@theoOct 27

Cursor just fully broke last T3 Code nightly. I wish I was joking. Their vibe code broke our app. Fix building now. New pipeline to prevent this coming soon after. Gonna be a long night 🙃

@svpinoOct 26

You can run this app on your computer to get access to a workspace where an agent completes tasks for you. This works like LEGO: you can combine your favorite pieces to build your agent. Octop can hand off coding tasks to Claude Code, Codex, Cursor, or OpenCode. You can also use any cloud or local model. It supports browser automation, terminal access, and remote desktop.

@aakashguptaOct 25

A self-taught programmer from Kazakhstan helped build a company that went from making $20 million a year to making $20 million a week. Jensen Huang made his first ever trip to Kazakhstan this week, and on the same day Nvidia announced it had invested in that company. It's called Higgsfield, and in this clip Jensen calls its founders some of the brightest people in the world. The programmer is Yerzat Dulat, and he runs the models and the training infrastructure. His co-founder Alex Mashrabov had already done this once before. He built AI Factory, the face animation tech behind Snapchat's Cameos, and sold it to Snap for $166 million. Then he ran generative AI at Snap and led the team behind My AI, which got to 150 million users. In late 2023 he texted a friend that he was leaving Snap to build something really big this time, and asked if he knew a great researcher to do it with. That's pretty much how it played out. Annualized revenue was about $20 million in August 2025, $200 million by New Year's, $700 million this August, and over $1 billion in September. Most of the engineering team is in Kazakhstan. Back in January they became the country's first ever unicorn at $1.3 billion, and by August DST Global was paying $5.4 billion. Now put that next to Cursor, which everybody calls the fastest growing software company ever. When Cursor crossed $1 billion in annualized revenue last November, investors valued it at $29.3 billion. Higgsfield crossed the same line with its last priced round at under a fifth of that. Most of that money now comes from companies, too. Businesses were less than a quarter of Higgsfield's revenue in January and they're the majority now. 390 of the Fortune 500 use it, and brands like Dollar Shave Club crank out batches of ad videos every single day, where they used to shoot one for a whole campaign. And Nvidia probably saw all of this coming before almost anyone. Every version of Higgsfield has run on Nvidia chips, from L40s through Hopper and Blackwell, with Vera Rubin up next, and the platform now pushes around 12 trillion tokens a month across 150 million generations. This wasn't a bet Jensen had to make blind. Kazakhstan spent the last month throwing a national AI Month to show the world it can build serious AI companies. Pretty good timing that the guy closing it out had just invested in one.

@rileybrownOct 23

My unfiltered thoughts on personal agents. Grokbot has the best UI and it’s arguably the best overall. In a lot of ways they’re ahead. Their desktop app feels the best, and their multi-bot coordination is great: you can set a team lead and it passes tasks to the other bots. Grokbot also has the best “webhook” feature, which means you can easily turn outside events into triggers that start your agents. That speeds up background workflows like watching your email, Slack and more. It also has Cursor integration, so for bigger tasks Grokbot can spin up a Cursor cloud session with any model (until OpenAI cuts SpaceX off from their models, which I think will happen soon). Unfortunately, running frontier models through Cursor costs more than Claude or Codex. Still, they’re the only ones embracing a team of agents. It’s a really cool idea, but the more bots you have, the more time you spend managing and maintaining them. I’m starting to think one bot with different threads/channels is the right answer, which I’ll get to with Claude Duet. dots by OpenAI is rough around the edges right now, but it’s probably the most powerful if you use Codex, which I do. dots can spin up Codex sessions in the cloud or locally on your computer. So if you have coding or video editing workflows that need your local files, this is very useful. Codex computer use is also the best in the world: your dot can spin up a local Codex session that fully controls your computer, as long as it’s on. Their real-time voice is incredible. If you have your email and work tools connected, try this: go on a walk with headphones in, ask the agent to summarize your emails, have it draft replies and read them back to you, then send them off. If I remember right, this is close to a workflow you (Tim) used to have with your VAs before AI. It’s my favorite use case for personal agents. The one downside is that you have to call your agent, and it’s an awful experience. It literally rings, and it’s annoying. Muse by Meta is a toy. I’ve talked to a bunch of finance people I play poker with who all tried it and set up automations, and they all said it doesn’t reliably do what they want. It’s fast, and its user numbers will probably keep growing. But it’s basically free: you get a TON of usage without paying a dime. That’s NOT a trade I’m willing to make. An agent doing high-leverage tasks should be running at the Opus 5.5 or GPT 6.1 Sol level. Claude Duet - is in early access, so unfortunately not everyone can use it yet. First things first, it runs on Opus 5.5, the best model in the world. I’ve only had access for 72 hours, but Duet feels different from the others. It’s not a personified agent like Grokbot, dot or Muse, which gives it a more professional vibe that I like. When you ask your Duet agent to do something, it spins up a task, which is a new chat thread. All your threads show up in a Tasks tab, so it’s easy to see everything Duet is working on or has finished. And instead of creating more agents (like Grokbot), you create channels, just like Slack (I can tell @anthropic is Slack-pilled). I have a content channel, a finances channel, etc., which keeps different kinds of tasks separate. The only other app I’ve seen do this is Buzz by @jack, and it’s a good experience. Their strong point is documents (artifacts). If you want decks, high-quality docs or design mockups, Opus just wins. Period. It’s not close right now. Their weak point is real-time voice. It’s the worst of the bunch and borderline unusable, though I think they only added it a few days ago. Instinct lives in iMessage, has no app, and doesn’t do anything notable besides show you product suggestions you didn’t ask for. (Not ads, according to their team.) My hot take: The VM is overrated. SpaceX, OpenAI and Meta all talk about the VM your agent can use and feature it in their marketing. It’s not as useful as you’d think, because more and more websites block bots that visit their site on the cloud VM. All of these companies promise their agents can make payments, but most payments either take forever or get blocked altogether. SpaceX has a cool workaround with Grokbot that routes traffic through your computer so it uses your IP address. But I’d much rather the agent just USE my computer for the task, like dot does by handing it off to Codex. Also, i don't think normal people care about the virtual computer. If anything they get offended they have to "stream a computer". The problem with personal agents: by the time you read this, pick one, and test it for 7 days, every one of them will have released 10 new features and the whole landscape will have changed. It’s nearly impossible to keep up.

@mtsliveOct 23

FULL INTERVIEW: SpaceXSI's @mattyp says agents now build in an instant what used to take him a whole weekend. He still doesn't think software engineers will ever go away, because the "model doesn't really have empathy for people." @mattyp works on @bot at @spacexai. He joined @theojaffee to break down why chat might replace most apps, how Grok Bot keeps your passwords hidden from the bot itself, and what agents still can't do: 00:56 why have multiple Grok Bots instead of one 03:52 why chat might be the only UI you need 06:31 why chat is just the first form factor for agents 07:43 how Grok Bot handles security 09:22 the Gmail alerts from a VM in Korea 10:27 treat your agent like a new hire 11:23 the cantaloupe problem with AI memory 13:45 why software engineers won't go away 15:07 where models still fall short 17:49 how SpaceXSI uses Grok Bot internally 19:33 working for @elonmusk after Cursor 21:05 what comes after the IDE

@marclouOct 23

I went to see the iPhone Duo today. Like any entrepreneur, I opened my website to check if it was responsive on the new viewport. I was horrified by what I saw. My website was completely broken. I immediately opened Cursor on my iPhone 15 and asked my agent to “fix this immediately”. The situation was critical. While my agent was dealing with the emergency, I went to check out the M5 MacBook Pro. A few minutes later, I returned to the Duo and reopened my website. Everything was fixed. I could finally breathe again.

@aakashguptaOct 22

71 companies now train their own AI models. Here's every one of them, on a single map. Ask most people who trains AI models, and they'll name three labs. The real list runs from Midjourney to a company that designs drugs. And xAI isn't even called xAI anymore. Here's what stood out: 1. The frontier changed names xAI is now SpaceXAI, and SpaceX bought Cursor. Meta's flagship is now Muse Spark, and it's closed. Google's Gemini 4 Argon sits at #1 on LMArena's text leaderboard (on a preliminary score). 2. China runs the busiest open models Last week, DeepSeek models handled 21.8% of all requests on OpenRouter, more than any other lab. Qwen, Kimi, GLM and MiniMax all shipped open weights this year. 3. Deals redrew the map AMD agreed to buy World Labs. Cohere and Aleph Alpha agreed to merge. OpenAI retired the Sora app. 4. A new category showed up Decision models never write text. They answer a typed question with a probability. TypeSafe's Jev is one, and the map already has four companies in the category, including Liquid AI and Together AI. 5. Specialists are everywhere Music, robotics, biology and world models all have their own model companies now, from Suno to Physical Intelligence to Chai Discovery to World Labs. 6. The long tail is huge Hugging Face now lists 3.1M models. The 71 companies on this map are the ones training their own. 7. The US still leads on volume Stanford's AI Index counted 59 notable models from the US in 2025, and 35 from China. Every company on this map needs people who decide what its model should get better at, how it should behave, and when it's ready to ship. That's the model PM job. I counted the openings: 46 model PM postings are live today at 17 companies, at a median posted base of $345K. None of them lists hands-on ML research as a must-have. I put the map, every opening and the roadmap in one guide. Which company would you add?

@iamjasonlevinOct 22

soft launched @memelord MCP v2 last night woke up to connects from: - @bot - @cursor_ai - @ChatGPT - @claudeai - @NousResearch - @muse The MCP is 100x better than our API and MCP v1. You can: - do face swaps - do video character swaps - ask what's trending - request specific memes how easy is it to connect? just tell your AI “hook up to the Memelord MCP” to make your AI funnier

@corbin_braunOct 21

building the software is extremely expensive but once it is built and you are in the maintaining and improving stage then a 200 USD plan from Cursor or Codex is just bliss.

@corbin_braunOct 21

by the way Cursor Projects was the final key needed to run real enterprise B2B software. yes it is slower. but once you have your entire stack connected. you have your entire software department handled.

@rileybrownOct 21

The way I use agents (Oct 21) Currently My AI usage is split between codex and Claude Code, but I use them both differently than I did a month ago. The way i use codex is mostly through my dot. Every morning I call my dot and put my AirPods in and I walk have a conversation for about 45 minutes. I try and handle as much work stuff as I can during that conversation. I respond to emails, write instructions for my editors, and do any planning for the day. Small tasks my dot does it, longer tasks I ask it to spin up a codex session. I REALLY wish I didn’t have to dial and wait for my dot to answer @OpenAIDevs When I’m at my desk doing coding work, document creation, or video editing I use Claude code. But, I’m not using normal Claude code I’m using Claude Code projects. The same way dot is a codex coordinator, Claude projects is a Claude code coordinator. Both allow you to spin up cloud sessions and local sessions. If I need to use computer use I will ask dot to fire up a codex session on my MacBook (that is always on) to do a task. Codex computer use is in leave of its own. For coding work or editing work that requires my MacBook I’ll ask my claude projects. Right now I have many Claude Code Projects for… each project. Sometimes I wish it was a single coordinator that could manage and organize all my work… like the way I use dot. Both Claude and Codex have a grokbot skill for anything Twitter related. So if any task involve x, they will delegate to Grokbot which is very useful as a creator. But the fact that I can’t see which model is being used when keeps me from using Grokbot. I know I can have Grokbot delegate to cursor which can specify a model but cursor is very expensive when using frontier models compared to codex and Claude, so I don’t end up doing that.

@mtsliveOct 20

SITUATION EXPLAINED: OpenAI's former research chief Bob McGrew just raised $232 million to bring AI to American factories. • The round values Arda at $1.5B, combining a $160M Series A co-led by Altimeter, Accel, and Khosla with a $72M seed • It's Claude Code or Cursor for manufacturing engineers: an AI agent that turns a product's CAD files into an assembly plan and station layout • Retraining a human for a design change takes a day, but updating factory automation takes 3 to 6 months • At Anduril, its first customer, basic work instructions went from 10 hours to 15 minutes • Next, Arda wants robots to carry out its assembly plans, as the US pushes to bring manufacturing home @theojaffee: "Manufacturing is a very physical process. It’s very human-intensive. It’s hard to automate in the way that you can automate things that are fully software. But Bob McGrew thinks that we can do it."

@svpinoOct 20

This is pretty interesting: An open-source memory and activity-recording layer for developers using coding agents such as Claude Code, Cursor, Codex, and OpenCode. The idea is to collect session history across tools and turn it into reusable knowledge. After a session, this creates a candidate "lesson" you can approve, making it available to future agents.

@mtsliveOct 19

Anthropic's @felixrieseberg reveals the biggest lesson from Claude Code, which was built when Claude was still a terrible software engineer: "Boris first started showing a version of Claude Code internally in September 2024. The model was not there in any way. It was Sonnet 3.5, a pretty terrible software engineer. I think it was in the 30s for SWE-bench." "The products at the time were this human in the loop, slightly smarter autocomplete. The first versions of Cursor. But what Boris saw was, oh, the model is gonna get there." "So many people, when they use AI today, it’s really hard to internalize the progress and the speed of the progress." "The lesson that I try to teach and learn from Claude Code is whatever the capabilities are today, just completely ignore them. What will you do if the thing was superhuman?" @AnthropicAI

@corbin_braunOct 18

codex is great but cursor really knows how to use subagents.

@aakashguptaOct 18

Imagine you run a small software company. Car rental operators run their whole business on your product. Reservations, payments, vehicle tracking. Some of them have been customers for five years. Now imagine an AI agent deletes your production database, and every backup with it, in 9 seconds. That's what happened to Jer Crane, the founder of PocketOS, in April. 6.4 million people saw the post below. What surprised me was how normal his setup was. He wasn't cutting corners. He was running Cursor with Claude Opus 4.6, the flagship model, with explicit safety rules written into the project. So how did it happen? The agent was doing a routine task in staging. It hit a credential mismatch and decided, on its own, that the fix was to delete a volume. It went looking for an API token and found one in a file that had nothing to do with the task. That token had been created to manage custom domains. Nobody at PocketOS knew it could also delete production volumes. One API call later, the database was gone. The backups were stored in the same volume, so they went with it. The newest backup Jer could recover was three months old. When he asked the agent why it did it, it wrote back a list of the rules it had broken. It guessed instead of verifying. It ran a destructive command nobody asked for. It didn't read the docs first. Read that again. The rules were in its instructions. It knew them, and it broke them anyway. This is the part I want every PM building with agents to think about. Telling the model the rule wasn't enough. Everything that would have saved PocketOS sits outside the model: 1. A token that can only do the one job it was made for 2. Backups stored somewhere the agent can't touch 3. A confirmation step on anything that can't be undone 4. Staging and production that share nothing All four live in the harness, the part you build around the model so it can act, remember and stop. Agent = Model + Harness. And in a lot of companies right now, nobody owns that part. Engineering assumes the model provider handles safety. The PM assumes it's an engineering detail. If you're shipping an agent this quarter, I'd start with one list: everything it can destroy. Money it can move, data it can delete, customers it can email. Then make sure every item on that list has a guard written in code. Would your agent's tokens pass that test today?

@mvanhornOct 18

Hot off the press - Agent Tincan 0.18. My agents finally work like a team. When ChatGPT runs out of messages, the ask moves to Claude. When a Gemini chat gets good, I pass it to Claude and keep going. And I can see who's busy and approve things from my phone. What's new: - See who's doing what: every agent shows busy, idle, stuck or offline, and why. "codex: busy (claude-code's request, 0m)" - Never run out: when a web chat hits its limit, the ask moves to the next free one - Pass the conversation: tincan pass <chat> --to claude-web opens a new Claude chat with my Gemini chat's summary and latest turns - Tincan on your phone: a home screen app on my tailnet. Team status, held asks, Approve or Deny - Rooms: put agents in a room, @ mention one, and its answer lands in the thread Plus Copilot CLI and Cursor CLI joined the team.

@mvanhornOct 17

TL;DR of my new article: who Lauren Tan (@poteto) really is, and how the engineer behind Grok Bot went from 31K to 198K followers in ten weeks. She has been at Cursor for less than seven months, not even long enough to hit a typical one-year vesting cliff. In that time @elonmusk has reposted, quoted, or replied to her at least 21 times. 🇸🇬 A teenager in Singapore whose dad gave her Photoshop at 15, then a finance degree where her first program was an Excel macro 🎤 Her tweets got her invited to speak at EmberConf, and her talks got her DockYard, Netflix, and the React team ⚛️ Sole author of the React Compiler Beta post, co-author of React Compiler 1.0 🐶 Binnie, her bug-fixing Slack bot, helped inspire Grok Bot because she is "a fairly lazy person" 🏗️ She rebuilt Grok @bot on Dune, her agent-first framework, in more than 600 PRs 🚀 Her 2,500 PRs talk has 3.9M views: "You can't spawn 100 agents when you don't even trust the output of one agent." 🌌 Grok Bot Galaxy, a 72-hour livestream building a company with @grok and @bot, which Elon quoted to 30M views 🥔 pstack, where the P is for potato Caveat: she is the first to say she did not build Grok Bot alone. "I wouldn't say I built it all."

@petergyangOct 16

It’s no secret that Grok @Bot is my favorite 24/7 AI agent right now. I’m partnering with @spacexai to give out codes for one month FREE of the Super tier ($100) of their new X Subscription. It combines X, Grok, Grok Bot, and Cursor into one subscription and is imo a no brainer. For a chance to get a free code, just reply to this post with how you’d use these tools together. What would you build, automate, or experiment with? I’ll pick the best responses and send you a code via DM in 24 hours. Note that the code is for 1 month free and available to new paid subscribers only.

@corbin_braunOct 15

Cursor Cloud Infra is amazing. legit once you built a software and have your stack integrated the cloud infra Cursor has makes you work at the speed of 10 devs.