@Lloyd

We had a good run 🫡

@benjitaylor That is sweet. Can we do something like $LON:RR (Rolls Royce) for other exchanges

You can now embed Cashtags directly within posts: $TSLA

b2b sales reps on their 10th email in a row:

Did anyone vibe code a better Fuji camera sync app yet

@benjitaylor any chance you could do something about the bookmarked folder flow? on mobile for example I'd love to be able to tap and hold the bookmark to jump to folder picker on desktop it is also clunky, multi-step, wait for the toast, click add to folder etc

Got a chuckle out of reading Brian’s Coinbase layoff memo where he says every manager will have 15 direct reports and also be an IC This is what I mean when I say that every tech executive has AI psychosis

🚨 New Experiment: Everyone thinks AI firms will look like little companies. A manager model decomposes the task and worker models do subtasks. The manager red-teams, revises, and recombines. A seemingly simple org chart. But when I ran the experiment, the current in-vogue org setup, manager-subagent, cost 4x more and performed worse than letting a rather simple market do the trick. I tested 3 ways to organize multiple AI models: 1. Solo: Onefrontier model does everything itself 2. Hub-Spoke: A "manager" model splits tasks, delegates, red-teams, revises 3. Market: Models bid on tasks, winner gets the job, reputation updates I also tested were 3 types of tasks - Coding, Reasoning and Synthesis. - Coding required most "global state" management, which the solo model did best at. In future @a1zhang's RLM will probably do even better here - Reasoning is the hardest to cleanly decompose, and the market worked the best here - Synthesis too, the market beat hub-spoke as the framing could be ambiguous The reason is, a hub isn't a "manager" as we know it. It's a model that must somehow know: - What the subtasks are - What good recomposition looks like And if either fails, as it does for complex or not-easily-decomposable tasks, competent workers still produce garbage. As we move from coding to letting multi-agent systems do work across the entire economy we'll end up with more not-easily-verifiable tasks with ambiguous settings and uncertain payoffs. In those, we won't be able to use the factory approach to get work done. The Coasean argument is that firms will get smaller, and the smaller firms will transact more, since the organisational premium reduces with AI. But how? Through central hubs, or markets? The fact is, Coase here needs Hayek. Setting up markets is not trivial, as @AndreyFradkin and I looked in our recent paper. Essay: strangeloopcanon.com/p/why-smart-pl…

@jwegener I would likely reach for an email worker? developers.cloudflare.com/email-routing/…

Is there an easy way to give my Claude an email address so I can forward emails I want handled?

This Noodle was 100% on device with Gemma 4 Apple Intelligence isn’t that bad either. It is great for the continuous summaries, but does tend towards making them too short

let’s go

@staysaasy @hnshah people don't want to think

The most important heuristic if you want to know what's going to happen with AI is "people don't want to type"

@petecodes @kylegawley When does the next edition go out?

I got 53 clicks for @kylegawley 's advert in No CS Degree He paid $99 in a newsletter ad sale The price is now $150 again. Anyone want to take that deal? #buildinpublic

From the front-page of HN - I agree! Noodle exists to help elevate your thinking, not replace it. koshyjohn.com/blog/ai-should…

`gemini-3.1-flash-lite-preview` is so fast I'm having to build deliberate delays into my flow so it feels right - the response speed makes it feel like something is broken

Claude worked independently for ~49 minutes to convert a project from OpenNext to @tan_stack - worked perfectly, zero issues

gooooood morning

Tweaking and tweaking

A neat little inspector so I can see if summaries failed-over to the cloud model

I've been working on an app called Noodle. Noodle is a thinking partner. When I Noodle, I take a walk and jabber away about a topic - I "Noodle on it" - and Noodle helps me dig deeper with suggested topics. When I'm done, Noodle generates a summary and some ideas for next time. There's something nice about rambling away with no interruptions. It's almost meditative. I've tried voice-mode agents, but they interrupt, the turn-taking is annoying - I don't get into any kind of flow. I'm trying to do uninterrupted thinking-out-loud. I can talk and ramble and only get input from Noodle if I need it. Swipe up to get Noodle to expand on a specific topic suggestion. Swipe a topic left to see fewer suggestions like that, and right to tell Noodle that was a good suggestion. It quickly catches on. AI certainly has the potential to make us lazier. But I'm hoping something like Noodle can make us smarter, help us explore more of the ideas we have, and ultimately think more clearly. It certainly helps me. Still plenty to refine, but Noodle is ready for some TestFlight users. So hit me up if you've got a stack of things to Noodle on!

@simonbs I am working on this, from a few different angles... do you have a TestFlight :-) Running locally is pretty tricky tbh. Especially once voice gets involved at all

Has anyone built a “second brain” that combines messy notes with AI memory? I want a place to dump ideas, notes, and docs, then have AI actually recall and surface them later when I search. Ideally runs locally on iPhone.

@filippkowalski Funny, I have seen similar takes / I often feel the opposite - codex catches things Claude didn’t I wonder if we are just subconsciously accommodating to the weaknesses of our primary model & switching messes that up

So every now and then, I see big accounts saying Codex > Claude. So I go on, try Codex, and more and more, I'm getting to the conclusion that those are undisclosed ads 🤡 Codex is amazing for review and spotting stuff that Claude misses, but it surprisingly often produces some crazy, junior-level code

@Cloudflare GitHub: github.com/lloydhumphreys…

Didn't want to miss out on all the @Cloudflare fun this week, so I'm finally shipping WorkerWidget Keep an eye on your builds/deployments from the menu bar + jump directly to a build in the dashboard with a click.

@Cloudflare Get it here: workerwidget.lloydhumphreys.com