want this to get rid of (sourced from China, not name brands by any stretch, only sell on Amazon). For the most part I would say this extension is not a great idea (obviously very biased!), since I purchase brands that have high quality products that typically have pretty poor branding/online presence that I can improve. My stuff is very frequently of the same quality (and sometimes from the same factories) as much pricier stuff but at a lower cost. To some of those suggesting you can get this stuff on Aliexpress, in some cases that is true, though of course the big benefit of buying from Amazon is that there's no risk to buying no-name stuff because if it's junk you can return it.
In any case, I gave this a try to see which of my brands it would filter out. It's weirdly inconsistent.
One of my brands was filtered out because there's no brand name at the beginning of the listing. That's just an outright bad rule, because Amazon generally decides whether or not the brand name appears first. This brand is trademarked and has Brand Registry, so it qualifies for that treatment, just not getting it right now. Also, a number of other brands without brand names did not get the same treatment (and these are very much the type of products this is designed to filter out).
On another one, it misunderstood the product model, which is at the beginning of the product name, as the brand and hid it based on that. That one was a bad one because the model is only three characters, which is extremely unlikely for a brand name.
One product I sell is a hunting accessory, so I did some searches there. It hid everything by the brand KUIU, which is a well-known and very high end hunting brand. Definitely wrong there.
So yeah, sort of an interesting idea, but the execution is pretty sloppy and the creator clearly doesn't have a full understanding of how Amazon listings work.
Another PL seller here with a few brands and a couple hundred products:
"No brand name" flag is tricky because the Amazon catalog team actively does A/B tests to hide brand names as part of their goal of commodifying all the sellers to increase price competition, when they see you're selling a commodity item.
Same goes to wellknown brands that get caught in the crossfire because they're using their brand name from another language but don't make sense in English.
Agreed that it's an interesting idea, but execution has a LOT of false positives.
> It might seem obvious to coders, but the difference between Claude Code and Claude.ai's chat is enormous, even if those two run the same model.
In my experience, Claude Code is vastly better for doing tasks, writing code, etc., but Claude.ai is better for analysis and high-level planning. When I'm working on a new project, I've started using the latter to do the initial planning, get feedback and draw up a spec, which then goes to Claude Code.
For this project, I probably would've done something similar - use CC to get whatever you need out of the image files, but have Claude.ai do the actual review/diagnosing.
Either way, I often think about how far behind most of the world is in really understanding AI. The overwhelming majority of people would never guess that you get vastly different outcomes from the exact same model in a different harness (tbf most people don't know what a harness is). I spend hours every day using AI for a broad range of tasks and still feel like I know a fraction of what there is to know. I haven't even tried the new GLM model (or really any of the open source Chinese ones of the most recent generation). With so many people thinking that the free version of ChatGPT is SOTA AI, a lot of folks are in for a very rude awakening at some point soon.
The experience of getting it there makes me pretty skeptical of the idea of a general business agent like this. First, because I still find myself having to review some categories of work for errors. These are decreasing over time, but they're still there. I fully believe that as models get better, errors will decline, but I am somewhat surprised to see some of the errors current models make given their intelligence. A common set is having Claude take some product photos and turn them into lifestyle ones using ChatGPT via Claude in Chrome. It's a pretty well-honed workflow at this point, but it'll still return images where the product is obviously not correct and seemingly not notice them.
Anyway, even if the agents are "perfect" in terms of their ability to execute tasks, there's still just an enormous amount of nuance and context in each business that takes a ton of time to convey. I've been at this for a couple of years now, and I'm still clarifying things. Now maybe an agent starting a business from scratch would have a better time since it's not inheriting all of this, but to have one run an existing business requires a very extended handoff, even if the agent is objectively amazing at all aspects of running a business.
reply