AdamJudelson

@askchapter My pleasure, was super fun to learn about how Medicare really works and the great impact Chapter is having.

Great conversation, thanks @Ajuggles. Link below.

Happy #cityhallselfieday! We’re building for you, civil servants of the future! And for the public too!

I think @Rippling meant to add “training” at the end of this sentence. Make sure to check your product UI copy not just with test data but real data too!

Extreme optimism and extreme pessimism both converge on laziness says Peter Thiel in a recent podcast episode. If you believe everything will just work out, no need to have agency yourself. If you believe nothing will work out, why bother trying? The best leaders live in the middle. Lots of optimism for what can be, but with great empathy for what is and how it must change to get there.

Think for yourself... Ask AI to think for you... Think together with AI. The smartest people are doing all three.

@karantalati Exactly, each new bet deserves the same crucible that a founder would apply to it.

Don't set OKRs if you want to launch a new product. Don't create sales collateral to train your sales team for a new product. A new product is a new startup. If you are doing anything differently to launch a new product than when you were hotdesking at a coworking space, or if your team is asking for anything else, you're doing it WRONG.

Building a brand so strong that people want to buy souvenir versions of the checkout experience is next level.

"AI is a type of labor" says @ssankar in his most recent talk. Love that framing for conceptualizing its power beyond thinking power and into its operational power.

@MichelleVolz Maybe the same, but on a private jet. :) The JSX experience is faux private and lovely.

There is no greater high than getting out of an uber and walking straight through security and onto your plane as they are closing the door. Bliss.

@clairevo Classy that they sent an email to users to inform them of this though vs the much more common silent sticker shock model most companies follow.

me: gpt-4.5 can’t be *that* expen… 😳

@clairevo The more the models reason, the more I find myself instructing them not to. Have you tried the thinking style chooser?

Claude x Sonnet-3.7 is a little too thinky for my tastes.

@alliekmiller It’s time for many of the labs to bring in some thoughtful product people. OpenAI and Anthropic both made fairly recent smart moves to start closing this gap, but culture takes time to change.

AI labs—both in public products and private products I'm testing—are not thinking enough about artifacts. If you're generating a report, I need to share it with others. Make it a PDF. Make it downloadable. Give it a title. Auto-attach to email. If you're generating research, I want all the links together. If you're generating 100 marketing captions, I need a CSV. And eventually, for agents, I will want the proactive action. If I'm generating a workout plan, offer to block off my calendar. If I'm generating a menu for a dinner party, offer to place all the items in my Instacart. Stop getting stuck on cool tech. Think about actual usage.

There is nothing more fun professionally than building something you’re passionate about with other amazing people you trust and who feel the same way.

@karantalati They actually have this feature but only in their AI plans, which basically doubles the cost of your company’s subscription, making it tough to experiment with. slack.com/help/articles/…

You know Slack will be displaced by another chat tool because it doesn't have an AI-powered "Summarize thread" feature yet.

@lennysan She is excellent! I can personally vouch.

Looking for an A++ copyeditor? DM me. I've got your person. Note: She's not a writer or editor—just a straight-up world-class copyeditor. She'll catch all of your typos and grammatical issues. If you've never noticed a typo in my newsletter, that's not because I'm that good—it's because of her. She catches and fixes 50+ issues every post 😵‍💫 She's worked at top tech and SaaS companies, and science, arts, and culture publications, and is both looking for a full-time role (ideally remote, but SF-based possible) and also accepting new freelance. DM me if interested in connecting.

@Kazanjy Also useful because if the deal pace slows down, no one will remember what was discussed a month later.

Founders: Document every verbal agreement immediately via email: 'Great call! Confirming what we discussed: - [X] seats at [Y] price - Implementation by [date] - Payment terms [Z]' Prevent 'selective memory' issues later.

@lennysan @Anthropic AI body dysmophia is real!

"In the early days of @Anthropic, we taught the model, 'Hey, you actually don't have a physical body to operate in the physical world.' But at the same time, it was being trained on data that told the model, 'This is how you set an alarm.' And so the model would get extremely confused about whether it could set an alarm— that it doesn't have a body in the physical world." — @karinanguyen, Research lead at @OpenAI

I often half jokingly give @OpenAI a hard time for focusing so much on the technology that the user experience is lacking, but today they surprised me. They've added a pretty novel set of quick actions to shorten answers, adjust the reading levels, increase the polish...and then... There is a button to emoji-fy the post. This seems utterly useless (or maybe I'm the one who is out of touch in not rushing to add emojis all over my writing)...but it's not... It's a way to show that their brand isn't 100% serious all the time. It's a way to get people talking about the technology, and it's sort of fun if you don't use it on a super long response. Good for you @kevinweil and team in allowing some light hearted stuff to shine through amid the normal 10,000 lb. brains features we've come to expect.

I think about this almost every day

“Making AI that is smarter than almost all humans at almost all things… is most likely to happen in 2026-2027.” - Anthropic CEO Dario Amodei