Category: links

  • This absolutely sends me

    https://gruhn.me/blog/2026-08-03/

    Too often I ask a question in Slack or leave feedback under a merge/pull request or argue with friends in a WhatsApp group and get back:

    > Claude said: [giant response verbatim]

    Please don’t do this. I mean, I’ve done this. But I’ve been on the receiving end too many times now. This is not adding value. I can talk to Claude myself. It’s going to be faster and I get to control the context. I don’t need a meat proxy in between.

    Honestly, if you do this I will 100% drag you. It might not be right away, but it will happen. Probably when you least expect it.

  • Understanding the amount of bullshit work

    AI is going to help humanity by revealing just much human time is spent on bullshit work.

    Consulting giant Accenture is trying to figure out how to stop non-technical workers from blowing through companies’ AI token budget on trivial tasks like converting PDFs to presentation slides, according to leaked audio obtained by 404 Media. Across the industry Accenture is seeing “soaring token spend,” according to the audio.

    The Tokenpocalypse Is Here: Companies Are Scrambling To Stop Spending So Much on AI

  • Local journalism

    South Shore news: AI-generated newsletter has paying audience (Boston Globe)

    This feels like the future. AI agents, directed by assignment editors, gather local intelligence and context. Other agents draft and edit stories. Humans handle quality control and accountability. Meanwhile, separate software tracks traffic and prominence, feeding signals back to adjust agent behavior — optimizing for the business goals or cultural impact that publication leadership is after.

  • The new skill: task imagination

    With models that can run for days, NLW argues we’ll all have to up-level ambition — and become token-efficiency optimizers who match models to use cases. Nate B. Jones’s framing: most of us have nothing that’s ever taken even an hour on AI, so the scarce skill is imagining tasks worth handing to a model that works for days.

    — The AI Daily Brief · https://aidailybrief.ai/e/2026-06-10#task-imagination

  • Be YouTube, not Qwest

    Today, no amount of model training is too much. No price for that training is too high. The builders bet is that inference will be orders of magnitude cheaper in 5 years. Build for that. Be Netflix or YouTube, not Qwest.

    This year, American tech companies will spend $300 billion to $400 billion on artificial intelligence, which is in nominal dollars more than any group of companies have ever spent to do anything. Notably, these companies are not remotely close to earning $400 billion on artificial intelligence.

    That’s why you’re starting to hear some people wonder whether the AI build-out is turning into the mother of all economic bubbles.

    The prospect of an AI bubble should scare us. Roughly half of last quarter’s GDP growth came from infrastructure spending on AI, and more than half of stock market appreciation in the last few years has come from companies associated with AI. If the AI spending project blows up in the next few years, as our next guest says it might, the implications for technology, the economy, and politics would be immense.

    This Is How the AI Bubble Could Burst – Plain English with Derek Thompson

  • Friction, difficulty and independence

    I think about this a lot with my own kids: making sure they have the right amount of friction in there lives, having time and space to be bored, and to even get into low-grade trouble.

    School hasn’t been a problem so far (acknowledging my privilege); they get challenge from other places. This piece is a good reminder of how difficulty – and potential boredom – are important parts of growing up.

  • Banks are slow-rolling open banking to protect payments revenue 🤯

    Financial institutions offer their customers a complex bundle of services.

    You might reasonably expect that Open Banking is a fight over the budgeting app space. The banks have, via the magic of account records, a large portion of the underlying data about a household’s finances. You could imagine software using Open Banking to allow it to slurp in transactions and then categorize them. That would compete against the lackluster offerings the large banks have in their apps.

    But Open Banking is not actually a fight over budgeting apps. Banks don’t make money on them and the best known standalone budgeting app, Mint, was acquired for a relatively small amount of money.

    Payments, on the other hand, are an enormous business. They are monetized both by banks and by a diverse ecosystem of fintech providers.

    The data banks find it annoying to make Open are, principally, account numbers. This is because, due to the long shadow of checks, possession of an account number (plus the routing number, identifying the bank) is sufficient to attempt to debit a bank account. Direct account-to-account transfers, including “pulls”, are a common payment method in many countries, but they are not a large share of consumer to business payments in the United States.

    Why not? One reason is that the user experience of asking someone for their account number is pretty awful. There is no way to check in real time whether an account actually exists. Credit card numbers, in addition to having infrastructure which allows you to query them in real time, are specifically formatted so that typos in them are easily catchable.

    Since you can’t know whether the account exists you certainly can’t know its current balance or whether a transaction posted against it today will succeed in a few days or be reversed for insufficient funds (or another reason). This means that businesses which use account transfers as a payment method would frequently suffer credit losses if they released goods or services at the time of “payment.” For many businesses, that isn’t a worthwhile tradeoff.

    So they keep using cards. Cards give much stronger (but not foolproof) real-time guarantees of funds availability and likelihood of a transaction going through successfully. The ergonomics of card acceptance, at the register, through your phone, or in a web browser, are also much more palatable to most customers.

    Several fintech companies, including Stripe, realized that they could use Open Banking to make account-to-account payments something customers would actually enjoy. The user is prompted at checkout whether they’d like to pay directly from their bank account. They log into their bank account and grants the fintech read access. This is a much stronger signal of authorization than simply knowing an account number. (We print those on every check, after all, and a check is designed to be handed to a cashier or waiter you’ll never meet again.) The fintech then grabs the account number and perhaps e.g. looks up the current balance.

    Then, they can pull money from the account, through an ACH debit.

    The ACH debit itself is not Open Banking. It is the ordinary operation of existing payment rails in the financial system. The ACH debit was just made much more convenient by Open Banking.

    Open banking and payments competition, Bits about Money

  • About Fox One

    News is very ideological. If you’re a diehard Fox News fan and that’s all you watch, and you don’t want your money going to CNN or MSNBC, you can take your bill down from $100 to $20

    So Fox One isn’t so much about helping Fox as it is about crippling CNN

  • Where are the customers’ yachts?

    The steady evolution of private equity into yet another set of high fee products for dentists and widows continues.

    Equity values are fleeting; fees are forever.